[103.158.136.

64] Project MUSE (2024-02-06 19:57 GMT) University of Edinburgh

non-­
­h uman
witness-­
ing
Thought
in the Act

a series
edited by

Erin
Manning

& Brian
Massumi

duke
university
press

durham
& london

2024
war,
data, and

ecol­o gy
­a fter

non­human
witnessing
the
end of

the world

michael
richardson
​​© 2024 Duke University Press
All rights reserved
Printed in the United States of
Amer­i­ca on acid-free paper ∞
Proj­e ct Editor: Michael Trudeau
Designed by Aimee C. Harrison
Typeset in Minion Pro and ibm Plex Mono
by Westchester Publishing ­Services

Library of Congress Cataloging-in-Publication Data

Names: Richardson, Michael, [date] author.
Title: Nonhuman witnessing : war, data, and ecology after the end
of the world / Michael Richardson.
Other titles: Thought in the act.
Description: Durham : Duke University Press, 2024. | Series:
Thought in the act | Includes bibliographical references and index.
Identifiers: lccn 2023015594 (print)
lccn 2023015595 (ebook)
isbn 9781478025641 (paperback)
isbn 9781478020905 (hardcover)
isbn 9781478027782 (ebook)
Subjects: lcsh: Mass media—Social aspects. | Mass media—­
Political aspects. | Mass media and technology. | Mass media—­
Influence. | Information society. | Communication and technology. |
Evidence. | Witnesses. | bisac: social science / Privacy &
­Surveillance (see also political science / Privacy & Surveillance) |
social science / Media Studies.
Classification: lcc p96.s65 r534 2024 (print) | lcc p96.s65 (ebook) |
ddc 302.2—dc23/eng/20231103
lc record available at https://lccn.loc.gov/2023015594
lc ebook record available at https://lccn.loc.gov/2023015595

Cover art: Yhonnie Scarce, Thunder Raining Poison, 2015. Blown-­

glass yams, stainless steel, reinforced wire, dimensions variable.
Installation view at Tarnanthi Festival 2015, Art Gallery of South
Australia. Photography: Janelle Low. Image courtesy of the artist
and this is no fantasy, Melbourne.
​for adrian
and sacha—

may you witness

many worlds
This page intentionally left blank
contents

Acknowl­edgments ​ix

Introduction 1
nonhuman witnessing

one 37
witnessing vio­lence
two ​80
witnessing algorithms
three ​112
witnessing ecologies
four ​150
witnessing absence

coda ​174
­ oward a politics
t
of nonhuman witnessing

Notes ​185
Bibliography ​207
Index 229
This page intentionally left blank
acknowl­e dgments

i acknowledge the unceded sovereignty of the Bidjigal and Gadigal

­people of the Eora Nation on whose lands this book was written. From tens
of thousands of years before the settler invasion through ­today, their knowl-
edge, culture, and customs have protected and nurtured this land, and ­will
continue to do so long into the ­future. Always was, always w ­ ill be.
This book responds to a pervading sense of crisis. Its writing began in
earnest with the Sydney air thick with bushfire smoke and ­these words of
acknowl­edgment are typed in the wake of catastrophic flooding and with yet
another pandemic wave surging. I was fortunate enough to hold an Austra-
lian Research Council Discovery Early ­Career Research Award from 2019 to
2021 (de190100486), a research fellowship that spared me the intensity and
­labor of teaching through the first two years of the covid-19 pandemic, al-
though not the upheavals, lost jobs, merged faculties, and other stresses of my
place of work in ­those years. That was and remains an ­immense privilege, one
not afforded to the many academics forced to eke out research time ­under the
strain of unsustainable teaching and administration loads.
If this book began anywhere, it was with two questions asked a year apart
on separate visits to Freie Universität, Berlin. The first was from Jonas Bens,
who asked ­whether drones might become nonhuman witnesses. The second
came ­after delivering a paper that began to answer that question, when Jan
Slaby asked w ­ hether the paper would become a book. I am grateful to both
Jonas and Jan, but their questions would not have been pos­si­ble without an
invitation from Kerstin Schankweiler to visit Berlin in 2017 for the inspiring
Image Testimonies: Witnessing in Times of Social Media symposium and
again a year l­ater as a visiting fellow at the Affective Socie­ties Collaborative
Research Centre. My deepest appreciation to Kerstin, who has since become
my coauthor, coeditor, and friend.
I benefited im­mensely from the chance to share ideas and words that
would become this book at conferences, often on more than one occasion,
including 4s, Crossroads in Cultural Studies, Deleuze Studies, Cultural
Studies Association of Australasia, the Australasian Society for Continental
Philosophy, International Communication Association, the Aesthetics of
Drone Warfare, Digital Intimacies, the Australian S­ ociological Associa-
tion, World of Drones, Data ­Futures, Drones in Society, Witnessing and
Worlds beyond the H ­ uman, and the worlding experience of the Affect
wtf and Capacious aims conferences ­organized by the shimmering Greg
Seigworth. Thank you also to the many ­people who invited me to speak on
drones, witnessing, and related ­matters. Kerstin Schankweiler for Image
Testimonies at Freie Universität; Elizabeth Stephens for F ­ uture Affects at
University of Queensland; Kerstin Schankweiler and Tobias Wendl for the
art history colloquium at Freie Universität; Lyria Bennett-­Moses, Danielle
Hynes, and Niamh Stephenson for the Data Injustices Roundtable at unsw;
Vered Maimon at Tel Aviv University; Catherine Rhodes at the Cambridge
University’s Centre for Existential Risk; Lilie Chouliaraki at the London
School of Economics; Carolyn Pedwell and Vince Miller at University of
Kent; Shela Sheikh at Goldsmiths; Jonas Wittelman at the Schaufler Lab at
tu Dresden; Naoko Abe, Justine Humphry, and Chris Cheshire for Mobile
Robotics in Public Space at the University of Sydney; Kerstin Schankwei-
ler (again!) at tu Dresden; Lindsay Kelley for unfiled at unsw; Timothy
Gregory and Jeremy Moses for the Weapons, Wounds, War seminar series
at Auckland University; and Andrew Brooks for the seminar series in my
home department at unsw Sydney. ­Those prompts to think through theories,
themes, and case studies and the generous engagement of the audiences at
­those talks with the limits and potentials of my thinking made for a much
stronger book.
­Earlier versions of parts of this book ­were also published in academic
journals but have since been revised extensively. I am im­mensely grateful to

x Acknowledgments
the editors and peer reviewers who guided ­these texts at a formative stage.
Parts of chapter 1 draw from “How to Witness a Drone Strike,” Digital War
(2022), published in a special issue edited by Olga Boichak and Andrew
Hoskins. What would eventually become sections of chapter 3 appeared in
“Climate Trauma, or the Affects of the Catastrophe to Come,” Environmen-
tal Humanities 10:1 (2018). An e­ arlier version of chapter 4 was published as
“Radical Absence: Encountering Traumatic Affect in Digitally Mediated
Disappearance,” Cultural Studies 32:1 (2018), in a special issue on media and
affect edited by Sarah Cefai.
I am doubly (if not triply) thankful to Sarah Cefai, who orchestrated my
visit to Goldsmiths in 2020, on the cusp of the pandemic, which led to con-
versations that pushed me to resolve some thorny questions with Joanna Zyl-
inska, Matt Fuller, Chris Woods, Ariel Caine, Eyal Weizman, Susan Schuppli,
Shela Sheikh, and Sarah herself.
Conversations large and small, online and off, with a host of interlocu-
tors contributed in large and subtle ways to the ideas in this book or helped
cheerlead it into existence. Thank you Adrian Mackenzie, Amy Gaeta, An-
drew Yip, Anthony McCosker, Ayesha Jehangir, Baden Pailthorpe, Boram
Jeong, Carolyn Pedwell, Casey Boyle, Chad Shomura, Charlotte Farrell,
Chris Agius, Chris O’Neill, Chrstine Parker, Crystal Abidin, Declan Kuch,
Donovan Schaefer, Edgar Gomez Cruz, Elke Schwarz, Emma Jane, Emma
Quilty, Fleur Johns, Gilbert Caluya, Hagit Keysar, Hannah Buck, Heather
Horst, Hussein Abbass, Jake Goldenfein, James Parker, Janet Chan, Ja-
than Sadowski, Jennifer Terry, Jenny Rice, Jodi Brooks, Joel Stern, Joseph
DeLappe, Joseph Pugliese, Jussi Parikka, Karin Sellberg, Kat Higgins, Kath-
rin Maurer, Kathryn Brimblecombe-­Fox, Kynan Tan, Larissa Hjorth, Liam
Grealy, Lilie Chouliaraki, Lisa Slater, Louise Ravelli, Lyria Bennett-­Moses,
Mark Andrejevic, Matthew Arthur, Mel Gregg, Michael Balfour, Michal Gi-
voni, Michele Barker, Milissa Dietz, Monique Mann, Nathaniel Rivers, Ned
Rossiter, Niamh Stephenson, Nisha Shah, Olga Boichak, Omar Al-­Ghazzy,
Paul Frosh, Poppy de Souza, Rachel Morley, Rowan Wilken, Sarah Truman,
Sonia Qadir, Sukhmani Khorana, Tess Lea, Tim Neale, Tom Sear, Verena
Staub, Willy Blomme, Xan Chacko, Yanai Toister, and all t­ hose that I have
inevitably forgotten. Some of the most rewarding conversations took place
with participants in the Drone Cultures symposium that changed incarnation
from in-­person to online over the course of 2020 and via the Drone F ­ utures
seminar series that I ran online through the first year of the pandemic. Spe-
cial thanks to Ronak Kapadia, Antoine Bousquet, Jairus Grove, J. D. Schnepf,

Acknowledgments xi
Kate Chandler, Mahwish Chishty, Thomas Stubblefield, Caren Kaplan, and
Alex Edney-­Browne. Your p ­ resentations and conversations (many of which
have continued) all shed new light.
Coauthors and collaborators on other proj­ects also helped me to work
through this one. I am im­mensely grateful to the legends J. D. Schnepf, Beryl
Pong, Adam Fish, Heather Ford, Kerstin Schankweiler, Amit Pinchevski,
Magdalena Zolkos, Madelene Veber, Anna Jackman, Kyla Allison, and An-
drew Brooks.
At unsw, I’m lucky to work with many wonderful academic and profes-
sional colleagues in the School of the Arts and Media, especially my cocon-
spirators in the Media F­ utures Hub. One of g­ reat joys of academic life has
been learning from my gradu­ate students Simon Taylor, Meng Xia, Asal
Mahmoodi, Theresa Pham, Rachel Rowe, Bron Miller, Kyla Allison, Maddie
Hichens, Katariina Rahikainen, and Maryam Alavi Nia.
Remarkable comrades in academic life have enriched my world. Deepest
thanks to Anna Gibbs, who set me on the path; Meera Atkinson, my first
academic partner in crime; Magdalena Zolkos, for introducing me to affect;
Andrew Murphie, for opening the worlds of media; Stephanie Springgay,
for wise guidance and late-­night bourbon; Rebecca Adelman, for finding
joys in email; Elizabeth Stephens, for lifting up o
­ thers, me included; Ramas-
wami Harindrinath, for getting me hired (and much more); Helen Groth, for
generous mentoring; Tessa Lunney, for laughs and lunches and always re-
membering; Nick Richardson, for unconditional cheerleading; Collin Chua,
for knowing what to read; Caren Kaplan, for solidarity and wisdom; Anna
Munster, for the creative energy and collaborative enthusiasm; Tanja Dreher,
for believing that change is pos­si­ble; Thao Phan, for the unerring capacity
to ask the right questions; and Greg Seigworth, for endless capaciousness.
Special thanks to the artists who generously gave permission to include
their work: Kathryn Brimblecombe-­Fox, Baden Pailthorpe, Kynan Tan, Ed-
ward Burtynsky, Grayson Cooke, Mahwish Chishty, and Yhonnie Scarce,
whose stunning Thunder Raining Poison appears on the cover. Par­tic­u­lar thanks
to Noor Behram for his striking photo­graph (figure 1.3), used h ­ ere ­under fair
use guidelines and (I hope) in allegiance with the politics of his work.
A book is nothing without readers, and the writing of this one is indebted
to the time, insight, critique, and support of some remarkable ones. The
Book Proposal Club of Astrida Neimanis and Lindsay Kelley helped build
the framework and find the book a home at Duke. Greg Seigworth, Nathan
Snaza, and Tanja Dreher read early, lumpy drafts of the introduction and
made it so much stronger through their generous attention to its unformed

xii Acknowledgments
thoughts and suggestions of further reading. Anna Munster read chapter 3
and homed in on its weakest points with unerring precision, which helped
strengthen that chapter greatly.
Research underpinning this book—­and especially chapters 1 and 3—­was
enormously assisted by Madelene Veber, who has worked with me over the
last four years and is, among other t­ hings, brilliant, indispensable, astute, and
creative. I’ve no doubt that Madelene ­will soon be writing acknowl­edgments
of her own.
For so many reasons, I am forever in debt to Andrew Brooks and Astrid
Lorange, staunchest pals and wisest of readers. Both read the manuscript top to
bottom and provided such thoughtful feedback and incisive edits. But more,
your fierce politics, love of thought and poetry and language, your convivial
joy in snacks, and your friendship have meant every­thing. Love you. gffs.
Two anonymous reviewers read both the proposal and eventual manu-
script. Their generous, critical, and constructive engagement with the proj­ect
at proposal stage had a transformative effect on the argument, framework, and
analy­sis. Without them, this book would be much poorer, or not a book at
all. All flaws that remain are mine alone. Thank you.
Brian Massumi and Erin Manning believed in this book from the first
loose description at a Montreal café in the dead of winter. Brian’s work led me
to fall for affect theory, for radical empiricism and ­process philosophy, and
for the relationality in all ­things, and you ­were both so welcoming to a clue-
less gradu­ate student at the Into the Diagram workshop in 2011. I am truly
honored to be published in your series, alongside so many brilliant o ­ thers.
At Duke University Press, I am immeasurably thankful to Courtney
Berger, whose editorial wisdom and unfailingly warm support for the book
made all the difference, editorial assistants Sandra Korn and Laura Jara-
millo, and Aimee Harrison for the stunning cover design. As all authors
know, a huge amount of unsung work goes into making a book and getting
it to readers. Thank you, Michael Trudeau, James Moore, Emily Lawrence,
Chad Royal, and the rest of the production and marketing crew for your
effort, enthusiasm, attention, and care. Big thanks to Cathy Hannabach
and Morgan Genevieve Blue from Ideas on Fire for stellar work producing
a rich index.
The Thread—­Na’ama, Phil, Astrid, Andrew, Nick, Annie, Zoe—­your soli-
darity, wisdom, humor, advice, and love are beyond ­measure.
To name all the ­family and friends who nurtured this book and its author
is an impossible task. You know who you are. Mum—­you taught me to love
learning and to seek justice. Dad—­you showed me how to pursue the exact

Acknowledgments xiii
and true. Daniel—­I only wish you w ­ ere nearer so that we could share more
joys together.
Fi­nally, my deepest thanks and love go to Zoe, Adrian, and Sacha. ­There
is no one like you, Zoe Horn, so smart, creative, caring, loving, and funny.
You keep me on track, you put ­things in perspective, you teach me again
­every day what ­matters most, and you always believe. Adrian and Sacha, this
book is for you, for every­thing y­ ou’ve taught me, and for the desperate need
to make new worlds to live and love within.

xiv Acknowledgments
introduction

nonhuman
witnessing
at 6:15 a.m., february 21, 2010, on a deserted stretch of road in the Uru-
zgan Province of A ­ fghanistan, a convoy of three vehicles slowed to a halt
and figures spilled out, clumping and milling as dawn light filtered through
the mountains. Captured by the Multi-­Spectral Targeting System (msts)
slung below the nose of the loitering mq-1 Predator, imagery of the con-
voy streamed across military networks to screens in the United States and
­Afghanistan. On the screens, engines and ­people glowed white against the
gray-­black landscape as indistinct heat signatures bled into one another in
the strange aesthetic of forward-­looking infrared (flir). Image and control
data flowed through the network, moving between dif­fer­ent devices, infra-
structures, and protocols. Connected by a ku-­band satellite link to Ramstein
Air Base in Germany, the Predator’s data then traveled down optical fiber
cable ­under the Atlantic to the Ground Control Station at Creech Air Force
Base outside Las Vegas, Nevada, to image analyst “screeners” in Florida, to
command posts and ground stations across the globe, and to an encrypted
server farm for archiving, where the video and its accompanying metadata
would be logged, recorded, and held for ­future analy­sis. Years ­later, ­these
time-­stamped pixel arrays of ones and zeros likely became part of the vast
video archive used to train machine learning algorithms to replace the l­ abor
of image analysts, a proj­ect initiated in partnership with Google and other
tech ­giants in a sign of strengthening ties between the architects of algorith-
mic enclosure and ­those of increasingly autonomous warfare.
On that pale morning, one place the video feed failed to reach was the US
Special Forces unit conducting an operation against a local Taliban leader
in nearby Khod. ­Afghanistan’s weak communications infrastructure and a
reliance on satellite bandwidth meant that the imagery never made it to the
ground, despite being subject to much debate as it was examined by screen-
ers, operators, and commanders. Conducted by radio and military internet
relay chat (mIRC) across discontinuous networks within the operational
apparatus, the debate over what the images showed angled ever more inexo-
rably ­toward vio­lence as the affective surge ­toward action cohered with the
indistinction of the drone’s mediations. Alongside the msts, the Predator was
equipped with gilgamesh, a sophisticated eavesdropping system capable of
blanket signal interception of nearby cellphones. Like the image screeners,
analysts combing its data oriented their interpretation ­toward perceiving the
convoy as a node within an ­enemy network. On the ground, two dozen men,
­women, and ­children spread prayer rugs on the dirt, while military personnel
on the other side of the planet argued over how to read the varied morpholo-
gies produced by the sensor-­network-­feed. Framed with military discourse,
­these uncertain bodies w ­ ere swiftly fixed as “military-­aged males” and thus
subject to potential elimination.
Prayers complete, the three vehicles continued along the road, veering
away from the Special Forces at Khod in what one of the drone crew inter-
preted as a “flanking” maneuver. The lurking Predator carried only a single
missile, so two Kiowa attack h ­ elicopters w
­ ere scrambled into position and
a l­ittle a­ fter 9 a.m., the convoy hit a treeless stretch of road. Guided by the
drone’s l­aser targeting system, two agm-114 Hellfire missiles launched from
the Kiowa h ­ elicopters and struck the first and third cars, explosive charges
in each detonating to fragment the shell casing. Metal and flesh tore apart
and fused together. Bodies w ­ ere everywhere, w ­ hole and in pieces. Nasim, a
mechanic who survived the blast, l­ ater recalled wrecked vehicles, a headless
corpse, another body cut in half. On the full-­color video feed that the crew
switched to ­after the strike, pixels re-­presented themselves as ­women and,
eventually, as ­children. ­Later, the Pentagon claimed sixteen dead, including
three ­children; villa­gers said twenty-­three, including two boys named Daoud
and Murtaza. A swiftly ordered US Department of Defense investigation
traced the tangled lines of communication, the pro­cesses of mediation, and
the failures of vision and transmission. Its report ran over two thousand

2 Introduction
pages. When eventually released ­under a Freedom of Information request
filed by the American Civil Liberties U ­ nion, the report provided rare insight
into the secretive inner workings of drone warfare. Much attention was paid
to the transcript of communications between the Predator crew and ground
command. ­L ater used to frame both journalistic and scholarly accounts,
the transcript distilled the hubris, faith in technology, and tendency ­toward
vio­lence that animates remote war. More than a d ­ ecade l­ater, the event and
its mediations and remediations remain a critical aperture into the drone
apparatus.1
In all of this, who—or what—­bears witness? H ­ uman witnesses abound:
the victims and survivors, whose flesh and words bear the scars and carry
the lived truth of hellfire from above; the ­pilot and sensor operator, the com-
manders, military ­lawyers, image analysts; the military investigators; the
documentarians and journalists who ­will tell the story of what happened, and
their audiences across the world; perhaps even the scholars, myself among
them, who turn to this moment to help make sense of remote war. Yet what
of our nonhuman counter­parts? Th ­ ere is the ground soaked in blood, the
roadway buckled by the explosive force of two warheads and blackened by
fire, and the dirt and stone of the roadside in a land wracked by war, and
the carbon-­rich atmosphere through which the missile and signals travel,
another in the countless pro­cesses contributing to the ecological catastro-
phe that consumes the planet. ­There is the drone itself, not only the aerial
vehicle and its payload of sensors capturing light across the spectrum but
its signals relays, and the complex network of technologies, pro­cesses, and
practices that make up the apparatus. And ­there are, too, the algorithmic
tools for snooping cellphones and scouring video; the data centers sucking
power for cooling and expelling heat for stack upon stack of rack-­mounted
computers; the undersea cables that carry military and civilian data alike.
If we extend the assemblage further, we arrive at lithium mines and orbital
satellites, image datasets and environmental sensors, cellphone manufactur-
ers and cloud ­services.
In most accounts of witnessing, much of this would be excluded alto-
gether, relegated to the status of evidence, or assigned the role of intermedi-
ary, dependent upon a h ­ uman expert or interpreter. Nonhuman Witnessing
refuses that relegation and instead deepens and widens the scope of wit-
nessing to include the nonhuman. Opening witnessing to the nonhuman
provides deeper, more finely tuned understandings of events for us h ­ umans.
But this book goes further, arguing that nonhuman witnessing enables the
communicative relations necessary for an alternative and pluriversal politics,

Nonhuman Witnessing 3
founded on the capacity of nonhuman entities of all kinds to witness and
through that witnessing compose new ethicopo­liti­cal forms. H ­ uman wit-
nessing is no longer up to the task of producing the knowledge and forms of
relations necessary to overcome the catastrophic crises within which we find
ourselves. Only through an embrace of nonhuman witnessing can we ­humans,
if indeed we are still or ever w
­ ere ­human, reckon with the world-­destroying
crises of war, data, and ecol­ogy that now envelop us.

nonhuman witnessing

Witnessing is fundamental to cultures, communities, and polities b­ ecause

it pushes events onto the stage of justice, helps determine significance and
truth, contributes to the making of shared knowledge, anchors ­political sub-
jectivity, and produces responsibility. Necessarily relational, witnessing forges
an intensive connection between witness and event, a registering of some-
thing happening that forms an address and insists upon a response. Not just
any encounter, witnessing exceeds itself and calls ­others into relation with it.
This is why witnessing co-­constitutes epistemic and moral communities, and
even ­political subjectivity itself. Witnessing pushes sense-­making to grapple
with traumas that refuse comprehension. Witnessing responds to vio­lence in
all its elusive and terrible forms, but also to won­der, beauty, and even banal-
ity. Witnessing precedes what comes to be deemed truth and gifts authority
to collective memory, but it also depends on the permeability and fluidity of
individual and collective subjectivities.
Witnessing and testimony are found in dif­fer­ent guises in court rooms,
church halls, h ­ uman rights tribunals, fiction and poetry, media reports, sci-
entific laboratories, and countless other places. Small won­der, then, that
witnessing figures centrally in academic thought, from philosophies of ethics
and ­political theory to media, literary, religious and science and technology
studies, to name but a few fields attentive to its normative effects, its constitu-
tive pro­cesses, and its historical specificities. Far from a static concept or set
of practices, witnessing itself has transformed throughout Western history,
taking on new forms and dynamics alongside and in response to changes
in technology, politics, sociality, and religion. As both theory and practice,
witnessing has proven pervasive and durable, as well as malleable and
exploitable. But in an era of interlocking crises of technoscientific war, eco-
logical catastrophe, and algorithmic enclosure, both the theory and practice
of witnessing need to reckon far more deeply with the nonhuman.

4 Introduction
Nonhuman Witnessing is about what happens when the frame of what
counts as witnessing expands, how more-­than-­human epistemic communi-
ties might form, and what this might mean for subjectivity, the nature of
justice, and the strug­gle for more just worlds. I develop nonhuman witness-
ing as an analytical concept that brings nonhuman entities and phenomena
into the space of witnessing and accords them an agency other­wise denied or
­limited by witnessing theory to date. This strategic gesture makes room for
excluded knowledges, subjectivities, and experiences within a wider frame-
work of cosmopo­liti­cal justice. It does so through an analy­sis of technolo-
gies, ecologies, events, bodies, materialities, and texts situated in crises of
military, algorithmic, and ecological vio­lence. While witnessing can certainly
occur separate from vio­lence, this book focuses predominantly on instances
of state and corporate vio­lence that occur across a variety of scales, speeds,
temporalities, and intensities. This book understands vio­lence as purposive
harm inflicted on ­people, animals, environments, and the ecological rela-
tions that make life and nonlife inextricable from one another. Vio­lence
thus d­ etermines the possibility, capacity, and nature of life for h
­ umans and
nonhumans alike. This instrumentality means that vio­lence is distinct from
mere force and cannot be neatly equated to destruction or death in general. In
the way I use it ­here, vio­lence captures environmental, ecological, structural,
technological, affective, discursive, and infrastructural forms of instrumental
harm, as well as the directly corporeal and material forms that are most
obvious and widely accepted. One of my central propositions, then, is that
nonhuman witnessing brings more excessive and elusive vio­lence into the
frame of witnessing in ways that ­human witnessing cannot.
What this book proposes is bold: to unknot witnessing, weave it anew as
inescapably entangled with the nonhuman, and within the warp and weft of
that weaving find a renewed ­political potential for witnessing ­after the end
of the world. My argument is that understanding witnessing as bound up
with nonhuman entities and pro­cesses provides new and potentially trans-
formative modes of relating to collective crisis and the role of the h ­ uman
within it. For many on this planet, crisis is neither a new experience nor an
exceptional event but rather forms the condition ­under which life is lived.
The book takes as its starting point the presumption that con­temporary
crises of war, algorithmic enclosure, and ecol­ogy are inseparable from the
enduring catastrophe of settler colonialism, w ­ hether in their connection to
extractive industries, colonial militarisms, techniques of control developed in
settler states, or the regimes of seeing, knowing, and being that underpin the
­European modernity that has spread unevenly, violently, and with varying

Nonhuman Witnessing 5
degrees of success across the planet. World-­ending crises are all too familiar
for First Nations p ­ eople, who live in what Potawatomi scholar and activist
Kyle Whyte calls “ancestral dystopias,” or pre­sent conditions that would
once have been apocalyptic ­futures.2 But I also want to emphasize that the
subject of History—­the figure that enacts and is produced by such structures
of vio­lence and control—is neither an accident nor a universal figure. Sylvia
Wynter calls this figure Man, the Western bourgeois figure “which overrepre-
sents itself as if it ­were the ­human itself.”3 This imposed image of the ­human
as Christian and ­middle class first emerged in the R ­ enaissance, only to be
amended in biological terms by the sciences of the nineteenth ­century.4 As
I ­will argue ­later in this introduction, it is precisely this figure of Man that
is the unexamined subject of witnessing. Nonhuman Witnessing argues that
this dangerous fiction of Man the Witness cannot hold u ­ nder the dual pres-
sures of existential catastrophe and its own violent contradictions.
The two terms of the main title thus signal the core theoretical interven-
tions and tensions of the book. By putting witnessing and the nonhuman in
conversation with each other, I aim to dismantle the humanist frame within
which witnessing has been understood ­until now. This revisioning of witness-
ing contributes to the larger critical and p ­ olitical proj­ect of interrogating
fundamental assumptions within the Western tradition and its proj­ect of
domination. It also speaks to the necessity of building new methods and
modes of knowing that can grapple with the injurious impacts of algorithmic
enclosure, technowar, and anthropogenic climate change at a time when the
illusion of a cohesive world cannot hold. In ­doing so, Nonhuman Witnessing
aims to be as generative as it is critical: it is a work of thought in action that
seeks new ways of making theory and building concepts.
As an analytic concept, nonhuman witnessing describes the varied mate-
rial, technical, media-­specific and situated relations through which ethicopo­
liti­cal knowledge, responsibilities and forms are produced in ways that can
include but neither require nor privilege h ­ uman actors. As I define and
elaborate the concept, nonhuman witnessing rests on a vitalist conception
of existence that understands technics, affects, and materialities as registering
and communicating experience in forms that can be deemed witnessing
in their own right, prior to and distinct from any semiotic translation or
interpretation. In this, I am indebted to philosophies of radical empiricism
that stress the movement and relationality through which existence takes
shape and meaning. My own intellectual roots are in the pro­cessual vitalism
of Gilles Deleuze, and particularly its incarnation in the heterogenous field
that has come to be known as affect theory. More specifically, my approach to

6 Introduction
relationality borrows from Brian Massumi’s theorizing of affect as intensities
of relation between bodies and worlds, w ­ hether h ­ uman or non, corporeal
or technical, dominant or fugitive.5 Parallel to the emphasis throughout on
relationality, this book also approaches nonhuman witnessing with a debt to
media and cultural studies approaches to media and mediation, as well as
witnessing more specifically.
Yet this book is also indebted to encounters with First Nations cosmo-­
epistemologies that understand animals, plants, rocks, sky, water, and land
as forms of life with inherent—­rather than granted—­rights, agencies, and
relations. Academic scholarship all too easily and often adopts an extractiv-
ist approach to such knowledges. As an uninvited settler living and working
on the unceded land of the Bidjigal and Gadigal ­peoples of the Eora Nation,
in what is now called Sydney, Australia, I engage with ­these knowledges in
a spirit of study without laying claim to traditions that ­aren’t mine. I want to
think and inquire with ­these knowledges, exploring their resonances with the
pro­cessual empiricism that anchors my own scholarly standpoint. My aim is
to show how the exclusion of the nonhuman from witnessing derives from
a distinct and narrow approach to both agency and knowledge, a limitation
that is endemic to the dominant strain of ­European philosophy that insists so
intently on the discrete and unitary over the relational and emergent.
Nonhuman witnessing elevates the status of the other-­than-­human in
bearing witness, refiguring witnessing as the entanglement of ­human and
nonhuman entities in the making of knowledge claims. In the air strike that
killed twenty-­three civilians in Uruzgan, ­Afghanistan, claims to knowledge
about what was happening on the ground w ­ ere animated within the mili-
tary apparatus by the interdependencies of media technics, environmental
conditions, and discursive practices. Vio­lence registers as datalogical and
informational before it is kinetic and lethal: witnessing the event of vio­lence
cannot be isolated to the drone operators or survivors or the infrared sensors,
but rather must be known through the registering of ­those complex relations
within and between ­human and nonhuman entities. Nonhuman witnessing
can be identified in ecological, biological, geological, and even chemical
manifestations, but also in technical and aesthetic forms, such as drone sen-
sor assemblages and machine learning algorithms. This means that nonhu-
man witnessing is inseparable from place, time, media, context, and the other
­human and nonhuman bodies through and alongside which it takes place.
Against the singular world of the scientific or juridical witness inherited from
the Enlightenment, nonhuman witnessing coheres with what Mario Blaser
and Marisol de la Cadena call a “a world of many worlds.”6 This is, then, one

Nonhuman Witnessing 7
meaning of the temporality of this book’s title: a theory of witnessing for a
world of many worlds, ­after the end of the illusion that ­there is only one.
One consequence is that nonhuman witnessing is rife with practical and
conceptual tensions. The very proposition contains within it the irresolvable
paradox of identifying a mode of witnessing that must necessarily exceed
the capacity to “know” inherited from Western epistemologies. Pursuing
witnessing as a relational p ­ rocess, rather than locating it in e­ ither the figure
of the witness or the act or object of testimony, raises the prob­lem of which
encounters constitute witnessing. Where, in other words, is the demarcation
between mere registering and the witnessing of an event’s occurrence? How
is the status of witnessing bestowed and u ­ nder what criteria? Tempting as it
might be to reconcile such tensions or produce checklists of qualification,
seeking to do so risks flattening nonhuman witnessing such that it loses
purchase on the specificity of media, materials, ecologies, technics, and con-
texts. Nor is nonhuman witnessing necessarily virtuous. Just as the soldier
can witness his own slaughter of innocents, so too the algorithmic witness
to drone strikes or environmental vio­lence can be understood as a witness-­
perpetrator. As with all witnessing, t­ here is no inherent justice to nonhuman
witnessing. The task at hand is to ask how nonhuman witnessing pries open
conceptual and practical space within how we h ­ umans do politics, ethics,
and aesthetics.
As a theory of ethical, p ­ olitical, and epistemic formation, nonhuman
witnessing responds to a twofold crisis in witnessing itself. Its humanist
form cannot reckon with the scale, complexity, intensity, and unknowabil-
ity of technoscientific war, algorithmic enclosure, and planetary ecological
catastrophe. Nor can witnessing hold in the wake of the disruption of “the
­human” by ecological, technological, and critical-­theoretical change, not least
­under the pressure of critiques by Black and First Nations scholarship. Faced
with this crisis of witnessing, we are left with a choice: to reserve witnessing
for h
­ uman contexts and find new concepts to address and respond to new
crises, or, as this book argues, reconceive witnessing as entangled with the
nonhuman by attending to registrations and relations in the stuff of existence
and experience. Precisely ­because witnessing is so crucial to ­human—­and
especially Western—­knowledge and politics, t­ here is a strategic imperative
to revising its vocabulary to analyze, strengthen, and generate transversal
relations with the nonhuman that are ethical, p ­ olitical, and communicative,
rather than simply informational or transactional.
This book, then, pursues what nonhuman witnessing is, but also what
nonhuman witnessing does as a concept for crafting knowledge out of which

8 Introduction
a more just politics might be formed and fought for. Rather than provide a
detached and abstract theory, it examines the media-­specificity of nonhu-
man witnessing across a motley archive: the temporal and spatial scales of
planetary crisis, the traces of nuclear testing on First Nations land, digital
infrastructures that produce traumas in the everyday, deepfakes, scientific
imaging that probes beyond the spectrum of the ­human sensorium, algo-
rithmic investigative tools, the unpre­ce­dented surveillance system that is
the global climate monitoring regime, and remote warfare enacted through
increasingly autonomous drones. It combines close analyses of events,
technologies, and ecologies with cultural studies readings of ­political and
creative texts. From poetry to video to sculpture to fiction, creative works
play a critical role in this book ­because they allow me to pursue nonhu-
man witnessing into speculative domains in which aesthetics and worlds
relate to one another in strange, unexpected ways. This approach aims to
show both the media dynamics and cultural consequences of nonhuman
witnessing. In ­doing so, nonhuman witnessing emerges as a relational theory
for understanding and responding to entangled crises, one that attends to
complexity and difference even as it works across divergent domains and
dizzying scales.
Rather than stitching together a g­ rand theory, t­ hese sites reveal the ne-
cessity of capacious, open, situated, and flexible approaches to nonhuman
witnessing. What this book pursues are the resonances, overlaps, and unex-
[103.158.136.64] Project MUSE (2024-02-06 19:57 GMT) University of Edinburgh

pected convergences in the kinds of grounded attachments and imaginaries

that undo the narrow frame of the ­human within which witnessing has been
understood for too long. This is why I am so insistent throughout on the ger-
und form of witnessing rather than witness as e­ ither noun or verb. My prin-
cipal concern is not the figure of the witness as such, although both ­human
and nonhuman witnesses play vital roles in its drama. Nor is this book overtly
focused on the forms that testimony can take, although testimonies of many
kinds occupy its pages. Nor is it about evidence and its forensic articulation,
although such terms are never far away. Rather, I am interested in witnessing
as a relational p
­ rocess, as a vital mode of world-­making that encompasses
both ­human and non.
In d
­ oing so, the book is as attentive as I could make it to differences of
capability and ­process, as well as circumstance. A mountain and a person
possess asymmetric capabilities, with one, for instance, able to endure across
eons and the other able to marshal linguistic resources that facilitate com-
munication with other ­people. Pushed to its most speculative ends, nonhu-
man witnessing might well extend into forms of witnessing that exclude and

Nonhuman Witnessing 9
elude the ­human entirely. It may well be that witnessing to which we are not
party is happening all the time, but of much greater significance are ­those
instances of nonhuman witnessing that seem to be addressed in some way
to the h
­ uman—­and through that address insist on both our response and
responsibility.
Each chapter is ­organized around a double meaning: the witnessing of vio­
lence, as well as the vio­lence that can be done by witnessing; the witnessing
performed by algorithms, as well as the need to witness what algorithms do;
witnessing of more-­than-­human ecologies, as well as ecologies of witnessing;
the witnessing of absence, as well as the absence of witnessing. ­These doublings
perform the relational dynamics of nonhuman witnessing itself, reflecting
its working as both a critical concept and an emergent phenomenon. But
each chapter also elaborates a distinct operative concept for understanding
the pro­cessual modalities of nonhuman witnessing. Chapter 1, “Witnessing
Vio­lence,” critiques the vio­lence of increasingly autonomous warfare as it is
mediated through technology, bodies, and environments, elaborating the
notion of violent mediation as constitutive of martial life. Chapter 2, “Wit-
nessing Algorithms,” pursues machine learning algorithms that produce
techno-­affective milieus of witnessing, articulating an account of the ma-
chinic affects that animate relations within and between technics, bodies, and
ecologies. Chapter 3, “Witnessing Ecologies,” attends to naturecultures ­under
the strain of climate catastrophe and nuclear war, conceptualizing a distinct
form of ecological trauma that ruptures vital relations between ­human and
nonhuman. Chapter 4, “Witnessing Absence,” conjoins the sites of war, al-
gorithm, and ecol­ogy to examine the traumatic absences that circulate in
the quotidian of digital media, developing the concept of radical absence to
show how nonhuman witnessing makes absence intensively pre­sent through
nonhuman infrastructures.
Each of ­these analytic concepts—­violent mediation, machinic affect, eco-
logical trauma, and radical absence—­explicate aspects of the pro­cessual dy-
namics of nonhuman witnessing. But while they intersect with one another
in many ways, they ­don’t snap neatly together to provide a unified theory
of nonhuman witnessing. Th ­ ese concepts instead name the relational pro­
cesses that constitute nonhuman witnessing across dif­fer­ent contexts. Not
all nonhuman witnessing entails violent mediation or radical absence, for
example, but the former plays a crucial role in war while the latter is vital
to understanding how nonhuman witnessing functions in digital cultures.
Throughout the book, I show how t­ hese dynamics converge and diverge in
productive tension with one another, marshalling varied constellations of

10 Introduction
them as they obtain to distinct sites of analy­sis. In the coda, I pull together the
conceptual threads of the book to outline in explicit terms how nonhuman
witnessing enables a more pluriversal politics that foregrounds communica-
tive justice for more-­than-­human entities and ecologies.
Upending the long history in theory and philosophy of reserving witness-
ing for the ­human subject, I argue that witnessing is and always has been
nonhuman. Our con­temporary conjuncture makes this much easier to see,
precisely ­because so much of Western ontology and epistemology has been
thrown into crisis. If crises of autonomous war, algorithmic enclosure, and
environmental catastrophe are indeed converging in the con­temporary mo-
ment, it is surely in no small part ­because their roots reach so deep into the
historical ground of militarism, capitalism, and settler colonialism. Nonhu-
man witnessing thus provides purchase on unfolding catastrophic ­futures,
but also on the catastrophes of the past—­and on the potential for radical
hope, historical acts of resistance, and the making and remaking of more
just worlds.

this mess w
­ e’re in

Amazon​.­com is an avatar for the interlocking crises of algorithmic enclo-

sure, ecological catastrophe, and autonomous warfare. Its recommender
algorithms and automated ware­houses combine with autonomously man-
aged global logistics systems to crowd out small producers and retailers. Its
­drivers and ware­house workers are tracked with biometric sensors, directed
in their movements by algorithmic overseers, expected to meet precisely
defined ­performance metrics, and kept in precarity by zero-­hour contracts
and just-­in-­time rostering, their bodies damaged by the dictatorial rule of
algorithmic management systems. Its smart-­home system, Alexa, provides
voice-­activated access to Amazon’s systems, even as it datafies the fabric of
daily life. More profitable even than its e-­commerce operations and as vital
to the infrastructure of the internet as Google, its Amazon Web ­Services
cloud computing platform powers every­thing from document storage to
facial recognition and is the single biggest s­ ervice provider to the US military.
All this comes at an astonishing cost to the planet. Amazon’s logistics
systems alone produce 51.17 million metric tons of carbon, while its aws
data centers produce roughly the same amount of carbon as nine coal-­
fired plants.7 Like most smart devices, Alexa relies on lithium and rare
earth metals mined at devastating cost to local ecologies and to Indigenous

Nonhuman Witnessing 11
c­ ommunities, such as the Atacan in Chile.8 ­Whether mining users for data
or land for lithium, Amazon is ruthlessly extractive, the high-­tech successor
to the colonial enterprises that coproduced racial capitalism.9 It has both
infiltrated and diverted countless facets of life, and its ­founder dreams of
extending that rapaciousness to the stars. Yet for all this, Amazon still retains
much of the veneer of techno-­utopian solutionism: a frictionless f­ uture of
goods, data, and currency flowing through global infrastructures in which
­human ­labor is obscured, if not erased from view altogether. The vision of a
transcendent ­future built on material waste and ­human sweat far more than
on computation and abstraction.
To state the obvious: Amazon is neither the architect nor the sole ben-
eficiary of the “modern world system of ‘racial capitalism’ dependent on
slavery, vio­lence, imperialism, and genocide,” as Robin D. G. Kelley describes
the current global regime.10 Nor is it the only exemplar of the convergence
of crises to which this book is addressed. Since the turn of the millennium
and the attacks of 9/11, war and military technologies have under­gone dra-
matic transformations, led by the United States but now sweeping across the
globe. Remotely pi­loted aerial systems, or drones, moved from the margins
to become instruments of killing and transform military strategy. ­Today, au-
tonomous and semiautonomous drones are used by more than one hundred
nations for surveillance and by a growing subset for lethal vio­lence, backed by
artificial intelligence systems powered by machine learning neural networks
that undertake real-­time analy­sis of impossibly large streams of remote sen-
sor and other data. Remote vehicles are used on and above ­every type of
terrain, as well as underwater and under­ground. Algorithmic se­lection and
targeting systems for drones and other weapons platforms are already h ­ ere,
with fully autonomous weapons systems already emergent, held back less
by technical capacity than by military, ­political, and public unease with the
notion of removing h ­ uman decision making from the act of killing. Th ­ ese
changes have, as Jeremy Packer and Joshua Reeves point out, transformed
“­enemy epistemology and e­ nemy production” in line with specific media
logics of “sensation, perception, reason, and comprehension tied to a given
medialogical environment.”11 The media-­technological production of en-
emies and knowledge about ­those enemies is itself inextricable from the
determination that certain populations must be controlled or can be killed,
­whether via the debilitating biopolitics that Jasbir Puar calls “the right to
maim” or in the necropolitics of remote warfare with which I opened this
book.12 Consequently, their martial media technics must be read within the
context of enduring colonialism.

12 Introduction
This martial transformation has been part and parcel of the wider enclo-
sure of life within computational systems and communications technologies,
­whether at the macroscale of public health databases, citizen registers, and
biometric surveillance, or at the personal level with the ubiquitous presence
of social media and smartphones across the planet. Logics of surveillance
and control that have crept into ­every dimension of social, ­political, and
economic life are also deeply entwined with histories of anti-­Black racism,
and the methods of domination applied during and ­after slavery in the Amer­
i­cas, as Simone Browne persuasively shows.13 Indebted to ­wartime initiatives
of the 1940s, ­decades of Cold War arms racing and antagonistic cultural
politics that legitimated significant military spending in the United States,
and active partnerships between the Defense Advanced Research Proj­ects
Agency (darpa) and what would become Silicon Valley, ­today’s communica-
tions technologies also bear the legacy of cybernetics and the effort to craft
“infrastructures of sensing and knowing,” as Orit Halpern puts it.14 At the
1970 World Exposition in Osaka, experimental multimedia environments
­were built to demonstrate the potential for actualizing cybernetic systems
in urban architecture and planning. Reflecting on the influence of Expo ’70,
Yuriko Furuhata argues that “regulatory mechanisms of policing and surveil-
lance, modeled as multimedia systems and aided by networked communica-
tions, form a much darker and somber counterpart to the types of artistic
multimedia environments that emerged in the 1960s.”15 ­Japanese architects,
theorists, and multimedia artists played a crucial role in this dynamic, inher-
iting and responding to a dif­fer­ent colonial legacy of vio­lence and control.
Algorithmic technologies are now embedded in every­thing from Ama-
zon’s purchase recommendations to the creation of art, from the mining
of personal and population data to the provision of welfare ­services to the
structuring of knowledge itself via the search results of Google. But what
Paul Edwards calls the “closed world” of Cold War computation also laid the
infrastructural foundations for the “vast machine” of atmospheric monitor-
ing that allowed anthropogenic climate change to become more vis­i­ble and
better understood, even as it became both contested and irreversible.16 ­Today,
ecosystems reel from hotter summers, extreme weather events, failing crops,
rising migration, ocean acidification, and atmospheric pollution, to name
but a handful of the more striking effects. W ­ hether marked in the geology
of the planet or in the biosphere, the sheer scale of ecological crisis (which
is r­ eally a set of interlocking crises) is its own catastrophe, leading to deni-
als of scientific knowledge, failures of politics, and global paralysis around
meaningful response.

Nonhuman Witnessing 13
Each alone would be more than enough to end countless worlds, but
­these three crises are also intensifying and accelerating, fueling and fueled
by the insatiable expansion of racial capitalism. Advances in machine learn-
ing have supercharged both algorithmic enclosure and autonomous warfare.
Reliance on mass computing in every­thing from image recognition to bitcoin
mining has combined with an exponential expansion in digital data stored
in servers and trafficked across networks to produce a huge carbon footprint
for computation. Built into the bedrock of the civilian internet as the host
of every­thing from ebooks to presidential election campaigns to banking,
­those infrastructures have a massive environmental impact in heat gener-
ated and fossil fuel consumed.17 ­Those same fossil fuels, of course, power
the energy appetite of the US military, the world’s largest carbon polluter.
Institutionally, eco­nom­ically, and ecologically, Amazon and its ilk are deeply
integrated with military apparatuses, especially in the United States where
big tech provides every­thing from enterprise software to cloud storage to
strategic guidance through bodies such as the Defense Innovation Board,
chaired by ex–­Google boss Eric Schmidt. Equivalent dynamics operate at
­every level, ­whether in the shared reliance on remote sensors by militarized
drones, urban surveillance, and environmental monitoring, or the centrality
of extraction to climate change, military industries, and the mining of data.
Despite this tight bind between technology, war, and climate change,
ever-­more innovation is proposed as the only solution by self-­interested
luminaries such as Bill Gates. In the most basic material sense, ­these crises
of war, data, and climate and the system of racial capitalism they maintain
and depend on are drawing down the finite resources of the planet. Taken to-
gether, they are both product and perpetrator of vio­lence, w ­ hether structural
or infrastructural, environmental or military, algorithmic or interpersonal,
kinetic or slow.18 The very existence of such lists speak to both the ubiquity
and variety of vio­lence ­today and its intimacy with crisis as the condition of
life for much of the planet. The explanatory force of nonhuman witnessing
resides in part in its capacity to register and communicate t­ hose forms of
vio­lence that might other­wise be rendered invisible.
Galvanizing the language of crisis, as I have done so far, is not without risk.
As Whyte argues, claims of crisis—of food, resources, space, security—­have
been frequently used to justify colonialism, both in the larger sense of the set-
tler enterprise and in specific instances such as the corporatization of tribal
governance in the United States as a response to an “emergency” of poverty.19
For Whyte, “crisis epistemologies” produce problematic politics that over-

14 Introduction
ride First Nations concerns, such as in the appropriation of tribal lands for
wind farms and other renewable initiatives in response to the exigencies of
the climate crisis. Such epistemologies depend on a conception of crisis as
aberrant and abnormal, a rupture that must be tamed and contained so that
the normal order of ­things can be restored. Rather than a break from order,
crisis is better understood as a condition of existence. “Crisis is not rupture, it
is fragmentation,” writes Henrik Vigh, “a state of somatic, social or existential
incoherence.” As such, crisis is “not a short-­term explosive situation but a
much more durable and per­sis­tent circumstance.”20 It is not an event, but the
condition and context of life. Lauren Berlant calls this “crisis ordinariness,”
in which “crisis is not exceptional to history or consciousness but a ­process
embedded in the ordinary that unfolds in stories about navigating what’s
overwhelming.”21 Thinking about crisis in this way does not require an aban-
donment of the notion of rupture. But crisis as condition does demand that
we see rupture, trauma, vio­lence, dispossession, precarity, and vulnerability
as at once pervasive and unevenly distributed. Crisis d ­ oesn’t punctuate time,
so much as shape its passage, lacking any distinct beginning or end, enfold-
ing past and ­future.
Crisis also enfolds and consumes events, entangles bodies, intensifies
the contexts of their occurrence, and cuts through forms of connection to
impose new (dis)­orders. Andrew Murphie calls this catastrophic multiplicity
“a complex storm of feeling, of aspects of world feeling each other in intense,
unexpected and constantly mutating ways.”22 Catastrophic multiplicity inten-
sifies, bewilders, and numbs feeling, which makes thinking with and through
prob­lems difficult, if not impossible.23 Knowledge-­making as a collective
endeavor becomes fraught and frayed. This generalized crisis environment
provides fertile conditions for states to harness ontopower, the power to bring
into being. ­Because ontopower targets life as it stirs into activity, it is a form
of power that both exceeds and precedes the h ­ uman. Massumi describes it
as the “power to incite and orient emergence that institutes itself into the
pores of the world where life is just stirring, on the verge of being what it
­will become, as yet barely ­there.”24 Ontopower operates at the pro­cessual
level of becoming itself. Deploying technoscientific apparatuses of war and
governance, states and other actors seek to harness ontopower in attempts to
preemptively control the f­ uture, as in the drone strike ordered in response
to the algorithmic analy­sis of phone calls and patterns of movement that
produce a “signature” deserving of eradication. But in d ­ oing so, ontopower
also produces crises that themselves escape control, through its continual

Nonhuman Witnessing 15
animation of the forces of state vio­lence, environmental extraction, and
algorithmic control. In this sense, ontopower does not replace biopower or
necropower but rather operates in concert with them
If I have drawn so many examples from martial contexts, this is b ­ ecause
Nonhuman Witnessing finds its way into data and climate through war. Like
the ­political theorist Jairus Grove, I take the view that war is a form of life as
much as it is a means of death: terrible, ruinous, and endlessly destructive,
yet also generative and creative. Applied to geopolitics and indeed to every­
thing from racism to capitalism, “war is not a ­metaphor; it is the intensive
fabric of relations” that form this historical era.25 What is needed is analy­sis
“characterized by inhuman encounters and deep relational pro­cesses across
geo­graph­i­cal scales rather than a form of p ­ olitical thinking that relies on
discreteness, causality, and an exceptional notion of ­human agency.”26 Also
like Grove, I am committed to decentering h ­ uman actors, but not d ­ oing away
with h ­ uman responsibility for the vast assemblages that continue to cause so
much damage. As concept, practice, and phenomena, nonhuman witness-
ing brings such encounters, pro­cesses, and scales into conjunction with the
relational formation of knowledge and subjectivity. But it does so through
committed attention to the pro­cesses of mediation that animate and bind
together crises of war, data, and climate.
Lively, temporal, and always in flux, mediation is never foreclosed or
­limited in its potential. Media studies scholarship has much to say on media-
tion. Sean Cubitt calls it the “effervescent commonality of ­human, technical,
and natu­ral pro­cesses.”27 For Sarah Kember and Joanna Zylinska, mediation
is crucial to “understanding and articulating our being, becoming with, the
technological world, our emergence and ways of intra-­acting with it, as well
as the acts and pro­cesses of temporarily stabilizing the world into media,
agents, relations, and networks.”28 In this sense, mediation is always rela-
tional, but it is also necessarily nonhuman: even the witness who speaks
their testimony entails the mediation of air so that wavelengths of sound
can carry from lips to ears. This vitalist understanding of mediation requires
an expansive understanding of media forms, one that sees every­thing from
clouds to usb drives to the planet itself as media.29 In keeping with the crucial
work of feminist scholars, this approach to mediation is avowedly material.
As Cubitt argues, “Media are finite, in the sense both that, as m ­ atter, they are
inevitably tied to physics, especially the dimension of time; and that their
constituent ele­ments—­matter and energy, information and entropy, time and
space, but especially the first pair—­are finite resources in the closed system
of planet Earth.”30 Crises of war, algorithm and ecol­ogy are thus also crises

16 Introduction
of media: of an accelerating consumption that only exacerbates all other cri-
ses. In the face of just such a trajectory, Cubitt calls for a renewed and more
differentially attuned mode of communication, one that resists the tendency
to extract information from nature but not speak back to it. Something like
this might be found in the radical empiricist tradition, which Chris Russill
argues offers an alternative intellectual history to communication theory
via William James, John Dewey, and George Herbert Mead that embraces
indeterminacy, incommensurability, and difference.31 Nonhuman witnessing
describes a critical concept and relational practice of a distinct mode of com-
munication, one constituted by an address that demands response but still
embraces opacity. It is a transversal opening onto the workings of vio­lence,
experiences of precarity, and the shattering of epistemologies; an aperture
through which communication might take place in ways that are necessary
for care and justice in the aftermath of ended and ending worlds.

­after the end of the world

Words that would become this introduction ­were first written amid bushfires
that ravaged Australia in the summer of 2019 and then labored over in the
long years of the pandemic. Throughout that summer of smoke and ash,
the sun glowed pale red and the density of particulate m ­ atter made the air
hazardous to breathe. More than a billion animals died, thousands of homes
­were lost, countless habitats erased. Across traditional and social media, in
corridor conversations and at dinner parties, all the talk was about apoca-
lypse, climate change, the failure of normal politics to do much of anything
at all. As the pandemic took hold in early 2020 and then wore on through
the years, life ­here began to come undone, but the fabric never tore so deeply,
so devastatingly, as it did across much of the globe. With Australia’s borders
closed for well over year, the sense of an ending world was impossible to
escape, even without the massive loss of life experienced in so many places
and borne so disproportionately by the already vulnerable and precarious.
The very networks of travel and trade that expanded “the world” to fill “the
globe” ­were now a threat to its continuation. What worlds would remain in
the aftermath?
Living and working on unceded and sovereign Aboriginal lands, I am
enmeshed in ended and ending worlds. Colonial expansion ended the worlds
of First Nations p ­ eoples in Australia long ago, beginning with the arrival
of Captain James Cook in 1770 and ­eighteen years ­later with the landing of

Nonhuman Witnessing 17
the First Fleet at Botany Bay, just a few bends of the coast south of my own
home. The lines of my own ­family are bound up with that dispossession, if not
at the point of a gun then through the construction of buildings, founding of
museums, plying of trade, and s­ ervice in the military. As my forebears settled
this land and built lives and families, the Traditional ­Owners experienced
massacre, epidemic, dispossession, incarceration, starvation, and the stealing
of ­children and the breaking of kinship formations.32 That ending of worlds
continues t­ oday, even as Aboriginal p ­ eople endure and resist in power­ful,
inspiring, and even beautiful ways. Preoccupations with an apocalypse that is
yet to come have a ­bitter irony in a place where First Nations have spent two-­
and-­a-­half centuries surviving the end of the world, struggling for new and old
ways of living in this place that always was and always ­will be Aboriginal land.
­After the end of the world: it is a temporality both commonplace and
strange. In Western ­popular culture, apocalypse has been in the air and on
the screen and page: zombies ­running amok, asteroid strikes, ai takeovers,
bioengineered crashes, alien invasions. ­Metaphors of late capitalism, or cli-
mate change, or global migration, t­ hese end-­times imaginaries are no longer
the preserve of niche subcultures or millenarian religions but at the heart
of the most profitable, most mainstream forms of ­popular culture. But the
estrangement felt from ­these imaginings, the lure of catharsis in the fictional
experience of the end of the world, relies on being situated in relation to a
specific telling of history. As Whyte points out, “The hardships many non-­
Indigenous ­people dread most of the climate crisis are ones that Indigenous
­peoples have endured already due to dif­fer­ent forms of colonialism: eco-
system collapse, species loss, economic crash, drastic relocation, and cul-
tural disintegration.”33 In this sense, the temporal location in the title of the
book—­­After the End of the World—­describes a shifting, situated temporality
that hinges on whose world has ended, to what purpose, and by what hands.
As Nick Estes so succinctly makes clear in describing the impact of the Pick-­
Sloan Dam on the Oceti Sakowin ­peoples of Dakota in the early twentieth
­century, “taking away land and ­water also took away the possibility of a ­viable
­future.”34 Now, that ending of worlds has come to the world enders, the colo-
nizers and empire builders who ­imagined into being a singular, global world
and made it so with the r­ ifle, the slave ship, the ledger, and the plantation.
Now, de la Cadena and Blaser write, ­there “is a new condition: now the colo-
nizers are as threatened as the worlds they displaced and destroyed when
they took over what they called terra nullius.”35 And yet ending worlds d ­ on’t
always fully end and can be reseeded, as the resilience and endurance of First
Nations p­ eoples across the planet makes clear.

18 Introduction
Naming this era is no s­ imple ­matter b­ ecause to name the prob­lem is also
to diagnosis it. Since its popularization by the atmospheric biochemist Paul
Crutzen and ecologist Eugene Stoermer in a short article from 2000, the term
Anthropocene has been widely ­adopted.36 While the label is useful ­because
it registers the impact of colonialism and industry on the planet’s biological
and geological systems, it also risks universalizing and misdiagnosing the
prob­lem by naming an undifferentiated Anthropos as the causal agent.37 In
this it serves an ideological function: flattening responsibility onto the ­human
in the broadest sense both hides the histories of extraction, pollution, and
vio­lence through which the planet has been transformed and obscures the
grossly unequal distribution of the spoils. Critics rightly argue that the term
Anthropocene risks occluding the originary vio­lence of settler colonialism,
without which our era of petrocarbons, plastics, terraforming, species loss,
and ocean death might never have been pos­si­ble at all. Alternatives now
abound, many of which attempt to name precisely distinct causal agents:
Capitalocene, Plantationocene, Eurocene.38 For me, deploying the term An-
thropocene is a necessary strategic decision despite its limitations. Sticking
with the Anthropocene allows me to center the Anthropos, understood as the
form of Man that has driven colonial and cap­i­tal­ist expansion and, crucially,
laid claim to the normative figure of the witness.39 Conceived in this way,
the Anthropocene and Man are co-­constitutive. Countering the idea that
the Anthropocene begins with the Industrial Revolution or nuclear bomb,
Heather Davis and Métis scholar Zoe S. Todd argue that “placing the golden
spike at 1610, or from the beginning of the colonial period, names the prob­
lem of colonialism as responsible for con­temporary environmental crisis.”40
Known as the Columbian Exchange, 1610 marks both the moment when
the exchange of biomatter between ­Europe and the Amer­ic­ as reshaped
ecosystems and when carbon dioxide levels dropped in the geologic layer
as a consequence of colonial genocide. Dating the Anthropocene in this
way ties it both conceptually and historically to Man, and to the ending of
worlds that is such an essential dimension of settler colonialism and racial
capitalism.
Situating this book a­ fter the end of the world is thus a conceptual claim,
as well as a historical one: the world has long since lost any claim to de-
scribe the totality of being. In its place are countless worlds without claim
to universality or unity. One of the ways in which the end of the world finds
hope is in recognizing that the world has always been multiple, a pluriverse
produced by the world-­making power of countless knowledge systems. Such
a multiplicity enables what Kathleen Stewart calls worldings, or the “intimate,

Nonhuman Witnessing 19
compositional ­process of dwelling in spaces that bears, gestures, gestates,
worlds.”41 Reflecting on war and its aftermaths, Caren Kaplan writes of the
“disturbance of conventions of distance and proximity, the presence of many
pasts and places in what we try to think of as the ­here and now” that make
“modernity’s everyday aftermaths—­the undeclared wars that grieve not only
the pre­sent absences but the absent pre­sents—­not so much a m ­ atter of ghosts
as multiple worlds that a singular worldview cannot accommodate.”42 The
unruly intensities and haunting disruptions of t­ hese martial aftermaths are just
as evident in the wake of ecological vio­lence, technological enclosure, and
colonial dispossession: time, place, space, experience and thought all resist
linearity, refuse o
­ rganization, unsettle the unfolding of life.43 As a form of
worlding ­after the end of the world, nonhuman witnessing is one means of
building a communicative politics that begins with ecological relations and
the inherent agencies of nonhuman ­things, animals, and places.44

witnessing and the nonhuman

As crises expand, intensify, and intersect, the capacity of witnessing and

testimony to respond has been amplified, multiplied, and diversified by the
adoption of new (and sometimes old) technologies, techniques, practices,
knowledges, and theories. Open-­source investigations led by agencies such
as Bellingcat, Airwars, and Forensic Architecture have shown how crowd-
sourcing, computational tools, 3d modeling, data analy­sis, remote sensors,
and other technologies and methodologies can be combined with situated
testimonies to generate alternative accounts of state and corporate vio­lence.
Satellites and drones provide h ­ uman rights and environmental monitors
with rich data that extends and exceeds the perceptual capacity of h ­ umans
in scale, vantage point, and visibility across a much wider band of the light
spectrum. Smartphones and social networks bring a far wider array of voices
and images to public attention, shaking the epistemic dominance of tradi-
tional media institutions. Cheaper and more accessible sensing technologies
have enabled citizen-­led proj­ects to monitor local ecologies. Growing recog-
nition within the scientific community about the communicative capacity of
plants and ecologies more broadly resonates with the push by First Nations
activists to have nonhuman entities recognized by state law, such as the suc-
cessful attribution of sentience to the Whanganui River in New Zealand
following more than a c­ entury of strug­gle by local Māori tribes, led by the
Ngāti Hāua. Artists, poets, activists, and creative prac­ti­tion­ers of all stripes

20 Introduction
now engage with technics, ecologies, and politics in a testimonial mode that
entangles h ­ uman and nonhuman actors.
Nonhuman Witnessing conceptualizes and theorizes t­ hese developments,
both as a means of making sense of ­these changes in situ and to connect
them into a larger proj­ect of reckoning with crisis, vio­lence, and trauma. It
joins a growing body of critical interventions into the connections between
aesthetics, witnessing, and forensics, prominent among them the ­legal, artis-
tic, and theoretical works of Eyal Weizman and his research agency ­Forensic
­Architecture, located at Goldsmiths, University of London. Weizman’s ­Forensic
Architecture theorizes the application of architectural techniques of siting,
sensing, mapping, modeling, and analyzing to the task of uncovering and
communicating “vio­lence at the threshold of detectability.”45 Attending to
material architectures, media objects, and situated testimonies, forensic ar-
chitecture is an operative concept that provides a method for investigation.
How that method articulates with wider transformations is the subject of
Weizman and Matthew Fuller’s Investigative Aesthetics, which explores how
resistant investigations assem­ble aesthetically to produce what they call an
“investigative commons” to challenge state-­and court-­sanctioned knowledge
production and ­counter the post-­truth “anti-­epistemologies” of misinforma-
tion and disinformation that have undermined trust in shared realities.46
Aesthetics in their terms comprises both sensing and sense-­making, and, as
such, is not exclusively ­human but rather found across all entities in their
relational milieus, as I explore in more detail in chapters 1, 2, and 3, includ-
ing with a close reading of the Forensic Architecture proj­ect ­Triple Chaser.
More closely attuned to the questions of witnessing that occupy this book,
Susan Schuppli’s Material Witness combines reflections on her artistic prac-
tice and work with Forensic Architecture, which draws on archival and eth-
nographic research to develop an account of how m ­ atter can obtain standing
as a witness within public fora such as war crimes tribunals. Her material
witnesses are “nonhuman entities and machinic ecologies that archive their
complex interactions with the world, producing ontological transformations
and informatic dispositions that can be forensically decoded and reassembled
back into a history.”47 Material witnesses can express themselves through a
technical sensibility rather than speech per se, but “­matter becomes a mate-
rial witness only when the complex histories entangled within objects are
unfolded, transformed into legible formats, and offered up for public consid-
eration and debate.”48 Material witnesses appear throughout this book, but
particularly in chapter 3 when I turn to the material traces of nuclear testing
and their mediation through art.

Nonhuman Witnessing 21
While Schuppli, Weizman, and Fuller ground their analy­sis in their own
investigative practices in and beyond the acad­emy, Pugliese’s Biopolitics of the
More-­Than-­Human shares this book’s imperative to develop an apparatus for
critiquing con­temporary warfare and the ruin it does to bodies and ecologies.
Discontented with existing practices of evidentiary analy­sis, Pugliese calls for
a “forensic ecol­ogy” that can “examine the physical remains, in par­tic­u­lar, of
more-­than-­human entities left in the aftermath of the vio­lence and destruc-
tion unleashed in militarized zones of occupation.”49 This is resonant with
the investigation of drone warfare and its violent mediations in chapter 1,
particularly in thinking through the entanglements of technics, bodies, and
ecologies.
Witnessing is also an impor­tant subfield of inquiry within media studies,
producing nuanced empirical and theoretical accounts of distinctive modes
and practices of witnessing and testimony. In an influential essay, John Dur-
ham Peters defines witnessing as “responsibility to the event” and points out
that media must wrestle with the “ground of doubt and distrust” that distance
adds to the “veracity gap” inherent to the relay of any testimony.50 Building
on this conception, Paul Frosh and Amit Pinchevski propose the concept of
“media witnessing,” or “witnessing performed in, by and through media” as
essential to con­temporary world-­making.51 Media witnessing, Lilie Chou-
liaraki argues, is a fraught proposition, veering easily into spectatorship as
distant audiences are presented with atrocity to which they have few or no
ave­nues of response.52 New witnessing practices emerged in concert with
new media technologies, producing what media studies scholars have vari-
ously called mobile witnessing, citizen-­camera witnessing, crowd-­sourced
evidence, digital witnessing, witnessing databases, and data witnessing.53
­These practices have enabled affected individuals and communities to nar-
rate crises in culturally distinctive ways and to self-­represent their witness-
ing, even if they have also produced new expert and intermediary functions
for h
­ uman rights o ­ rganizations.54 Throughout Nonhuman Witnessing, this
research provides valuable insights into distinct witnessing practices related
to my lines of inquiry, but also serves as a springboard for thinking past the
limits of the ­human in ways that I hope ­will in turn be generative for scholar-
ship in media studies.
The works highlighted in the preceding pages share with mine a com-
mitment to interrogating the shibboleths of testimony, evidence, and
their relation to politics, technology, and justice. But t­ here are also criti-
cal departures. Where Weizman elucidates an existing practice of forensic
architecture, this book theorizes a more expansive, ontoepistemological

22 Introduction
reconception of witnessing as an encounter with and response to vio­lence.
Where Fuller and Weizman focus on the theory and p ­ rocess of investiga-
tion as a mechanism for assembling aesthetics, this book attends to how the
sensing and sense-­making of aesthetics produces a witnessing relation that
is not dependent upon an investigative team, method, or apparatus. Where
Schuppli insists on contestation within public fora as a condition for material
witnessing, my approach to nonhuman witnessing insists on witnessing as
an experiential relation that can produce contestation but is not dependent
on it for its existence or even politics. Where Pugliese centers the law and its
enmeshment with military power and colonial structures, my concern is with
pro­cesses distinct from the juridical domain, and that fail to appear or cohere
within ­legal frames. Where media studies research delves into the complex
ensembles of media and h ­ uman that produce distinct forms of witnessing, it
reserves ethical and ­political standing for h­ uman witnesses, intermediaries,
and audiences and leaves nonhuman agencies largely out of frame. In short,
Nonhuman Witnessing contributes to an active proj­ect within critical thought
in which debates over key concepts remain vibrant. And while the forms of
vio­lence and modes of intervention with which all t­ hese works are concerned
are largely new, they are also embedded in a long history of transformation
in the forms and practices of witnessing, who counts as a witness, and how
shared knowledge is produced.
In the earliest foundations of the Western ­legal tradition in Athens and
Rome, the wounded body was considered the most reliable witness, which
meant torture was central to l­egal proceedings. Who could be tortured in
the name of truth was a m ­ atter of importance: the enslaved w­ ere often the
subject of torture to provoke truthful testimony, not the power­ful and prop-
ertied.55 Witnessing was borne on the body up ­until the Enlightenment, when
the law of proof emerged in conjunction with the ocular revolution of the
­Renaissance and the humanist conception of the dignity of Man.56 In 1846
the United Kingdom abolished the law of the deodand, a relic of old E ­ nglish
jurisprudence that held that an object in motion that has killed a ­human must
be held to account. Consequently, writes Su Ballard, “where once they ­were
able to take responsibility for the harm they have caused, now objects are
just another group of silenced witnesses.”57 This sentencing of the memory
of objects to evidence accompanied the modern juridical witness taking
familiar form: structured by norms, ordered in narrative, and verified by
accompanying evidence.58
The figure of the witness thus becomes synonymous with Man, which
meant certain bodies w ­ ere again excluded: the enslaved, Indigenous and

Nonhuman Witnessing 23
Black p ­ eople, and, often, w
­ omen and the unpropertied. Unable to become
witnesses before the law due to explicit rule or fear of retaliation, their flesh
could be made to speak through violent punishment. Hortense Spillers calls
the flesh that “zero degree of social conceptualization,” left ­behind in the
“theft of the body” that occurred in transatlantic slavery and Indigenous
dispossession: “a willful and violent (and unimaginable from this distance)
severing of the captive body from its motive ­will, its active desire.”59 Without
­will or body, the enslaved and First Nations ­were rendered illegible to the
law as persons, figured as property or inhuman objects. As the philosophi-
cal underpinning of imperial and settler colonialism, Man depended on the
construction of Black Africans as the ultimate other, the slave, and the assimi-
lation of all dark skinned ­peoples into the category of “native” as the negative
inversion of the i­magined normal h ­ uman.60 As such, they w­ ere also denied
witnessing before the law, refused the right to attest to the vio­lence done to
them.61 Thus the humanist figure of the witness fused new notions of the in-
dividual, unitary subject of rights and responsibilities with existing regimes
of humanity and inhumanity. But it also carried the legacies of ­monotheistic
religion, in which the figure of the witness claims intimacy with the divine.62
While the testimony of preachers figures prominently in American religious
culture, the martyr or blood witness is rooted in the early years of Chris­tian­
ity and carries through—if in radically dif­fer­ent ways—to the pre­sent in the
dead of Auschwitz and the suicide bombers of isis.
But the Enlightenment and its rearticulation of Man also produced a
new and divergent form of witnessing, one that emerged in the eigh­teenth
and especially into the nineteenth ­century as markedly ­free from overt ties
to vio­lence and law. With the invention of the scientific method and the
establishment of practices of experimentation and observation, science and
scientists both in­ven­ted and claimed mastery over the natu­ral world through
the production of knowledge about it. As Lorraine Daston and Peter Galison
cata­log, the emergence of a new “epistemic virtue” of scientific objectivity was
a complex ­process related to transformations in perspective, understandings
of self, and much more.63 Within this framework, the scientist bears witness,
and it is upon their testimony that knowledge builds. Hypothesis, experi-
ment, rec­ord, replication, verification, peer review, and scholarly publica-
tion built normative guard rails to ensure objectivity, like the swearing of
an oath in court.64 But the scientific witness depended on a host of erasures.
­Women ­were excluded, as was embodiment, in the invention of an affectless
and cultureless objectivity.65 Haraway writes that this “gentleman-­witness”
becomes “the legitimate and authorized ventriloquist for the object world,

24 Introduction
adding nothing from his mere opinions, from his biasing embodiment.”66
By constructing expert knowledge divorced from opinion and transcendent
authority alike, the scientist—by default white and male—­became endowed
with “the remarkable power to establish facts. He bears witness: he is objec-
tive; he guarantees the clarity and purity of objects.”67 This is the figure of the
witness capable of the “God-­trick” of scientific rationality, which claims an
objective, ahistorical, and unbiased viewpoint on the world.68 This modest
witness wins his authority through the performative disavowal of power, and
in d
­ oing so entrenches science—­new though it is—as the authoritative mode
of apprehending the world. Against the rich multiplicity of worlds that jostled
and warred with one another, this new science and its modest witnesses re-
made the world as a singular, knowable t­ hing, conquered by colonialism and
made profitable by capitalism.
If modern science heightens the power of Man the Witness, then the
roughly concurrent emergence of print and then technical media amplifies
and extends that authority in time and space, even as it enables new forms
and practices of nonhuman witnessing. Media technology had always been
bound up with witnessing—­consider Moses, who descends from Mount
Sinai with the word of God engraved in stone—­but the advent of modern
communications made bearing witness a form of informational sociality
around which shared truths form. No longer a ­matter for courts, churches,
and laboratories alone, witnessing through the printing press, telegraph, and
radio ­imagined nations into being and rendered distant events immedi-
ate. No surprise, then, that media studies has had so much to say about
witnessing. For John Ellis, t­ elevision had an even more profound effect on
witnessing by placing the viewer in the position of the witness.69 Mass media
made witnessing, as Frosh and Pinchevski put it, a “generalized mode of re-
lating to the world.”70 But this proliferation of media witnessing amplified the
“veracity gap” that must be bridged to grant the media narrative its author-
ity as truth, as John Durham Peters explains.71 Liveness, that new quality of
televisual media, stood in as truth’s guarantor: How could what is unfolding
now before one’s very eyes be anything but truth? Yet liveness is no guarantor
of the complete picture or the reliability of the witness, nor even—as I w ­ ill
show in chapter 2’s examination of deepfake technologies—of the existence
of the witness. Liveness, like all media coverage of suffering and vio­lence,
can produce spectatorship that dispels action rather than spurs it, present-
ing mere seeing as sufficient response.72 Still, media witnessing is often not
intended to spark action; its purpose is to bind communities around shared
understandings of events, such as the world-­shattering nature of the 9/11

Nonhuman Witnessing 25
attacks for Amer­i­ca and much of the West, or the extended intractability
of the covid-19 pandemic. Increasingly, this binding takes place not only
through the consumption of images, but also through actively participating
in their production and circulation.
In both science and media, witnessing serves as a sociotechnical appara-
tus that refracts experiment into authority, reportage into truth, science and
broadcasting into power.73 In the twentieth c­ entury, a shift took place from
transcendental knowledge, continuous media, and analogue technologies
to mathematical grids and models, discrete media, and statistical technolo-
gies.74 In The Practice of Light, Cubitt argues that the emergence of technical
media requires and constitutes a transformation in the pro­cesses through
which (especially visual) media are produced and the under­lying epistemic
framework.75 Enumeration, probability, and statistical inference and analy­
sis take hold, backed by mathematical theories of information and markets.
With the arrival of the postwar datalogical turn and the claims to potential
omniscience that flow from a seeming infinitude of information, the “com-
municative objectivity” of the cybernetic revolution documented by Halpern
began to bind both science and governance ever more tightly to networked
systems and screen interfaces. Networked computation applied to a data-
fied world produced a new kind of observer, one who followed the rules of
the new cybernetic order but saw the world through increasingly inhuman
modalities of perception.76 The witness as cyborg, harnessing and harnessed
to new technologies of vision began to shape how data was presented and
deployed.77 But it also signaled a deeper infiltration and extension of h
­ uman
perception and action via machine. This technological transformation laid
the foundation for smartphones, drones, remote sensors, and even artificial
intelligence to become instruments of witnessing, even as they transform the
relationship between witnessing and the ground truth against which it is so
often ­measured.78
What t­ hese changes in media and mediation make clear is that witness-
ing is a relational p­ rocess that probes, exposes, and undoes the limits of
repre­sen­ta­tional modes of knowing and being.79 Rather than reinstantiating
the authority of the unitary subject or even of language, con­temporary wit-
nessing exposes the primacy of relations between bodies, events, environ-
ments, worlds, and objects, even if they are obscured, denied, disavowed, or
absent. While testimony might take the form of language or a fixed image,
the experience of witnessing is always affective, occurring in the encounters
through which bodies and worlds emerge within and alongside one another.
Witnessing, writes Kelly Oliver, is “the heart of the circulation of energy

26 Introduction
that connects us, and obligates us, to each other.”80 But now witnessing must
reckon with the unravelling of the ontological and epistemological grounds
of knowledge by radical theory on the one hand and the interlocking crises
of the con­temporary world on the other.
In an evocative, searching essay on the relation between testimony and
the witness, Michal Givoni writes that rather than an age of testimony, “ours
is an era of becoming a witness, a time in which individuals are called, in
greater numbers and intensity and at a growing rate, to fashion themselves
as witnesses, while their witness position is never guaranteed and their mode
of witnessing is questioned.”81 If becoming-­witness is the task set for the
­human, then what of the agencies that make up more-­than-­human worlds?
If we shift the ­angle with which we approach witnessing and the ­human, the
scene might be dif­fer­ent: Could we not think of witnessing as yet another
pressure applied to the h ­ uman, another dissolving agent working to undo
the narrowly inscribed figure of knowing and being that has both enabled
remarkable advancement but also done terrible, enduring, and world-­ending
vio­lence? Or, to put this differently, what if it is not only ­today’s insistent
presence of the nonhuman that demands a new understanding of witnessing,
but that witnessing carries within itself an unrevealed history, a constitutive
nonhumanity?
This choice to bring witnessing into conjunction with the “nonhuman”
rather than the more-­than-­, post-­, in-­or even de-­human was not easily ar-
rived at. For me, nonhuman emphasizes distinction and difference from the
­human, but retains its necessarily entangled relation to the ­human and thus
asserts the necessity of keeping the h
­ uman in the frame.82 As Richard Grusin
observes, “The ­human has always coevolved, coexisted and collaborated
with the nonhuman,” and, as such, “the ­human is characterized precisely
by this indistinction from the nonhuman.”83 The h ­ uman is, in this sense,
constitutively dependent on complex relations with the nonhuman. This
relationality is central to moving to conceptualize nonhuman witnessing,
since witnessing itself is a relational practice. But I also find the nonhuman
beneficial b
­ ecause it implies no time before, a­ fter, or beyond the h­ uman.84
“Nonhuman” thus avoids the potential to read post­human as an uncritical
desire to move “beyond the ­human,” as Zakkiyah Iman Jackson puts it, which
can be an impossible endeavor for t­ hose never fully afforded the category
of ­human to begin with, and who might not now wish to receive it, even if
only in passing.85 As Karen Barad points out, attending to the nonhuman
“calls into question the givenness of the differential categories of ‘­human’
and ‘nonhuman,’ examining the practices through which ­these differential

Nonhuman Witnessing 27
bound­aries are stabilized and destabilized.”86 As such, Dana Luciano and
Mel Y. Chen argue that “the nonhuman turn marks, for many critics, not a
venture ‘beyond’ the h ­ uman but a new mode of critical realism, a recognition
that the nature of ‘real­ity’ itself is changing as power moves away from the
individual.” D­ oing so has material consequences.87 For Shela Sheikh, “where
care for both ­human and nonhuman life is at stake, witness collectivities
necessarily entail an expansion beyond the category of the ­human.”88 This
questioning of categories, bound­aries, and differences is not only a ­matter of
language, but of the affects, materialities, and mediations of forces, bodies,
meanings, experiences, energies, and ecologies.
In this light, nonhuman should not be read as a dismissal of the related
terms outlined h ­ ere, nor as a disavowal of the species we call h
­ uman as a key
locus for the strug­gle for justice. Established practices of witnessing have
stratified distinctions between h ­ uman and the non through an inability to
give materiality and relationality their due. Zylinska argues that “embracing
nonhuman vision as both a concept and a mode of being in the world ­will
allow h ­ umans to see beyond the humanist limitations of their current phi-
losophies and worldview, to unsee themselves in their godlike positioning
of both everywhere and nowhere, and to become reanchored and reattached
again.”89 As I conceive it, nonhuman witnessing is both a par­tic­u­lar form of
perception and something ­else besides, a communicative form ­shaped by the
materiality and affectivity of the world as medium: an ethicopo­liti­cal mode
of relation for grounding anew how meaning comes to ­matter in the making
and remaking of worlds. Nonhuman witnessing is not an ahistorical or tran-
scendental concept, but rather the naming of a set of interconnected practices
and pro­cesses of witnessing bound up with evolving epistemic frameworks
and forms of mediation.
Nonhuman witnessing is not a free-­floating concept but an injunction to
the h­ uman to become with and alongside the non in far more attentive and at-
tuned ways. Cubitt argues that fundamentally transformed practices of com-
munication offer “the possibility of changing the conduct of relations between
­human beings and nature, and between both of them and the technologies
that so profoundly and multifariously mediate between them.”90 Nonhuman
witnessing is thus a historical p­ rocess, one that has—­I would contend—­always
operated in conjunction with h ­ uman ethics, politics, and meaning-­making but
that manifests in new forms, practices, intensities, and dynamics as epistemes
and media technics change through time. Nonhuman witnessing in the con­
temporary conjuncture is thus a response to Man the Witness, but exploits,
escapes, and exists beyond the dominance of technical media. Tracing its

28 Introduction
occurrence in instances as diverse as edge computing weapons targeting and
glass-­blown art, this book shows how nonhuman witnessing addresses power
as ­process, not solely biopower or necropower, but the ontopower that brings
becoming within its ambit. As a modality that operates across multiple levels
of sense-­, truth-­and world-­making, opening witnessing to the nonhuman
takes up the task of producing new communicative aesthetics, ecologies, and
politics in the face of vio­lence and its traumatic aftermaths.

witnessing trauma, witnessing vio­lence

To testify is, in the most basic sense, to insist that something be remem-
bered by someone or something other than the witness. Memory is shared
across species, technics, and materials: it is h ­ uman and animal recall, but
also information stored in computation, ammonites fossilized in stone, scars
on gumtrees ­after summer fires. Its politics must be forged; its collectivity
brought into being. One means of making memory collective is witnessing.
Memorials to wars past bear witness, and statues of slave ­owners, Confed-
erate generals, and colonial “heroes” remind us of the vio­lence that can be
entailed in being called to witness and remember u ­ nder the normative rule
of empire.91 Memory itself is not normative, but rather attains its ethical or
moral weight through its marshalling to cultural or ­political ends. Witness-
ing, by contrast, is an ethicopo­liti­cal ­process: it is always and already on the
brink of becoming-­political, even if its politics remain latent or geared very
far from justice. Witnessing orients ­toward the f­ uture, even if it reaches back
into the past. This book, then, is not “about” memory, even if memory and
its uncertainties feature often. Instead, I am interested in the registering of
experience that precedes memory, and of the intimate relation between this
witnessing and the vio­lence and trauma to which it so often responds.
For trauma studies in the humanities, the witness to trauma—­and to
historical trauma and atrocity in particular—­lives with the violent event
written on and through the body, such that the past is in fact never past at all.
Fragments of experience cling to the pre­sent and refuse to become memory,
continuing as lived remnants of vio­lence. Testimony exposes the failure of
language, the stuttering of repre­sen­ta­tion, and the shattering of experience at
the heart of trauma.92 Testimony is thus vital and necessary, even as it cannot
ever provide a full accounting of trauma, nor be enough on its own to work
through the traumatic event and reconstitute the subject. This is part of why
trauma theory has had such influence on literary, film, and cultural theory: art

Nonhuman Witnessing 29
addresses ­those incidents of history that refuse comprehension, seeking to
overcome the collapse of meaning through aesthetic and imaginative force.
In this sense, trauma theory is unabashedly anthropocentric. It might not
celebrate a classical humanism, but it is dedicated to the h ­ uman (in)capacity
to speak in the face of that which refuses or resists speech: ­those traumatic
events that most demand voice are also exactly t­ hose that refuse repre­sen­ta­
tion.93 If the relation between testimony and traumatic event is necessarily
fractured, then how can the witness testify to historical facts? How can his-
tory even be written?94 This fragmenting of the connection between writing
or speech and the event throws testimony into crisis: witnessing becomes
precisely the urgent task of pursuing the event that w ­ ill not give itself up to
knowing, whose full scope and meaning always eludes the grasp.95 This neces-
sary failure of witnessing within trauma theory marks the failure of the h ­ uman:
witnessing signals the limit point of what the ­human can know of itself and
what it can become.96 Trauma can never appear as itself to the knowing sub-
ject, it can never be known and rendered speakable. Consequently, the ­human
itself is always bound by this failure to reckon with the traumatic. Witnessing
cannot exceed or extend beyond the h ­ uman b ­ ecause it is constitutive of an
incapacity for the ­human to be fully ­human in the face of trauma. Positioning
both trauma and testimony as operating on the line between h ­ uman and less-­
than-­human, as trauma theory does, implies that the nonhuman cannot be
accorded ­either trauma or testimony. If witnessing enacts the paradox of the
­human failure to be fully h­ uman, what room is t­ here for the animal, the plant,
the stone scorched by exploding fragments of a Hellfire missile? Yet trauma
escapes the confines of the subject. It can be climatic, atmospheric, collective,
and it can be transmitted between p ­ eople and across generations. As chap-
ters 3 and 4 argue, trauma can be both affective and ecological. Trauma con-
tinually exceeds the h ­ uman subject, which means that reading the failure of
witnessing as a falling short of the h
­ uman cannot hold. This very proposition
is an obscured anthropocentrism that predetermines what witnessing can be.
But all this discussion of testimony and trauma implies an original vio­
lence. While trauma and witnessing are often yoked together by theory,
relations between vio­lence and witnessing are often assumed, unstated, or
unresolved. In part, this is ­because vio­lence itself is a slippery concept: perva-
sive, elusive, varied, and resistant to neat formulations. But it is also b ­ ecause
witnessing and vio­lence converge and diverge, coming together in some con-
texts but not at all or only thinly in ­others. Consider the difference between
witnessing police killings and witnessing a volcanic eruption. Both might
involve the destruction of life, but only one constitutes vio­lence as such.

30 Introduction
Hannah Arendt makes this distinction clear. “Vio­lence,” she writes, “is dis-
tinguished by its instrumental character,” whereas force describes “the energy
released by physical or social movements.”97 If vio­lence is instrumental, it is
also relational. It might well be that vio­lence is intrinsic to being a body. “The
body implies mortality, vulnerability, agency: the skin and the flesh expose
us to the gaze of o ­ thers,” observes Judith Butler, “but also to touch, and to
vio­lence, and bodies put us at risk of becoming the agency and instrument
of all t­ hese as well.”98
But vio­lence can be structural, as well as direct and immediate, “exerted
systematically—­that is, indirectly—by every­one who belongs to a certain
social order,” as Paul Farmer observes.99 Structural vio­lence resists neat as-
criptions of blame or responsibility. Its effects are diffuse yet deeply harm-
ful, enabling oppression and working to maintain existing hierarchies of
wealth and power.100 Capitalism and colonialism are forms of structural
vio­lence, even if they can also manifest in more kinetic, martial, and im-
mediate forms. This is why Patrick Wolfe describes settler colonialism as a
structure, not an event.101 But other forms of distributed vio­lence feature
in this book: symbolic, discursive, infrastructural, environmental, and
algorithmic vio­lence, for example. Lacking an obvious originating agent,
such vio­lence takes place through institutions, linguistic exclusions, tech-
nocratic programs, extractive industries, and other such assemblages, often
harnessed to state and corporate power but at times filtered through more
ambiguous actors.102
Vio­lence is not only distributed, but also differentially experienced. As
Saidiya Hartman, Hortense Spillers, and other scholars of slavery and Black
life teach us, vio­lence strips away the body and exposes the flesh to injury,
often in diffuse and difficult-­to-­detect ways that permeate the quotidian.103
Racial vio­lence exemplifies this dynamic b ­ ecause it coalesces the capricious-
ness of law, the exclusionary force of Man, and the harnessing of relation to
produce subjects not governed by the law. Writing on the killing of p ­ eople
of color in Brazil’s favelas, Denise Ferreira da Silva argues that “raciality im-
mediately justifies the state’s decision to kill” b ­ ecause such “bodies and the
territories they inhabit always-­already signify vio­lence.”104 Vio­lence exposes
the vulnerability of the body, but it distributes that vulnerability in radically
unequal ways. To say, then, that the body is defined by its vulnerability to
vio­lence makes a necessarily ­political claim about who gets to possess a body
to encase their flesh. This is a question rooted in the Enlightenment concep-
tion of the subject, the figure of Man that Wynter ties to ­European colonial
expansion. Binding witnessing to the ­human means that who witnesses is

Nonhuman Witnessing 31
always contested ground—­and witnessing itself can be complicit in the le-
gitimation of vio­lence. ­After all, can the figure denied humanity bear witness
if witnessing belongs to the h ­ uman? Preceding the body, flesh marked by
vio­lence offers a way outside of Man, a fugitive witnessing enabled through
the generativity of flesh that refuses to give up its vitality and seeks solidarity,
­resistance, and joy.
Vio­lence, in other words, is a malleable phenomenon. In war, it can be
mechanized and automated, but also intensely intimate. It can unfold
slowly, as in the degradation of bodies exposed to radiation or the col-
lapse of environments polluted by toxic. “Vio­lence unfolds on dif­fer­ent
scales, over dif­fer­ent durations, and at dif­fer­ent speeds,” writes Weizman. “It
manifests itself in the instantaneous, eruptive force of the incident, evolves
in patterns and repetitions across built-up areas, and then manifests itself
in the slower, incremental degradation of large territories along extended
timescales.”105 Nor are ­those forms, modalities, intensities, and speeds sepa-
rate from one another. Vio­lence flows between states. Buzzing in the sky
above, the drone generates fear and abiding anxiety, a kind of diffuse and
atmospheric vio­lence, even as its surveillance systems engage in the vio­lence
of datafication, transforming the textures of life into metadata. And then,
when a target is acquired and a missile launched, vio­lence becomes horrify-
ingly kinetic. P­ eople living u­ nder drones in A ­ fghanistan, Yemen, Gaza, or
Ukraine witness this vio­lence, as do members of the military apparatus from
[103.158.136.64] Project MUSE (2024-02-06 19:57 GMT) University of Edinburgh

operators to intelligence analysts to authorizing officers. But when vio­lence

is so inseparable from environments and technoscientific systems, ­there is
much that testimonies of the nonhuman can offer. For all the moral force
that resides in h ­ uman rights testimony and the humanitarian witness, the
entanglement of the nonhuman in vio­lence suggests the need for nonhuman
witnessing as fundamental to healing and repair for h ­ uman, nonhuman, and
the worlds we share.
Oliver’s generative attempt at disentangling trauma, vio­lence, and wit-
nessing offers a way through this knot. Writing against the notion that social
strug­gles are strug­gles for recognition of difference, Oliver develops a theory
of the relational formation of subjectivity that turns on the ethicopo­liti­cal
imperative of witnessing. Drawing on a Levinasian ethic that privileges the
other over the self, Oliver argues that “the speaking subject is a subject by
virtue of address-­ability and response-­ability.”106 Both address-­ability and
response-­ability are at the heart of witnessing, and so witnessing is “the basis
for all subjectivity; and oppression and subordination work to destroy the
possibility of witnessing and thereby undermine subjectivity.”107 The inversion

32 Introduction
­ ere is crucial: witnessing is not simply a response to vio­lence, but what
h
vio­lence destroys. “While trauma undermines subjectivity and witnessing
restores it,” she writes, “the ­process of witnessing is not reduced to the testi-
mony to trauma.”108 Trauma cannot be the foundation of subjectivity b ­ ecause
such a move could only engender an impoverished ­political life. Disaggre-
gated from trauma, witnessing forges bonds that exceed any given situation
or singular act of witnessing.
Witnessing is always an open-­ended, recursive, and necessarily active
­process of becoming. But the impor­tant move that Oliver makes is to situ-
ate witnessing within a relational milieu, arguing that the self develops its
capacity as an internal witness through being witnessed by the other and
that is how subjectivity emerges from and with social relations. Working
within a psychoanalytic framework, Oliver argues that witnessing is essential
to working-­through hostilities that stem from fear and anxiety over differ-
ence. This is a “profoundly ethical operation insofar as it forces us not only
to acknowledge our relations and obligations to ­others” but to transform
them.109 Working-­through connects witnessing to sociality and makes trans-
formations—of love, of justice, of re­spect—­pos­si­ble. Unsurprisingly, Oliver’s
witnessing is unquestionably h ­ uman: a p­ rocess that involves “language and
gestures” and an act of “love” in the face of the other and against the de-
humanizing power of oppression and vio­lence. Witnessing is intrinsically
­human such that ­human subjectivity itself is the “result of a continual ­process
of witnessing.”110 Objects have no capacity to witness precisely ­because the
object cannot speak or gesture.
Despite this avowed humanism, Oliver’s account helps elucidate some of
the interventions this book makes in thinking witnessing with the nonhu-
man. First, the rejection of a symbiotic relationship to trauma opens wit-
nessing to world-­making in ways that invite richer and more generative
potential while not at all foreclosing the necessity of witnessing in response
to trauma and vio­lence. Second, the insistence on the relationality of wit-
nessing as enacted through address and response provides a way into what
witnessing might be if address and response involve nonhuman animals, ma-
chines, entities, and environments, and so on, as long as we understand both
address and response outside their familiar anthropocentric frames. Third,
the conception of relationality as fundamentally biosocial, affective, and en-
ergetic already contains within it a permeability that is almost ecological in
its insistence on complexity and p ­ rocess. Fourth, the notion that witnessing
forges relations that make working-­through hostilities to difference pos­si­ble
offers a way of understanding the dynamism of witnessing and why it makes

Nonhuman Witnessing 33
transformation pos­si­ble. Taken together, ­these four implications offer points
of departure from the ­human witness and into the unruly domain of nonhu-
man witnessing.

In the painting Theatre of War: Photons Do Not Care, (figure I.1), Kathryn
Brimblecombe-­Fox depicts the machinic attempt to make planetary environ-
ments subject to martial enclosure. A cluster of drones, networked by fine red
lines, looms over a pale dot in a field of rich blues and reds reminiscent of
scientific visualizations of cosmic evolution. Viewing the painting, we reside
in the cosmic distance, thrown far from any conceivable ­human perception of
the Earth or its technologies of war. And yet the painting calls for us to attend

figure I.1. Theatre of War: Photons Do Not Care, oil on linen 92 × 112 cm, Kathryn
Brimblecombe-­Fox, 2021. Courtesy of the artist.

34 Introduction
to the planetary nature of military technologies, to their growing tendency
to render space-­time itself as a site of martial contest. Photons do not care:
­these massless particles are the raw stuff of the electromagnetic spectrum,
transcending national bound­aries, the h ­ uman, and the planet itself. And yet
they are also, increasingly, the site of military contestation and intervention,
as autonomous and cyber warfare infuses all other forms of martial conflict.
Military media, networked systems, and algorithmic assemblages all seek
mastery, and in ­doing so tug us into an age in which the world as target has
given way to the planet as an operative medium for targeting any point on
or above its surface.111 The hand of the artist is evident in the occasional un-
blended brush stroke of oil on linen, and in the uneven stippled dots arranged
into the pixelated drones. Th ­ ese pixelated silhouettes of looming drones blur
computational mediation with organic repre­sen­ta­tion, h ­ uman hand, and
galactic scale. ­There is no escaping the ­human, the painting insists, no release
into an existence without responsibility for the crises wrought in the name of
economic growth, colonial expansion, state power, and military supremacy.
The question is what ­will happen, down on that pale blue dot, ­toward survival
and a new flourishing of life?
If crisis is the ­political and ecological condition within which much of
the planet lives, the unraveling of the fantasy of a unified, cohesive, and
knowable world offers some potential for more just and equitable f­ utures.
The enmeshed desire of states and other actors to both produce and control
crisis—­crisis as a modality of governance that allows for the abrogation of
demo­cratic and other responsibilities—is not solely about discourses, institu-
tions, or even technologies that target individuals and populations, ­whether
as biopo­liti­cal life-­in-­the-­making or necropo­liti­cal death-­in-­waiting. Onto-
power heightens the stakes of con­temporary technopo­liti­cal power, enabling
states and other actors to target the stirring of life within the bare activity
of existence. Techniques of ontopower seek to direct being as it becomes, to
harness emergence itself to the ends of the already dominant forces of pro-
duction and control. Such are the promises of the algorithmic technologies
of war, governance, culture, and ecol­ogy that this book explores, but so too
is ­there the potential in resistant harnessing of technics and aesthetics, algo-
rithmic and other­wise, to produce new modes of surviving with and living
beyond the World of Man.
Addressing h ­ uman responsibility for the existential crises within which
we find ourselves—­and reckoning with the radically unequal distribution
of both responsibility and the effects of crisis—­requires us to hold onto the
­human. But this holding onto the h ­ uman must also undo the blind privilege,

Nonhuman Witnessing 35
the narrowness of vision, and the closed imagination that undergird an An-
thropos that is bound to Man. Oliver writes that “being together is the chaotic
adventure of subjectivity.”112 This book calls for witnessing as the foundation
of a renewed becoming-­together—­becoming-­environmental, becoming-­
machinic, becoming-­imperceptible—­that coheres not on h ­ uman subjectivity
but on the chaotic dance of life and nonlife.

36 Introduction
Chapter
One

witnessing
vio­lence
two mq-9 reapers confront each other nose to nose, simulated aerial ve-
hicles floating above simulated mountainous country. Light bends across
the mirrored surface of one; the other is gray and black, a digital replica
of its physical counterpart. Interspersed by spinning reflective planes and
suspended in inscrutable contemplation, the two machines seem possessed
of their own needs and desires. What takes place in this communion of mili-
tarized drones? While the drone skinned in military tones and textures is
disconcerting if familiar, the mirrored drone is both alluringly beautiful
and horrifyingly alien, an other-­than-­human object across which the gaze
slides and fails to stick. Its mirroring offers no clear reflections, but rather
refracts its surroundings into distorted fragments—­a nonhuman resurfacing,
the world rendered into the materiality of the drone as it seeks to become
imperceptible. This moment in Australian artist Baden Pailthorpe’s mq-9
Reaper I–­III (2014–16) captures something of what makes military drones
fascinating, disturbing, and urgently in need of critical attention. At once
threatening and seductive, the Reaper drone promises an omniscient and
yet nonhuman capacity to perceive, know, and kill, one that sanitizes war by
making it datalogical, computational, and spatially and affectively remote.
For the militarized drone of the last two ­decades—­exemplified by (but far
from ­limited to) the Predators, Reapers, and Global Hawks operated by the
United States, or the Turkish Bayraktar tb2 used by Ukraine—­this capacity
has depended on its near invisibility, its ability to operate untouched from the
atmosphere. As war becomes increasingly autonomous and more centered on
­great power conflict, the forms and applications of drones are becoming far
more varied, ubiquitous, and dependent upon artificial intelligence.
Exhibited at Centre Pompidou, Art Basel Hong Kong, and numerous
festivals and galleries, mq-9 Reaper I–­III pre­sents drone warfare as vio­lence
enacted through the computational simulation of real­ity (figure 1.1). Built in
the modeling program Autodesk 3ds Max, Pailthorpe’s proj­ect reimagines
key locations within the drone apparatus into the air above an environment
that references the mountainous terrain of ­Afghanistan, over which drone
warfare took its con­temporary form. Shipping containers rotate slowly in
the clouds, walls cantilevering open on hydraulics to reveal ground control
station cockpits loaded with the screens, controllers, and interfaces needed to
crew the Reaper and its siblings. Or they open to expose spare living rooms
in which uniformed men perch on beige couches or do jumping jacks, trans-
planting the suburban life that bookends on-­base shifts operating drones
from the domestic United States to the atmospheric zone of war. Graphics
are realistic but heightened, surreal simulacra of the computational space
of war and an aesthetic familiar to both video games and the promotional
videos produced by arms manufacturers. Their sterility mimics the rhe­toric
of precision and hygiene that accumulates around remote warfare and infuses
the technocratic and corporate discourses that elide the vio­lence inflicted by
lethal strikes.
More than this, the computational materiality of mq-9 Reaper is a stark re-
minder of the layers of simulation, data, modeling, and algorithms connected
by distinct logics and pro­cesses that constitute the martial con­temporary.
Estranging relations between ele­ments within the drone apparatus while
insisting on the distortions and reflections produced by its operations, Pailt-
horpe lays bare the circulatory, diagrammatic flows of the system by shifting
the locus of agency away from the h ­ uman and to networked relations. When
soldiers appear on screen to shadowbox and sit at their control stations, they
have also entered the space of the drone and become its subject. Yet in taking
up the toolkit of modeling, computation, and simulation, Pailthorpe know-
ingly enters the epistemic regime of con­temporary war and so is bound to
its informational logics and repre­sen­ta­tional modalities even as they come
­under scrutiny. How, then, to witness this increasingly autonomous form of

38 Chapter One
figure 1.1. Still from mq-9 Reaper (III), Baden Pailthorpe, 2016. Courtesy of the
artist.

war? How to grasp the vio­lence its witnessing might do? While Pailthorpe’s
aesthetic intervention makes for an instructive entry point into the entangle-
ment of aesthetics, war, and computation, this chapter is not about drone art
per se.1 Rather, it pursues ­these questions of witnessing vio­lence by tracing
the violent mediation that is essential to perception, knowledge-­making, and
communication in con­temporary war.

drone war’s violent mediation

Violent mediation names ­those material pro­cesses that are constitutively

harmful, w ­ hether b­ ecause they cut, target, exclude, define, categorize, or
classify in ways that are injurious to h ­ uman or nonhuman entities and en-
vironments. Weapons targeting systems are one such violent mediation in
which the flux of light, molecules, and energy captured by computer vision
systems are directed through interfaces that enable the se­lection of entities
for lethal assault. But so too are mugshots, colonial land registers, and pesti-
cides. Mediation itself is ambivalent, as Sean Cubitt notes, its flux preceding
“all separations, all distinctions, all thingliness, objects, and objectivity.”2 In
this sense, “mediation as the very fabric of change, of mutation, is a builder of

Witnessing Vio­lence 39
differences, but as ­bearer of communication, it also establishes ­organizational
forms with varying degrees of longevity.”3 While mediation can be transfor-
mative and generative, enabling deep communication and the flourishing
of rich ecologies, it is not bound by moral standards nor intrinsically ethi-
cal. Mediation is thus not a normative ­process. With this concept of violent
mediation, I want to distinguish between mediation in general and t­ hose
instances in which it animates ­human desires to control, extract, dominate,
oppress, and kill. Violent mediation is often most evident through technical
systems that subjugate life and nonlife to their ends, but it is also at work in
datafication and computation, and in a host of biogeophysical interactions
instigated by ­humans to bring ecologies to heel or direct them to ­human
ends. In this chapter, my focus is on the violent mediations of drone warfare,
enacted through its sociotechnical apparatus. Violent mediation is not ancil-
lary to drone warfare, but constitutive of it.
In this, drone warfare is not an outlier within war more generally but
rather symptomatic of its media saturation. Martial operations are intensely
mediated, bound together through recursive informational flows structured
and ­organized by media technics. “Military knowledge,” as Packer and Reeves
put it, is primarily “a media prob­lem, as warfare is o ­ rganized, studied, pre-
pared for, and conducted according to communicative capacities.”4 Military
strategy, logistics, and operations are all determined by media technological
capacity, but also shape t­ hose technologies in turn. The necessity of com-
munication across distance produces semaphore, the t­ elegram, satellites,
and the internet, and t­ hese then enable naval formations, the coordination
of mass armies, the deployment of missile batteries, and the networking of
the battlefield via tactical drones, wearables, and mapping systems. This co-­
constitution of war and media means that h ­ uman soldiers, p­ ilots, analysts,
and even commanders are increasingly ancillary to the workings of the
systems themselves. If this was already true in the logistics or command-­and-­
control infrastructures of e­ arlier wars, the intensification and proliferation
of automation marks an acceleration of the removal of h ­ uman agency. No
longer the essential component in waging war, the h ­ uman is increasingly
seen as ­either its most fallible ele­ment or its datalogical target. The ballistics
revolution reor­ga­nized battlefield perception around wider geographies and
enabled the infliction of vio­lence at considerable distance, while the nuclear
revolution introduced a planetary perception coupled with the potential for
vio­lence at a planetary scale. But the emergent ai revolution is reconfiguring
perception to be everywhere and nowhere, with the capacity for vio­lence so
tightly bound to perception that it too can take place anywhere at any time.

40 Chapter One
Warfare transforms not only in connection with technological, strategic, or
even ­political change, but also in concert with epistemic shifts in the foun-
dational frameworks, assumptions, and ­metaphors of scientific knowledge.5
From its inception, artillery targeting entailed mediation: the se­lection
of targets, ­measuring of distances, the translation to maps, the adjustment
of machinery, the firing of the gun. But with the emergence of autonomous
systems of war—­exemplified by the adoption, development, augmentation,
and transformation of remotely pi­loted systems such as drones—­mediation
takes on a new complexity founded on the ­imagined and presumed exclu-
sion of the ­human from its workings. Wide area motion imagery systems
track areas as large as small cities at high resolution, identifying and follow-
ing targets of potential interest that would be difficult if not impossible for
­human analysts to comprehensively account for. As such systems develop
in capacity and autonomy, automated pro­cesses of mediation w ­ ill locate,
select, track, and even execute threats that only exist within the framework
of the system. Military media are thus “constantly producing new enemies,
and new methods of ­enemy identification stimulate the development of new
weapons technologies designed to kill t­ hose newly identified enemies.”6 This
interconnection between media and what Packer and Reeves call “­enemy
epistemology” and “­enemy production” is not only a question of stabilized
media technologies intersecting with military strategic imperatives. It also
occurs through material pro­cesses of mediation, bounded by instrumental
technologies but let loose on the complex terrain of life.
As I theorize it, violent mediation is embedded in a material-­ecological
understanding of war and the role of technologies of perception within it. In
this, it shares much with what Antoine Bousquet terms the “martial gaze,”
which aligns “perception and destruction” through “sensing, imaging and
mapping” that encompasses not just the visual but “the entire range of senso-
rial capabilities relevant to the conduct of war.”7 As perception and vio­lence
are increasingly twinned, mediation functions within ­those apparatuses to
produce vio­lence. Violent mediation is thus intrinsic to the martial gaze. We
might think of violent mediation as the connective tissue of such systems,
constituting sensing at the material level of technical operation but also
stitching sensing into the larger apparatus: the thermal camera of the drone
sensing its environment entails vio­lence within its mediating pro­cesses, but
also in the translation from sensing (thermographic camera) to imaging (de-
coding for optical display) to targeting (fixing of the reticule on an agglom-
eration of pixels). Pro­cesses of mediation occur within each stage, but also
across them and throughout the kill chain. Attending to violent mediation

Witnessing Vio­lence 41
thus means focusing on the movement, use, and structuring of information
within the military apparatus, as well as within the ele­ments that compose
it. As with the martial gaze, much of this mediation is not visual—or only
presented visually for the benefit of h ­ uman actors within the system. Much of
what is violent in such mediation is bound up with the technical pro­cesses
of datafication, abstraction, analy­sis, and instrumentalization that increas-
ingly animate military technologies of perception.
This chapter asks how witnessing might take place through the violent
mediations of the martial gaze, and how t­ hose mediations—­and the corpo-
real, ecological, and affective vio­lence they engender—­might be witnessed.
It locates remote and increasingly autonomous warfare as both a driver and
beneficiary of algorithmic enclosure, while recognizing that it si­mul­ta­neously
responds to and produces ecological crises.8 Nonhuman witnessing provides
an analytic framework for conceiving and excavating the witnessing that
takes place in, by, through and, crucially, of the drone assemblage. War has al-
ways been a form of life, as Grove maintains, but its emergent con­temporary
forms possess a ubiquity, complexity, variability, autonomy, and technicity
unpre­ce­dented in ­human experience. Reckoning with this becoming-­war ­will
require a refiguring of the h­ uman relation to it, but also a transformative shift
in what counts as ethical and p ­ olitical claims to knowledge. This chapter thus
lays conceptual foundations for the examinations of algorithms, ecologies,
and absences that follow by showing how violent mediation is constitutively
imbricated with war.
By attending to the nonhuman of witnessing, I am not dismissing or mar-
ginalizing the Afghans, Yemenis, Somalis, Palestinians, Pakistanis, Syrians,
Iraqis and o ­ thers who have given and w ­ ill continue to give testimonies to
reporters and h ­ uman rights o­ rganizations.9 As Madiha Tahir forcefully points
out, “­every ­thing is speaking and talking and witnessing and testifying ­these
days, it seems, except the p ­ eople whose f­ amily members and neighbors
have been blown to bits in this war.”10 Hearing t­ hose voices louder and
in more forums is unquestionably a vital task. Factual in orientation and
presented as narrative, many of t­ hese testimonies are s­ haped by the expecta-
tions of h
­ uman rights convent and the norms of tribunals and courts.11 Their
very familiarity, their echoing of testimonies of torture or rape or migration,
speaks to the “becoming witness” of international humanitarian politics
in the latter half of the twentieth ­century.12 Such testimonies intentionally
reinforce the humanist, rights-­bearing subject ­because their very efficacy
and legitimacy depends on recognition by the institutions and conventions
of international humanitarian law, which are themselves interwoven with

42 Chapter One
neoliberal attempts to develop a moral framework for cap­i­tal­ist relations in
the wake of World War II.13 Yet in d ­ oing so they seek to make recognizable
encounters with nonhuman systems of vio­lence—­networked, autonomous,
highly technical, and massively distributed in space—­that resist the forms of
knowing and speaking available to the eyewitness. ­There is a tension, then,
between the necessity and possibility of making drone vio­lence legible within
the conventions of human-­centered forums, ­whether international humani-
tarian law or rights discourses more generally. Within such a framework,
drones and their data can only be made evidence, rather than recognized as
witnessing in themselves. That is, h ­ uman witnessing takes p ­ recedence and
priority, relegating the nonhuman to the status of evidence that must be in-
terpreted. While Pugliese provides a power­ful case for a counterforensics that
reckons with the more-­than-­human and Schuppli shows how material wit-
nesses can obtain standing within public and ­legal fora, this chapter adopts
a strategic agnosticism t­ oward the agencies that animate the drone apparatus
and to the potential for any instance of witnessing taking f­ uture shape as
testimony. It refuses to deny potential standing as witness to the system (the
entire military drone network, for example) nor any given ele­ments of such
systems (automated image analy­sis software, for example), even if they ­will
be hostile witnesses. And it understands nonhuman witnessing as preceding
the existence of fora for testimony, and so sees witnessing as i­ndependent
from such fora. This chapter thus attends to the constitutive entanglement
of ­human and nonhuman witnessing as a relational p ­ rocess of mediation
through which vio­lence is both registered and enacted on ­people, places, and
ecologies, no ­matter w ­ hether testimony is ever called for.
In the remainder of this chapter, I examine nonhuman witnessing within
the widening frame of increasingly autonomous martial systems. First, I
consider the multiplying aftermaths of drone vio­lence, attending to the in-
terplay of the survivor testimony, war’s material and cultural traces, and the
way drone sensors and computational systems perform their own nonhu-
man witnessing. As a counterpoint to this bleak vision, I then turn to look
at drone and remote sensor witnessing of Aleppo, Syria, in the aftermath
of war. Moving from the drone war of recent ­decades to more autonomous
­futures, I then examine the violent mediations of augmented sensor systems
in the case of the Agile Condor targeting system, which I read as an instance
of automated media that displaces and disperses witnessing across military
architectures and into the preemptive technics of edge-­computing targeting
systems. Fi­nally, the chapter closes with an extended discussion of witness-
ing, autonomy, and the martial ­future of violent mediation.

Witnessing Vio­lence 43
martial drone empiricism

Drone is the colloquial term for an unmanned or remotely pi­loted aerial

vehicle or—­more properly—an unmanned or remotely pi­loted aerial sys-
tem. At a minimum, the vehicle requires a controller, network, and signal to
operate. Hobby drones typically form a wifi network with a smartphone as
controller. Small military drones such as the AeroVironment rq-11 Raven,
a fixed-­wing drone designed for tactical battlefield awareness, are launched
by hand, and networked to a hardened laptop. The Predator and Reaper
commonly associated with drone warfare are more complex, employing
a “remote split” system in which the drone is launched from one location
before control is handed off via satellite link to an operations crew, typically
located in the continental United States. Data feeds from ­those systems can
flow across an array of military institutions and actors, with communications
inputs streamed back into the control station via voice and irc-­style text
chat. Swarming drones are more complicated still, communicating with one
another in the ­service of a predefined mission and thus even more dependent
on software and sensors.
Drone systems are complex media architectures subject to continual
transformation, which means they are best understood as hybrid collections
of h
­ uman and nonhuman agents and the relations that bind them.14 As An-
thony McCosker and Rowan Wilken observe, “Drones have emerged as a set
of technologies that throw orbital power off its axis through their unfixed,
unruly trajectories, their accessibility to ordinary users and their multidi-
rectional motility.”15 ­Whatever their form, as Lisa Parks and Caren Kaplan
write, drones “are loaded with certain assumptions and ideologies.”16 Yet
while it is tempting to think of drones as radical departures—as exceptional
technologies—­undue focus on their newness obscures their debt to histories
of airpower, racializing surveillance, and colonial-­imperial practices of clas-
sification and control.17
Figured within the long history of airpower and its relations to visual
culture, drones d ­ on’t so much mark a radical break in the evolution of the
martial gaze as coalesce a set of tendencies residing within the technics, imagi-
naries, and conduct of modern war.18 This coalescence is particularly evident
in their operational combination with the “kill box,” the US military term for
a temporally ­limited, geo­graph­i­cally specific, and volumetrically defined zone
in which deadly force is preauthorized.19 Defined by a grid reference system
and managed computationally through militarized communications systems,
the kill box neatly encapsulates the violent mediation constitutive of con­

44 Chapter One
temporary war in general and of drone war in par­tic­u­lar. The kill box itself
is a mediation: an operative transfiguration of world into media. In taking
up life and refiguring its relation to death, this mediation is constitutively
violent even before it kills, reworking the ontoepistemological status of t­ hose
within its ambit from life to not-­yet-­death. ­Whether in concert with the kill
box or operating in a less preauthorized context, the kill chain of the drone
is distributed, dispersed, and mobile, producing and responding to emergent
threats actualized within and through the network.
In this chapter, I approach the prob­lem of witnessing (drone) vio­lence by
understanding it in relation and response to the becoming of war, rather than
beginning with an ­imagined fixity or boundedness to war. Against the idea
that the nature of war is given or known in advance, Antoine Bousquet, Jairus
Grove, and Nisha Shah propose embracing “war’s incessant becoming” such
that “its creativity, mutability and polyvalence” are as central to analy­sis as its
destruction.20 Their “martial empiricism” references philosophies of radical
empiricism—­particularly Whitehead, James, and Deleuze—­that resist any
preferential focus on ­either ontology or epistemology in ­favor of an open-­
ended embrace of experience in all its generative mutability. Martial empiri-
cism orients critique ­toward the pro­cesses, relations, affects, sensations, and
technicities through which war autopoetically emerges. Such an approach
necessarily involves an openness to the incapacity to provide ultimate or de-
finitive answers and demands instead that martial vio­lence be apprehended
“as a p­ rocess of becoming that is suspended between potentiality and actual-
ity,” in which the task of critique is “scrutinizing the enfolding of intensities,
relations and attributes that give rise to war’s givenness.”21 In the context of
increasingly autonomous warfare, one starting point for a martial empiri-
cism might be the perceptual relations that cohere around the figure of the
drone, itself understood as an unstable and hybrid assemblage through which
knowledge is produced and operationalized to violent ends.
My concern h ­ ere, however, is less the emergent dynamics of autonomous
warfare as such but rather how witnessing occurs within this condition of
martial vio­lence, and how nonhuman entities and pro­cesses engage and enfold
­human experiencing and witnessing. My pursuit of nonhuman witnessing
within this becoming-­war takes place through attention to violent media-
tion as a transversal ­process that both occurs within and connects distinct
formations of martial vio­lence, as well as the bodies, technologies, and situ-
ations that compose them. Attending to violent mediations as pro­cesses of
knowledge-­making and communicating opens the terrain on which witnessing
can and must take place. As I theorize it ­here, nonhuman witnessing provides

Witnessing Vio­lence 45
a mode of inquiry into the tensions between ­actual and virtual in the flux of
becoming as it is interrupted, redirected, and mutated by martial vio­lence.
Let us begin, then, with the violent mediations that animate the drone war
assemblage by attending first to the shift from optical to datalogical media-
tions. In their first operational incarnation above the skies of Kosovo in the
1990s and then A ­ fghanistan a­ fter 9/11, Predator drones w
­ ere primarily optical
technologies. With full motion video (fmv) and (usually) thermographic
sensors, ­these drones “produce a special kind of intimacy that consistently
privileges the view of the hunter-­killer,” as Derek Gregory puts it in an early and
influential critique of drone vio­lence.22 One operator describes the view from
above as “looking through a soda straw” that cuts context and complexity
and tends to lock focus on ­whatever stays within its narrow targeting frame.23
Limitations of bandwidth and multiple stages of encoding and decoding
meant that video imagery was often not received by operators at anywhere
close to the high definition in which it was recorded, while the atmospheric
location of the sensors meant that p ­ eople w
­ ere principally seen from directly
above or at a very acute a­ ngle, dehumanized pixels rather than recognizable
persons. This violent mediation cut, reduced, and blurred complexity in ways
that encouraged the infliction of force: rather than generating uncertainty that
might discourage lethal action, the mediation of events in the world through
the technical apparatus produced degraded information that was read as a
threat within the system. While the perceptual capacity of drone sensors
has advanced in the last ­decade, the under­lying dynamics of using degraded
information to produce threats remains very much in place in con­temporary
Reapers, Global Hawks, and similar lethal surveillance platforms.
To make sense of the drone as paradigmatic of a par­tic­u­lar strand of con­
temporary war, I want to tease out the relational pro­cesses that underpin
drone vio­lence and in d ­ oing so shift the locus of inquiry from image and
repre­sen­ta­tion to mediation. Drone vision is digital vision, enabled through
sensors that transform light into binary data rather than an analog imprint.
Such vision operates through change and transmission of code, mathematical
arrangements that can be rendered into pixels for display to ­human operators.
Drone vision is thus operative and actionable, rather than merely repre­sen­ta­
tional.24 That is, we can think of the drone assemblage as not only perceiving
but also producing slices of the world upon which operations can be per-
formed. Drones are automated media, oriented ­toward the ­future and governed
by a logic of preemption that seeks to define and control threat. “Pre-­emption
operates in the register of the urgency of the imminent threat,” writes Mark
Andrejevic.25 Privileging visual repre­sen­ta­tions risks instantiating problem-

46 Chapter One
atic imaginings of the temporal and spatial dynamics of drone warfare at
the expense of properly grasping its networked, mediated, pro­cessual, and
computational logics as a sociotechnical assemblage. Mediation is the per-
formative transformation of a perceptual encounter, one that occurs in time
and exceeds its content. It is a vital ­process, as well as a technical one: indeed,
its technicity is itself a form of life.
Drone mediations are enmeshed with terrestrial surfaces and substrates,
aerial atmospheres, built environments, multiple spectrums, and corporeal ac-
tivities. Parks calls this vertical mediation: “a ­process that far exceeds the screen
and involves the capacity to register the dynamism of occurrences within,
upon or in relation to myriad materials, objects, sites, surfaces or bodies on
earth.”26 As mediating technologies, “drones do not simply float above—­they
rewrite and re-­form life on earth in a most material way,” extending to “where
­people move and how they communicate, which buildings stand and which
are destroyed, who s­ hall live and who s­ hall die.”27 In the context of war, the
mediations of the drone apparatus are not solely vertical but also violent, and
that vio­lence is bound up with verticality. In perceiving and capturing slices
of existence through its perceptual technics, the drone assemblage is at once
reductive and productive. Reductive, in that it frames and subordinates life
within the narrow aperture, a­ ngle, and classificatory mechanisms of milita-
rized knowing. Productive, in that it transforms that life into actionable data
crowded with virtual ­futures of per­sis­tent surveillance, active control, and
even arbitrary death.
Both the soda straw and bandwidth prob­lems spurred technological de-
velopments that marked an impor­tant shift in the sensory apparatus of war
and an intensification of its violent and vertical mediations. To c­ ounter the
narrow field of view, darpa facilitated a series of wide area motion imagery
(wami) initiatives to equip drones with sensors capable of recording and
analyzing hundreds of city blocks within a single frame.28 In its early forms,
wami promised to capture every­thing, but in ­doing so produced an aston-
ishing amount of data. Automated image analy­sis tools sought to exploit the
totality of the feed, a feat what would require hundreds, if not thousands,
of h
­ uman analysts working in real-­time. But bandwidth issues also meant
that wami was difficult to make operational via the ad-­hoc satellite, optical
fiber, and wireless relays that compose military network infrastructures.
wami thus produced spatial and temporal expansions in potential capability
and in ­labor, network, and computational demands. Take the Autonomous
Real-­Time Ground Ubiquitous Surveillance Imaging System, or argus-­i s,
which combined 368 overlapping high-­definition sensors into the equivalent

Witnessing Vio­lence 47
of a 1.8-­billion-­pixel camera to provide a high-­resolution, full-­motion video
of up to ten square miles at a ground resolution of six inches per pixel from
an altitude of twenty thousand feet (figure 1.2). As it was hyped in the 2013
pbs documentary Rise of the Drones, analysts would be able to create video
win­dows, track vehicles, generate 3d models, and access location-­specific
archives to compare prior activities and track environmental change.
The volume of data produced by the system was astonishing: up to one
billion gigabytes of data in twenty-­four hours r­ unning at full capacity. Such
potential perception far outstripped ­human visual capacities, promising to
transform the world and its inhabitants into actionable data that can be called
up on demand and rolled back and forward through time. But that techno-
logical capacity was never realized in practice due to the massive bandwidth
and computational power required to make the system effective. For wami
to provide its promised ubiquitous surveillance, the prob­lem of getting data
to ­humans in swiftly actionable form needed to be resolved. The obvious
answer was to reduce the reliance on h ­ umans: new systems are thus built

figure 1.2. Interfacial image from argus-­i s ­presentation, 2013

48 Chapter One
around on-­board packages that automatically analyze sensor data for items
of interest and then push a selected subset of data through to h ­ uman analysts
and operators. ­These edge computing systems, such as the Agile Condor
pod that I discuss ­later in this chapter, mark an intensified operative role for
computation, one in which autonomous software systems not only rec­ord
and analyze but also pre­sent data as actionable, where action can lead to kill-
ing. Mediation ­here takes on an overtly violent tendency, not simply through
what it excludes or removes but through the lives that it pre­sents as (poten-
tially) requiring the application of lethal vio­lence. As wami, edge computing,
machine vision, photogrammetry, and autonomous targeting and navigating
systems in general show, violent mediation is increasingly complex, distrib-
uted, and thick.29 The identification, se­lection, targeting and execution of
­people depends upon a growing number of systems and technics involving
increasingly interoperable components, while at the same time becoming
opaquer in its workings. Making remote and increasingly autonomous war
sensible—­that is, making it graspable and addressable within the terrain of
politics rather than its irruption into martial conflict—­requires finding ways
to witness the workings of t­ hese violent mediations. Yet the perceptual op-
erations of violent mediation can themselves produce witnessing: registering
and responding to vio­lence, including their own.

tenuous aftermaths

Drone warfare seems not to want to produce lasting aftermaths. Drone wars
persist, carried on through the open-­ended generation of threat, the low
cost of involvement for aerial powers, and the ease with which they can be
returned to the air above places and populations. This distended temporality
is punctuated both by intense periods and sharp instances of vio­lence and
textured by the ever-­present potential of death from above. ­Wartime, writes
Beryl Pong, “constitutes its own violent, recalcitrant temporality.”30 Living
with drone war means living in enduring aftermaths, troughs of grief and
ruin that follow from drone strikes and shadow operations yet can never
mark an end to w­ artime. Drone war’s aftermaths are rarely spectacular, trans-
lated into narrow idioms that commemorate and reinstantiate a lost, yet
mythical, past. Instead, the aftermaths of drone war are intimate, contested,
and unruly; ­etched in stones, buildings, gardens, and bodies; seared into the
fabric of communities and cultures.

Witnessing Vio­lence 49
The photojournalism of Noor Behram captures t­ hese entangled effects
of drone vio­lence throughout Waziristan on the Pakistani border with
­Afghanistan. Haunted ­faces of survivors, shattered bodies of victims, broken
homes, and fragments of Hellfire missiles—­the ­people and objects docu-
mented since 2007 by Behram refuse to go unseen.31 Among the many ar-
resting images are ­those of survivors in the ruins of their homes, cracked
metal from the shaft of a Hellfire held in their hands like the weight of it might
break them all over again. H ­ ere is the materiality of remote war, stark m
­ atter
that belies claims of surgical precision even as, according to Thomas Stub-
blefield, “­these photo­graphs at the same time acknowledge a certain inad-
equacy of (­human) narrative in this system of drone vision.” In one potent
image, ­children stare into the lens, pieces of rubble offered to the camera and
the remnants of buildings (a home, a school?) all around (figure 1.3). Mark
Dorrian argues that the belatedness of the photo­graphs to the act of vio­
lence—­bodies, homes, and missiles already destroyed—­signals the “violent
cancellation of the possibility of witnessing” in the face of remote war.32 But
I want instead to suggest that ­these images confront the limits of ­human wit-
nessing as the Hellfire fragments, ruined homes, and haunted survivors insist
on richly textured, intimate relations shattered by war.33 They both assert the
radical absence of the technical apparatus of the drone on the ground, but
also insist on that absence as a site of witnessing: its absence is itself a violent
mediation. Against the violent delimitations of the algorithmic systems and
militarized modes of analy­sis that dehumanize ­people into targets, homes
into safe h ­ ouses, and social relations into signs of threat, the material and
affective relations that circulate within and leap from t­ hese photo­graphs
manifest the more-­than-­human wounding and trauma that accompanies
“precision” warfare—­and the inability of military infrastructures to reckon
with or even acknowledge its ongoing presence.
Aftermaths such as t­ hese almost never disturb Western culture or politics,
held at a distance by an apathy t­ oward the unseen. Drone war per­sis­tently
happens over ­there, despite the ramifications of its racializing technopolitics
for publics at home.34 In her history of war’s aerial aftermaths, Caren Kaplan
calls for close attention to “unpredictable yet repetitive intensities of time
and space, disturbing the singular linear or bounded world that we take for
‘real­ity’ in Western culture.”35 Such “rogue intensities” are characteristic of
­wartime, holding the potential to “disturb the everyday experiences of ­those
who might other­wise believe that they are unscathed or untouched folds
places and times onto each other while opening up pos­si­ble affiliations and
historical accountability.”36 Careful attention to the ambivalence, contradiction,

50 Chapter One
figure 1.3. Photo­graph from Dande Darpa Khel, August 21, 2009,
by Noor Behram

r­ esistance, and uncertainty that marks the visual history of the aerial view
is crucial. But this same care can be extended beyond the aerial to its ter-
restrial reverberations and, in par­tic­u­lar, its material, cultural, and affective
registrations. Drone war’s tenuous aftermaths become more response-­able
and address-­able when their witnessing is not a ­human proj­ect alone, but
also heard in the discordant strains of nonhuman witnessing.
In operation, drones flicker on the edge of perception. For ­people living
­under drones, encountering them within the visual field is not uncommon, but
neither can sight be relied upon to warn of an operation in pro­gress.37 On
American missions, militarized drones usually fly high enough not to be seen
at all, or to be caught only in the glint of sunlight in de-­icing fluid as it slides
across the wings and fuselage of the vehicle. Rain can keep them grounded,
while cloudy weather sometimes means lower flights and greater visibility
and tends to be avoided by commanders keen not to alert the surveilled to

Witnessing Vio­lence 51
their presence. But the aural presence of drones is far more constant: a whirr
that cuts through the hum of daily life and grinds against the mind. One man
describes the sound of the drones as “a wave of terror” that sweeps through the
community. Another links their buzz both to the permanent affective state of
fear and to the strain placed on communal gatherings. “When w ­ e’re sitting
together to have a meeting, ­we’re scared ­there might be a strike,” he tells the
researchers. “When you can hear the drone circling in the sky, you think it
might strike you. W ­ e’re always scared. We always have this fear in our head.”38
Alex Edney-­Browne notes that one Afghani slang for drones is bnngina, ­after
the bnng noise that the drones make.39 That buzz works its way into bod-
ies. As Mohammad Kausar, f­ ather of three, says, “Drones are always on my
mind. It makes it difficult to sleep. They are like a mosquito. Even when you
­don’t see them, you can hear them, you know they are t­ here.” If the everyday
disruptions and anx­i­eties of life u ­ nder drones is most pre­sent in their aural
intrusion, then might witnessing not also take place at this level of ears,
sound, and material vibration?
For Schuppli, this earwitnessing strains the limits of what can count as the
material witness of conflict ­because it leaves no trace, even if when “­these low-­
frequency emissions combine with physical ­matter, they vibrate the tympanic
membrane of the ear, so that hearing becomes a kind of barometer for read-
ing the atmospheric pressure of drone surveillance on the body public.”40 Yet
while the lack of trace limits the potential of this aural witnessing to enter the
­legal domain, we nevertheless need to reckon with its registering in the body
as a critical point of contact in witnessing relations. Aural witnessing entails
bodily mediation in the now, yet what it mediates is the virtuality of f­ uture vio­
lence: not simply a warning of potential drone strikes, but an impingement of
the ­future on the sensorium in the pre­sent. Bnngina is the crowding presence
of the aftermath to come, the violent mediation of a pos­si­ble f­ uture.
Witnessing drone warfare from below is as much about making sensible the
enduring, gradual, and uneven vio­lence done to the fabric of life as it is about
registering the spectacular, kinetic vio­lence of the lethal strike. Surviving entails
reworking relations of community and the movements of daily life in counter-
rhythm to the algorithmic operations of intelligence gathering and analy­sis.
Disruptions to daily life and its communal governance are m ­ atters of space
and movement, as well as custom, ritual, and routine. No longer socializing
­after dark, no longer holding community gatherings, no longer undertaking
funeral rites: t­ hese are restrictions on mobility dictated by the uncertainty of
vio­lence from the air.41 They also reflect intensive, shared learning in response
to drone vio­lence, a communal pedagogy of atmospheric war. That pedagogy

52 Chapter One
not only entails re­orienting daily life away from t­ hose activities that the drone
apparatus might mark as threatening—it also involves integrating responsive-
ness to intelligence gathering into daily life such that the potential presence of
the drone reweaves the cultural fabric. This reweaving becomes quite literal in
the incorporation of drone iconography into traditional Afghan war rugs, with
silhouettes of Predators and Reapers replacing the Soviet tanks and Stinger
missiles that found their way into ­these woven images in the 1980s.
In works by Pakistani American artist Mahwish Chishty, this cultural
imbrication of drone vio­lence takes on a more direct critical dimension.
Trained in miniature painting at the National College of Arts in Lahore,
Chishty turned her attention to drone vio­lence following a visit home in
2011. Combining her training in painting with the ornate folk traditions of
Pakistani truck art, Chishty’s Drone Art Paintings (2011–16) and accompany-
ing installations and video works refigure drone technologies as splendidly
vis­i­ble, captured in the vibrant color and gold leaf of finely wrought bricolage
against tea-­stained backgrounds. Painted in opaque gouache, the works in-
sist, as Ronak Kapadia points out, on the permanent visibility of the drone:
materialized not as technoscientific monstrosity but as contained and owned

figure 1.4. Reaper, Mahwish Chishty, gouache and gold flakes on paper, 2015.
Courtesy of the artist.

Witnessing Vio­lence 53
figure 1.5. Image from Drone Shadows installation, Mahwish Chishty, 2015.
Courtesy of the artist.

by the body of the artist, the thick pigment of the paint, and the textured
surface of the paper, exemplified by the painting Reaper (figure 1.4).42 ­These
works are emblematic of what Kapadia calls an “insurgent aesthetics” that
seeks to unsettle the racialized, gendered, and colonial dynamics of empire.
Against the smooth, blank dimensionality of the militarized drone, Ch-
ishty’s paintings segment surfaces into blocks of color, flowers, flag motifs, and
eyes and mouths. For her Drone Shadows (2015) installation, Chishty painted
plastic model kits of Reaper and Predator drones in the bright reds, greens,
and yellows of truck art (figure 1.5). Suspending them in Perspex containers
and using gallery lights to cast shadows, Chishty puts the (in)visibility of drone
warfare in tension with the hypervisibility of the miniatures. In Chishty’s work,
­there is an insistence on returning the nonhuman technics of drone warfare
to the embodied scale of craft and paint. In wrestling with how to figure such
nonhuman vio­lence, Chishty undertakes a kind of nonhuman witnessing
in reverse: testifying aesthetically to the possibility and necessity of making
the seemingly invisible technoscientific mechanisms of vio­lence the briefly
tamed object of art.43 This making vis­i­ble and identifiable is, of course, al-

54 Chapter One
ways only propositional: an address to imagine other­wise, and to resist the
sanitizing discourse that surrounds and obscures drone vio­lence in practice.
[103.158.136.64] Project MUSE (2024-02-06 19:57 GMT) University of Edinburgh

Over ­Afghanistan and in the Federally Administered Tribal Areas (fata)

of Pakistan, drone strikes fit into two broad categories.44 While “personality
strikes” target specific individuals identified by the US as threats (alleged
terrorist or insurgent leaders, for example), “signature strikes” are activated
by emergent patterns in accumulated data about movement and communica-
tion cross a certain threshold on a predefined decision matrix. Collected by
drones carry­ing the gilgamesh cell phone snooping equipment, metadata
from cellphone tower check-­ins, calls, and texts is analyzed by skynet soft-
ware to identify “patterns of life” that could be mapped to potential threats
or targets of interest.45 But cell phones have strange lives—­s im cards can
be swapped, phones shared—­and in many places the status of vari­ous per-
sons can be multiple and contextual—­local elder in one context, warlord
in another. Preemption obscures such specificities in ­favor of what can be-
come operationally subject to tools that identify risk and act to eliminate
it.46 Even if the technical details ­aren’t known on the ground, the felt force
of potential vio­lence permeates daily life. Camouflage is phones changing
hands, sim cards circulating, gatherings avoided. Life’s textures transform,

Witnessing Vio­lence 55
binding existence more tightly to war. ­People on the ground speculated to
the Stanford and nyu researchers about paid ­informants, worried about
“chips” and “sims” placed in cars and h ­ ouses, and complained of eroding
community trust and an atmosphere of paranoia.47 The learned protective
practices of p ­ eople ­under the martial gaze testify to an intensive relation to
the potential for death that manifests in the drone apparatus. Movements,
changing cultural patterns and practices, a folding into life of the vagaries
of the algorithm—­these are a kind of collective witnessing to the nonhuman
assemblages of signature strikes and their algorithmic architectures of intel-
ligence gathering and objectification.
Surviving drone warfare is, however, as much a ­matter of chance as anything
­else.48 How surveillance analysts and signals intelligence pro­cesses capture and
classify bodies, movements, and social relations is deeply contingent. Death,
too, entails the randomized destruction of living bodies into ruined flesh. Algo-
rithmic killing, or “death by metadata” in Pugliese’s formulation, is far from the
technocratic ideal. While the language of surgical strikes and precision warfare
suggests some sanitized form of vio­lence, the real­ity on the ground is very dif­
fer­ent. Lethal strike survivor Idris Farid describes “pieces—­body pieces—­lying
around” and the effort to “identify the pieces and the body parts” to determine
“the right parts of the body and the right person.”49 Delving into the horrific
vio­lence of an attack on a village in Yemen, Pugliese writes that distinguishing
between animal, child, and adult was often impossible, bodies fused into a
“composite residue of inextricable flesh. The one melts into the other. The one
is buried with the other.”50 While the targeting systems and discursive logic of
drone warfare dehumanizes through techniques of gendering and racializing,
its vio­lence strips its victims of any corporeal distinction from other animals.
Reducing the living to “scattered fragments of undifferentiated flesh,” animal
and ­human bodies become what da Silva calls “no-­bodies” and Pugliese labels
“nothing less than generic, anomic, and wholly killable flesh.”51 Even the land
is scarred. As one survivor put it, “The entire place looked as if it was burned
completely,” so much so that “all the stones in the vicinity had become black.”52
This ruination to h ­ uman, animal, plant, and inanimate entities signals the
limits of a witnessing that centers the h ­ uman: How can a narrow humanism
account for vio­lence that strikes at the very vitality of more-­than-­human
ecologies? This enfolding of more-­than-­human environments with ­human
flesh demands what Pugliese calls forensic ecol­ogy. His vision of a radical fo-
rensics sees testimony as “a relational assemblage of heterogeneous materials
that, collectively, is mobilized to speak an evidentiary truth.”53 While mobili-
zation within a framework of laws typically depends upon a speaking subject,

56 Chapter One
the registration of vio­lence enacted on the sites of drone strikes constitute a
form of witnessing that both precedes and exceeds the ­human. It precedes
the h­ uman b ­ ecause the air’s mediation of light in the collection of sensing
data and of force in the on-­rush of Hellfire missiles is already witnessing ru-
ined flesh, scarred rock, and shattered plant life in the instant of explosion.
It exceeds the ­human ­because this witnessing occurs below the threshold of
detectability—in the faint striations of dirt subject to passing shrapnel, in
the misting of viscera, in the ephemerality of heat—­and far outside it, too,
in the elusive scale of the drone apparatus itself. Translating such witness-
ing into frameworks of individualized responsibility is impossible, not least
when the drone apparatus itself is so dispersed as to make no one singularly
responsible for any given strike.54 Yet while this combination of fused flesh
and machinic occlusion of responsibility certainly signals the limitations
of rights-­based frameworks for dealing with the vio­lence of increasingly
autonomous warfare, it also suggests the necessity of a conceptual means of
dwelling with the thick and messy confluence of forces that produces t­ hese
horrors. Such a dwelling-­with can only be pro­cessual and can only reckon
with the vio­lence of drone war as ­process. As a ­process of registration—­which
is to say, of the violent event mediated into the more-­than-­human flesh of the
world—­nonhuman witnessing offers critical purchase, insisting on attending
to both the thick knots that bind vio­lence, as well as the tenuous strands of re-
lation that shimmer out of reach within ecologies and technical systems alike.
On the other side of the drone sensor array, aftermaths of vio­lence are me-
diated very differently. In the form that has dominated the last twenty years
of remote warfare, drone sensors display sensing data in visual images on the
screens of operators located in ground control stations far from the battle-
field. Replicated on the terminals of ­lawyers, commanders, image analysts
and, in certain situations, officers commanding troops on the ground, the
principal lens for drone operations is e­ ither optical or thermal full-­motion
video overlaid with gis, timestamp, targeting, and other key information.
With the arrival of the war on terror, Parks shows how media coverage “made
vertical space intelligible to global publics in new ways and powerfully re-
vealed what is at stake in being able to control the vertical field.”55 Media
coverage of the invasions of ­Afghanistan and then Iraq rendered the aerial
view familiar, training publics to recognize and decode new ways of seeing.56
According to Roger Stahl, drone vision “invited publics to see the drone war
through the very apparatus that prosecuted it,” and in ­doing so “framed out
­those populations who must live and die u ­ nder this new regime of aerial
occupation,” rendering them vulnerable, invisible, and ungrievable.57 When

Witnessing Vio­lence 57
drone war does intrude on the mediascape of the United States, the United
Kingdom, Australia, France, Denmark, or elsewhere in the West, it does so
through the existing profusion of screens, stories, images, and mediated
encounters. Drone warfare pre­sents distinct challenges for witnessing that
take place in, by, and through journalistic media.
Vis­i­ble in YouTube videos of Reaper strikes and part of the visual rhe­
toric of films such as Eye in the Sky (2015), the event of a missile strike over-
comes the sensory capacity of the drone: a burst of white, intensities of light
that overwhelms the optical camera and of heat that undoes the thermo-
graphic sensor.58 Focalizing infrared radiation through the lens and onto
the microbolometers assembled one-­per-­pixel into the sensor itself, ther-
mographic cameras have to manage wider wavelengths than their opti-
cal counter­parts. For Nicole Starosielski, “the infrared camera is not just
another thermal medium alongside thermostats, sweatboxes, and heat
ray guns: it is a technology whose sensing capacities work to transform
all m
­ atter, ­whether bodies or buildings, into thermal media itself.” The im-
ages it produces depend upon the recasting of “the world as a landscape of
infrared reflectors and infrared emitters—as a field of thermal communica-
tion.”59 Sometimes, that field overwhelms the camera’s thermoceptive capac-
ity. When a missile strikes, the combination of ­limited resolution and intense
heat prevents infrared sensors from ­doing anything but assigning maximal
intensities—­computer vision cannot resolve what it cannot sense. ­Whether
in optical or infrared, this incapacity to capture the event of the strike means
that drone sensors necessarily repeat the erasure of life at the level of sensor
­process. From within the drone apparatus, the aftermath is always obscured
by the destruction itself, the wreckage of buildings and bodies, thick smoke,
and the heat of melted ­matter. Inhuman vision reveals its inhuman sensoria,
yet what h ­ uman sensorium would not be shocked and undone by witnessing
such a t­ hing? In the aftermath, sensor operators typically shift to infrared
to identify the movements of bodies and the still-­warm flesh of the dead.
Prescribed by the requirement to count all dead as military-­aged males, as
threats ­until proven other­wise, military personnel decipher the aftermath
according to a rubric designed to repeat visceral, material vio­lence in infor-
mational form. This reading of the scene—­a kind of brute forensics—is often
yoked to the question of additional strikes. ­These so-­called double taps are
often conducted at a delay intended to flush out further threats, but are far
more likely to kill or wound anyone who rushes to assist at the scene, a fact
that means bystanders often choose to listen to their neighbors die rather
than risk being killed themselves.

58 Chapter One
Not only are ­these sensors overwhelmed, but network latency also means
that the drone apparatus can only ever witness on a two-­to six-­second delay.
­Whatever appears on screen does so with the event already in the past, not
quite real-­time but still live in the sense that the drone system always experi-
ences liveness on delay. Distance vanishes, but time dilates. Drone systems
intensify this tension between occurrence and technical mediation: an elastic
temporality brimming with vio­lence. Yet this latency also contains within it a
certain necessary trauma, a deferral of the traumatic event into the durational
virtuality of an arrived and arriving f­ uture. Produced by the combination of
distance and transcoding between components of the network, this latency is
one temporality of violent mediation, a time in which nonhuman witnessing
takes place in the ambivalent space of the drone apparatus itself. This mode
of nonhuman witnessing has ­little corporeal immediacy or ­political valence,
but it is witnessing that registers vio­lence distributed in both time and space.
Seen in this way, the violent mediations of the drone apparatus remind us
that nonhuman witnessing carries no inherent ethics, no necessary tendency
­toward justice, only an insistence on the complexity of registering an event
as knowable. For ethics, morality, or justice to enter the frame, the question
has to become one of testimony—of the bearing of witness ­after the event
of witnessing itself. If the drone apparatus is, in its own ambivalent way, a
witnessing machine, if a hostile one, then it is one that must in turn be wit-
nessed. That challenge is amplified by new technologies that augment the
sensory capacity of the drone through on-­board advanced computing. But
before turning to one such technology, Agile Condor, I want to first consider
nonhuman witnessing in the aftermath of war in Aleppo, Syria.

witnessing aleppo

While the aerial view of war is rightly associated with surveillance, control,
and vio­lence, remote sensing systems and civic drones can also be harnessed
as witnessing apparatuses for publics and researchers.60 Such uses of sensing
technologies reveal their partial, contested, and contingent nature, as well as
the fraught politics of control that suffuse both atmospheric sensing and digi-
tal infrastructures.61 Aleppo, in Syria, is a case in point. In March 2011 and
amid the Arab Spring, prodemocracy protests in Daraa against the regime
of Bashir al-Assad ­were brutally suppressed. When anti-Assad supporters
rebelled across the country, Syria swiftly fell into civil war, which in turn
produced power vacuums in vari­ous regions and enabled the Islamic State in

Witnessing Vio­lence 59
Iraq and Syria (isis) to take root. Fought across four years from 2012 to 2016,
the ­Battle of Aleppo saw what the United Nations called “crimes of historic
proportions” committed by Syrian, rebel, and international forces, including
via ­Russian, American, and Turkish air strikes from crewed and uncrewed
aircraft.62 By the time the city was retaken by the Assad regime, some 31,273
civilians ­were reported dead and numerous culturally significant sites w ­ ere
destroyed or damaged according to a unesco conversation report, including
the destruction of the ­Great Mosque and the eleventh-­century minaret of the
Ummayad Mosque. Aerial and artillery bombardment ruined roads, homes,
schools, hospitals, and entire neighborhoods, reshaping the city in fundamental
ways and transforming life for its h ­ uman and nonhuman inhabitants.
Rather than containing the vio­lence, the application of “precision” weap-
ons such as drones and guided missiles seemed only to intensify the destruc-
tion: imagery of Aleppo in 2016 bears a remarkable similarity to that of Berlin
in 1945. W­ hether a missile was launched from a drone or manned h ­ elicopter
is in some ways immaterial to the destruction it ­causes on the ground: the
dead remain dead, homes remain ruined. But in Aleppo the view from above
has afforded a more ambivalent relation to aerial aftermaths than is always
the case, a phenomenon revealed in dif­fer­ent ways by the Conflict Urbanism:
Aleppo proj­ect from the Center for Spatial Research at Columbia University
and drone video by Aleppo Media Center, an antigovernment activist group
responsible for widely shared and republished footage.
Conflict Urbanism uses remote sensing imagery, geolocation data, and
open-­source software tools to create an accessible digital platform for track-
ing the city’s ­wartime aftermaths. As artist, academic, and proj­ect lead Laura
Kurgan points out, “while war demolishes, it also reshapes a city, and, how-
ever difficult it is to imagine rebuilding in the midst of a war, Aleppo is being
restructured and w ­ ill be rebuilt.”63 The core of the proj­ect is an interactive
map that reveals damage to the city’s urban fabric by layering high-­resolution
satellite images with data from unitar’s unosat (the United Nations Satel-
lite Center, run by the United Nations Institute for Training and Research). In
its remediation of satellite imagery into an activist-­aesthetic context, Con-
flict Urbanism: Aleppo continues Kurgan’s long-­standing research practice
engagement with the politics of remote sensing imagery.64
From the main site hosted by the Center for Spatial Research, users are
able to engage with the city at the neighborhood scale, moving through
time and at dif­fer­ent resolutions to track the damage to the city (figure 1.6).
This use of technics to make vis­i­ble other­wise obscured transformations
to the more-­than-­human environment of the city succinctly encapsulates

60 Chapter One
figure 1.6. Image showing areas of intense damage, Conflict Urbanism: Aleppo

the ambivalence of nonhuman witnessing of vio­lence. Death, displacement,

and destruction are rendered legible beyond the structural and infrastruc-
tural damage to the city itself, with individualized accounts from YouTube
videos geolocated onto the map to provide an alternative ground truth.
Seeking to intervene in the politics of war by making spatial and temporal
scales of vio­lence knowable to ­humans, the proj­ect shows how nonhuman
witnessing—­satellite sensing, drone vision, material scarring, ecological
disruption—­can broaden what counts as testimony within h ­ uman polities.
But it also lays bare the power that resides in control over access to and tasking
of remote sensing satellites, as well as who and what counts as witnessing,
witness, or testimony.
While Conflict Urbanism provides a kind of nonhuman witnessing infra-
structure in its own right, the proj­ect is also concerned with interrogating
the limitations of that infrastructure and developing transferrable techniques
that might be deployed to understand other urban conflicts (figure 1.7). A
crucial ele­ment of the proj­ect is thus probing the repre­sen­ta­tional politics
of satellite imagery made evident through constraints of access, resolution,
legibility, and literacy. With some l­imited exceptions, remote sensing satellites
that produce public data and imagery are e­ ither operated by the US gov-
ernment (such as nasa’s Landsat) or ­under its auspices, as in the case of the
­ikinos satellite and its successors. While Landsat’s mission is the continuous

Witnessing Vio­lence 61
capture of multispectral data of the earth, private satellite infrastructures
only take images they are tasked to collect. Users need to purchase satellite
time and specify locations. While the images produced can then be pur-
chased by ­others, the costs of tasking and purchasing can be prohibitive
for noncommercial or nonstate actors such as h ­ uman rights o­ rganizations.
Depending on the satellite, resolutions down to around 0.25m are available
for public purchase, but for ­decades the US government ­limited commercial
resolutions to 0.5m to keep h­ uman bodies illegible.65 This can make the work of
conflict monitoring more difficult, obscuring the movement of p ­ eople but also
the damage to buildings from non-­incendiary missiles launched by drones.
Through an experimental approach, the proj­ect produced an algorithmic
dataset using open-­access satellite images to m ­ easure brightness in pixels
between successive images. This stitching together of spatial images across
66

temporalities allows the tracking of damage done to the city. Ground truth
for the proj­ect imagery was established via high-­resolution satellite imagery,
as well as through the calibration and geo­graph­i­cal location necessary to
the operation of remote sensing satellites. But the proj­ect also produces
a relational ground truth as images are compared, synthesized, and syn-
chronized.67 By foregrounding how this method is “messy and riddled with
ambiguity,” the proj­ect exposes the constructed and frictional nature of such
relational ground truthing. It reveals material, nonhuman traces of the wit-
nessing apparatus itself, a violent mediation within the witnessing of the city’s
destruction, in which low resolution obscures texture and specificity.

figure 1.7. Image of interactive map, Conflict Urbanism: Aleppo

62 Chapter One
Alongside its tracking of ­human activity, such as the displacement of
­people from ruined sections of the city to settlements on its outskirts, the
proj­ect also witnesses the complex interplay between urban environment,
vio­lence, media, mobility, and renewal. Rather than focusing tightly to spe-
cific sites of airstrikes, Conflict Urbanism attends to “what surrounds the
circles—­the areas contiguous to the damaged sites—in order to ask ques-
tions on an urban scale.”68 Such an approach enables a witnessing of vio­lence
that centers the intentional and incidental destruction of cultural memory,
urban history, and community ecologies. This witnessing exceeds the h ­ uman
but does not abandon it. By foregrounding the limitations of the platform,
keeping it open to collaboration and development, and directly addressing
issues of data neutrality, the proj­ect exemplifies the necessary contingency of
nonhuman witnessing. In Aleppo, urban vio­lence registers its traces in Sch-
uppli’s material witnesses: wood, concrete, steel, glass, and asphalt as much
as in remote sensing systems, or indeed in the testimony of ­those displaced
residents of the city. In an environment in which ­people have been driven
from their homes, t­ hose nonhuman material witnesses capture something
that the displaced have left b ­ ehind: the material and affective traces of de-
struction, loss, and absence of life.
Integrated into the online platform are YouTube videos captured on the
ground, what Lilie Chouliaraki and Omar Al-­Ghazzi call the “flesh witness-
ing” of digital materials recorded and shared by ­people in conflict zones.69
­These videos capture the ­angles, color, texture, and immediacy lacking in
the layered sensor data. Among them are drone videos produced by activists
from the Aleppo Media Center. Shot at the now-­familiar but still uncanny
vantage of the drone—­hovering above or just below rooftop, moving with in-
human smoothness, footage rendered with an almost too-­sharp definition—­
this footage mediates the vio­lence of the aftermath. While mainstream media
coverage of Aleppo’s destruction featured drone footage from a range of
sources, including the ­Russian military, the video shot by the Aleppo Media
Center insists on capturing ruined streets, homes, shops, and squares, and
in d
­ oing so both reveals and obscures the vio­lence (figure 1.8). While drone
footage is always imbricated in the militarism of the aerial view, it can none-
theless be deeply affecting. As Kaplan writes: “We absorb ­these views to
such a degree that they seem to become a part of our bodies, to constitute
a natu­ral way of seeing.”70 This capacity to enfold nonhuman vantages into
the ­human sensorium speaks to the malleability of our perception, but also
to our cyborg existence, to the always more-­than-­human nature of h ­ uman
sensoria and knowledge-­making.71

Witnessing Vio­lence 63
figure 1.8. Still from drone footage, Aleppo Media Center

As nonhuman agents of technological perception, drones transect space

and time to si­mul­ta­neously draw us nearer to p­ eople and places and amplify
or highlight our separation. Drone witnessing enables mediated intimacy
with distant events, yet it also reinforces remoteness, placing the viewer in an
uncanny relation to what enters the frame of the drone’s camera. If the aerial
view of war has become a natu­ral way of seeing, what is outside the frame or
within but obscured bears close scrutiny. In a provocative essay on the mass
rape of ­women in Berlin ­after the fall of the city at the end of World War
II, Ariella Azoulay argues that the absence of sexual vio­lence from photo­
graphs of the ruined city means that witnessing depends upon attending
to the affective and sonic registers of images. For Azoulay, photo­graphs of
damaged buildings, off-­duty soldiers, and wrecked cars obscure vio­lence and
injustice. “[Rape] was ubiquitous,” she writes, “but still, it did not appear as a
prime object for the gaze of ­these photog­raphers, in the way the large-­scale
destruction of cities did.”72 While mass rape at scale might not be an object
that the photo­graph can capture, some of the tens of thousands of individual
rapes could have appeared in photo­graphs. The blown out second story of
an apartment building might have been the site of rape; a w ­ oman might be
raped even as the photo­graph is being taken. Yet this vio­lence never appears
in the images. This absence of sexual vio­lence calls for a reckoning with the

64 Chapter One
violent mediation that makes it pos­si­ble: attending to what is pre­sent in such
photo­graphs as participating in an affective production of that which is not.
While a certain material intimacy exists between the film negative photo­
graphs of postwar Berlin and the city and its vio­lence, this task of witness-
ing the absence of vio­lence is complicated by the machinic vision of drone
video in Aleppo. For Azoulay, photo­graphs of spaces in which widespread
and systemic vio­lence took place but is not shown pre­sent an injunction to
witness the photo­graphs through the historical knowledge of an absence of
visual evidence. She thus reads “­these perforated h­ ouses, heaps of torn walls,
empty frames, uprooted doors, piles of rubble—­all ­those ele­ments that used
to be pieces of homes—as the necessary spatial conditions ­under which a
huge number of w ­ omen could be transformed into an unprotected popula-
tion prone to violation.”73 Drone imagery from Aleppo shares much with the
photo­graphs analyzed by Azoulay: perforated walls, piles of rubble, blasted
win­dows, shattered sidewalks, distended roadways. It obscures the 31,273
civilians dead, the many more displaced, and the rape, theft, wounding, and
loss that accompanies such undoing of a city. Unlike the analog photography
of postwar Berlin, machinic vision does not imprint the light of the world
in the gelatin material of the film negative, but rather translates the fleeting
response of the optical sensor directly into pixels, stored as code and only
rendered in visual form for the benefit of the ­pilot and, l­ ater, the audience of
any distributed recordings.
Drone footage of w ­ artime’s aftermaths in Aleppo mirrors pro­cessually
the vio­lence of aerial war, with its digital targeting systems, guided muni-
tions, and sensor capture of the environment. But it reveals l­ittle of t­ hose
workings: drone footage of Aleppo is what remains within the machinic
frame but hidden both by the depopulated city and the technics of the sensor
itself. Integrated into the Conflict Urbanism mapping apparatus, this foot-
age both grounds and is grounded by multispectral satellite images. Drone
footage introduces a more-­than-­human visuality that is nonetheless tied to
line-­of-­sight operation and the practical constraints of battery life and signal
strength: it returns the aerial view almost to the body and yet also retains a
nonhuman detachment that heightens the witnessing of war’s aftermath.
Within the aftermaths of con­temporary war’s violent mediations, witnessing
must pursue the tactile and affective, but also the machinic, technical and
networked architectures of seriality and sensing. Yet the nonhuman percep-
tion of drones and remote sensors is increasingly not only an extension of
­human sense-­making, but also an augmentation at the level of identification
and decision.

Witnessing Vio­lence 65
augmenting the drone apparatus

Developed by src Inc. and flight-­tested by General Atomics on its mq-9

Reaper drone, Agile Condor is an on-­board targeting system designed to re-
solve both network bandwidth and analytical resourcing prob­lems that limit
the efficacy of remotely pi­loted systems.74 Built to analyze large quantities of
data from the drone’s sensor apparatus in real time, this computer system is
embedded in a pod that can be fitted to the wing of a drone, replacing one
Hellfire missile from its payload. Developed in conjunction with the Air
Force Research Laboratory (afrl), its makers claim that Agile Condor is
an artificial intelligence targeting system capable of analyzing video footage,
synthetic aperture radar imagery, or infrared camera imagery with the capac-
ity to detect, categorize, prioritize, and track potential targets (figure 1.9). By
undertaking so much image pro­cessing autonomously at the edge of the
network, Agile Condor only sends imagery it deems to meet a threshold of
value, cutting down latency, and relieving pressure from overstretched mili-
tary networks. By only delivering sensor data of potential interest, the system
also alleviates the accelerating need for highly skilled image analysts and al-
lows them to focus on potential targets rather than sift through vast amounts
of irrelevant imagery. While Agile Condor cannot make a determination to

figure 1.9. Agile Condor Operations Concept, Air Force Research Lab

66 Chapter One
strike, it sets the background conditions for what might be worthy of closer
attention and potential lethal action. It thus exemplifies both the violent
mediation of the drone apparatus, but also its liminal status between h ­ uman
operation and lethal autonomy.
More autonomous data pro­cessing at the point of perception marks a
qualitative shift in the agential composition of warfare. Autonomous military
systems are not in themselves new—­loitering munitions have been used by
Israel since the 1970s; the sage system designed to monitor Soviet nuclear
launches was built in the 1950s—­but Agile Condor integrates autonomous
perception into an already complex kill chain, inserting a machinic intel-
ligence that preemptively shapes the fields of possibility for ­human analysts
and operators. Agile Condor thus constitutes a kind of liminal, nonhuman
witness: it (pre)determines the meaning and significance of objects and
events, presenting them as open to address by the remote warfare system.
Through the operative role of its on-­board high-­performance computer, the
ai pod siphons off ­human agency in the name of efficiency. No longer ­will
­human analysts be concerned with discerning the figure of threat against
the ground of life, but only with the array of figures presented as action-
able. In the transcript of the drone strike that opened this book, it becomes
clear that almost two dozen p ­ eople ­were killed in no small part b
­ ecause the
figures in view obtained an affective potency divorced from the milieu in
which the convoy moved. That is, mission atmosphere oriented the operators
and every­one ­else involved ­toward vio­lence. Agile Condor entrenches this
orientation ­toward identifying foes and not friends into the milieu itself: a
machinic perpetrator, its witnessing tends ­toward vio­lence.
In an oft-­cited passage of War and Cinema, Paul Virilio writes that “along-
side the ‘war machine,’ ­there has always existed an ocular (and l­ater optical
and electrooptical) ‘watching machine’ capable of providing soldiers, and
particularly commanders, with a visual perspective on the military action
underway. From the original watchtower through the anchored balloon to
the reconnaissance aircraft and remote-­sensing satellites, one and the same
function has been in­def­initely repeated, the eye’s function being the function
of a weapon.”75 This mechanization of perception involves “the splitting of
viewpoint, the sharing of perception of the environment between the animate
(the living subject) and the inanimate (the object, the sensing machine).”76
This splitting of perception entails not only the ­human and lens, but also an
entire technical apparatus that is motorized, electrical, computational, and
increasingly autonomous: what Virilio calls the “logistics of perception.”77
While not on the same order of magnitude as the arrival of networked warfare

Witnessing Vio­lence 67
itself, edge computing is an impor­tant intervention in t­ hese logistics b ­ ecause
it yokes the ontopower of perception to the necropo­liti­cal capacity to make
die. Unlike the s­ imple reactive relation between sensing and killing found
in an improvised explosive device (ied) or land mine, the ai intermediary
enabled by high-­performance edge computing means that deterministic opera-
tions happen at a spatial and temporal remove from ­human agents.
But while president of General Atomics David R. Alexander claims that
Agile Condor’s “ability to autonomously fuse and interpret sensor data to de-
termine targets of interest is at the forefront of unmanned systems technology,”
edge computing is not confined to military applications or even to drones.
In fact, it originates in commercial prob­lems of bandwidth and latency pro-
duced by the move t­ oward cloud computing architectures. Early edge com-
puting can be found in “cloudlets” such as content delivery networks that
cache web data closer to users so that, for example, ads can be served faster
and more responsively, preventing delays in page loading caused by the need
to pull data from distant, centralized data centers. Edge computation now ex-
ists in every­thing from networked security cameras to automated agricultural
systems, reducing the flow of data to central control points. Against the push
to centralize control via the capacity of networks to distribute information,
edge computing offers the potential to decentralize control while retaining
centralized authority. Such a tendency can only produce ever more radical ab-
sence, as experiences of the world are distributed, remediated, and rendered
computational even as they become operative and immediate. In war, this
fusion of sensing, classifying, and selecting within black-­boxed technologies
signals an increasing a­ cceptance of computational agencies on and above the
battlefield, a machinic corollary to the shift of the US military to special forces
operations. Where the military media technologies of the twentieth ­century
­shaped and ­were ­shaped by mass, t­ hose of the twenty first are devolved, di-
vidual, and distributed. Like power itself, military media have pushed more
and more computation to the edge of the logistics of perception.78
Artificial intelligence is particularly appealing for dealing with sensor data
­because the first action required is to sift for items of interest, something that
machine learning is—in theory, at least—­particularly well situated to do. But
standard methods of machine learning analy­sis require power­ful graphics
pro­cessing units (gpus), particularly if the system ­will also learn on the fly.
That means significant power loads and accompanying heat. Consequently,
huge dividends can be achieved through computational techniques—­both in
terms of hardware and software—­that reduce the need for power, via both
more efficient cir­cuit design and learning systems that only fire when needed.

68 Chapter One
According to both its marketing material and vari­ous technical papers pub-
lished by the development team from src and afrl, Agile Condor uses a
neuromorphic architecture modeled on ­human neural systems.79 In other
words, its capacity for discriminating perception is intended to mimic neu-
robiology in contrast to typical parallel pro­cessing architecture. Both the ibm
TrueNorth and Intel Loihi experimental pro­cessors used by the Agile Condor
can be traced to a darpa proj­ect called Systems of Neuromorphic Adaptive
Plastic Scalable Electronics (SyNAPSE), launched in 2008 to develop revolu-
tionary new neuromorphic pro­cessors and design tools. In contrast to typical
machine learning image analy­sis that addresses entire images, neuromorphic
systems such as Spiking Neural Network architectures are designed so that
individual “neurons” within the system can fire in­de­pen­dently and directly
change the states of other neurons. ­Because information can be encoded di-
rectly into the signals themselves, spiking networks are not l­ imited to binary
states and can thus produce something closer to the analogue workings of the
brain, more proximate to the early cybernetic dream before it veered ­toward
an altogether dif­fer­ent computational rationality.80 ­Because ­these neurons
only work when “spiked,” the network consumes significantly less power and
can autonomously gear up to higher capacity as needed. Neuromorphic sys-
tems such as Agile Condor are prime examples of what Andrejevic calls “au-
tomated media,” or “communication and information technologies that rely
on computerized pro­cesses governed by digital code to shape the produc-
tion, distribution, and use of information.”81 Harnessed to the martial gaze,
automated media reveal how, as Bousquet puts it, “the ­human sensorium has
been slowly and surely directed, mediated, and supplanted in s­ ervice to the
ultimate imperative of targeting.”82
As with so much emergent military technology, exactly how Agile Condor
might function in a battlefield context is impossible to ascertain. In a series of
articles published in vari­ous ieee forums between 2015 and 2020, the research
team from afrl and src Inc. reveal snippets of insight about the compu-
tational architecture and machine learning techniques used in the system.83
Using a mix of machine learning model types, including spiking neural net-
works and the MobileNet architecture, the researchers demonstrate a bal-
ance between accuracy and efficiency across a series of prototypes built on
ibm and Intel pro­cessors. Working with a range of test datasets that include
optical satellite imagery from the United States Geological Survey, vari­ous
experiments achieve object recognition accuracy of more than 90 ­percent,
depending on the specific technical arrangement. A similar accuracy was
maintained using imagery from the Moving and Stationary Target Acquisition

Witnessing Vio­lence 69
and Recognition (mstar), a joint darpa and afrl program that collected
and pro­cessed sar imagery of vari­ous military targets. But that dataset, while
public, was produced in 1995 and its resolution has been far exceeded by
con­temporary satellite imagery. As such, even though the technical informa-
tion about vari­ous chip, pro­cessor, and model configurations is in­ter­est­ing,
­these publications give no indication of how the Agile Condor targeting
system would work in practice. What data w ­ ill it be trained on? How w­ ill
it be verified and ground-­truthed? How are its determinations presented to
operators and analysts? Does it have its own interface or is it integrated into
existing ground control station control systems? What information is made
available back through the system about modeling, probability, and so on?
How much on-­the-­fly learning is the system capable of executing, and what
quality control mechanisms are in place to verify accuracy or intervene in
the learning ­process?
With the in-­practice workings of the apparatus itself largely foreclosed,
we can turn instead to the promotional materials for an articulation of the
military imaginary that animates Agile Condor. In a two-­minute video pro-
duced by src Inc., Agile Condor is presented as a power­ful tool for saving
lives and preventing vio­lence.84 Rendered in computer graphics that share
­ opular video games such as the Call of Duty
the gritty, lens-­flare aesthetic of p
series, a General Atomics Reaper drone takes off from a mountainous air
force base to a dark techno soundtrack. Cruising at night above a dense urban
environment, its sensor system identifies vari­ous objects, marking them with
glowing green squares. Then Agile Condor kicks in, automatically analyzing
incoming imagery (figure 1.10). Dramatized as a clichéd array of image feeds
entering the hardened box of the computer itself and headlined in multiple
places with the term “neuromorphic computing,” the Agile Condor swiftly
does its magic and an alert flashes up: threat detected. Cut to a swarthy
figure with an rpg on his shoulder, then a convoy of vehicles, and back to
the aerial view. Now, the convoy vehicles are marked in blue and the threat
in red. The sensor pulls focus onto the threat and zooms in tight, resolving a
high-­resolution image that it then runs through a facial recognition system
to obtain a 98 ­percent match (figure 1.11). Signal streams back to command,
where “Agile Condor with neuromorphics enabled has detected an imminent
threat.” The convoy can now be diverted and ­helicopters sent to arrest the
would-be assailant.
Hyping the efficacy of the system in producing a swift, bloodless reso-
lution is not unusual for this genre of military technology videos, but the
­presentation of its technics is revealing of the imaginaries that animate

70 Chapter One
figure 1.10. Agile with “neuromorphics enabled,” still from “Agile Condor™ High-­
Performance Embedded Computing Architecture,” YouTube video, October 15, 2016

figure 1.11. Still from “Agile Condor™ High-­Performance Embedded Computing

Architecture,” YouTube video, October 15, 2016

military desires for ai systems such as Agile Condor. Sensor capture, image
analy­sis, threat determination, geolocation, signals transfer, and operational
actualization are all presented as seamless, frictionless pro­cesses. Wide area
surveillance captures data at scale, which is then immediately transduced
into the Agile Condor analytic engine to identify and locate an imminent and
incontrovertible threat. How t­ hose analytics take place is obscured: Does the

Witnessing Vio­lence 71
system first identify a static figure in the dark? Does it map onto the con-
voy? What are the relations between ­those ­things? Is it correlating between
dif­fer­ent sensor feeds in real t­ ime? Once the threat is detected, the capacity
to recognize a face—­something not mentioned in the technical papers pub-
lished to date by the Agile Condor team—­provides a granular, individualized
level of analy­sis. This slippage in scale—­from the unknown, impenetrable
urban environment to the named identity of an individual—­exemplifies
the god trick that animates both the militarized view from above and the
artificial intelligence system. Agile Condor figures as a watchful guardian,
capable of oscillating between scales and presenting immediately actionable
information to a hyperresponsive command center. Despite the immediate
threat of vio­lence, the response is ­measured and clinical. Precision warfare
performed through automated media promises to facilitate bloodless control.
As is often the case in military promotional materials, the use cases pre-
sented for public consumption veer closer to policing than mass or even
“precision” vio­lence. Nevertheless, we can observe what Andrejevic calls the
“cascading logic of automation” in which “automated data collection leads to
automated data pro­cessing, which, in turn, leads to automated response.”85
This cascading logic has an inherent connection to the death drive, exempli-
fied by the development of Lethal Autonomous Weapons, but evident in
technologies such as Agile Condor, which are not only designed to facilitate
the application of lethal force but also to be part of the ­process of tipping over
the threshold into ever more complete autonomy. More specifically, Agile
Condor can be understood as operating in the mode of preemption, which
“dispenses with the question of causality: it takes as given the events it targets,
relying on comprehensive monitoring and predictive analytics to stop them
in their tracks.”86 Neural network analy­sis of sensor data is preemptive in
this way, filtering through data streams for sets of image characteristics that
map to par­tic­u­lar models. Presenting the correlative outputs of this analy­sis
works to preempt interpretation, framing every­thing presented as potentially
actionable. This direct intervention in the becoming-­target of p ­ eople, struc-
tures, and ecologies, reveals Agile Condor as an operative expression of what
Massumi calls “ontopower”: the power to bring into being. Agile Condor and
all such autonomous systems do not simply identify targets but produce them
through their violent mediation of the world around them, binding affect and
encounter into the knowledge apparatus of the ai-­enabled drone.
Agile Condor points to the existence of a machinic witnessing operat-
ing exclusively within an algorithmic domain inaccessible to the ­human.
This machinic witnessing occurs alongside the preemptive determinations

72 Chapter One
that the system makes. Diffracted through the hunt for emergent threat and
within the loop of sensor capture and algorithmic identification, classifica-
tion, filtering, and prioritizing, this mode of relation to the event probes the
limits of witnessing, as the next chapter examines in detail in the context of
learning algorithms. Within the broader milieu of drone warfare, any wit-
nessing that occurs within the Agile Condor system needs to be understood
in relation to its consequences for the witnessing that takes place within the
wider apparatus and in conjunction with its h ­ uman actors. “Prosthetically
tethered to the war machine,” writes Bousquet, “the combatant’s cognitive
and neurological ­labors are hitched ever more tightly to cybernetic control
loops, mind and body subsumed into complex assemblages that render the
locus of agency increasingly diffuse and uncertain.”87 This dispersal of agency
throughout the system means that witnessing—as a mode of relation that
binds agencies to events—is also diffused. This diffusion concentrates within
par­tic­u­lar pockets of intra-­activity, sites of intensity where perceptual trans-
ductions take place, and where determinations are rendered in relation to
the data produced. If preemptive technologies such as Agile Condor seek
to cut through the inefficiencies of symbolic, narrative, and causal analy­sis,
they also undo the grounds of evidence itself by presenting the (potential)
need for action through an operative frame detached from the complexities
of the world beyond the sensor.88 In this sense, the machinic witnessing at
work within the technical constellation of Agile Condor pod, sensor array,
and aerial drone constitutes a kind of witnessing without evidence. For the
­human operators, analysts, and commanders looped into such cybernetic
controls system to varying degrees of intimacy, witnessing is already violently
mediated by the preemptive shaping and techno-­authority of the targeting
system. For ­those operators “seeing” war through the machinic eye of auto-
mated imaging and analy­sis systems, witnessing drone vio­lence is inescap-
ably nonhuman. Not only ­because the apparatus mediates what is captured
by its sensors, but also ­because ­human witnessing is already preemptively
entangled within the machine vision system.

witnessing autonomy

“If we disregard for a moment the fact that robotic intelligence w

­ ill prob­ably
not follow the anthropomorphic line of development prepared for it by sci-
ence fiction,” writes Manuel DeLanda in his 1991 book War in the Age of Intel-
ligent Machines, “we may without much difficulty imagine a f­ uture generation

Witnessing Vio­lence 73
of killer robots dedicated to understanding their historical origins.”89 Such a
robot historian, DeLanda speculates, would compose a very dif­fer­ent history
of their own emergence than a h ­ uman might, one far more concerned with
how machines shape ­human evolution ­toward their own autonomy than
with the agency of h ­ umans in assembling them. In the evolution of armies,
“it would see h ­ umans as no more than pieces of a larger military-­industrial
machine: a war machine.”90 Seeking to trace its own emergence, the historian
of a world of autonomous, weaponized robots would turn not to ­human
historical witnesses but to instances of machinic, signaletic, energetic, and
elemental witnessing registered in material rec­ords and relics, in the transfor-
mation of motors, fuel cells, transponders, mining equipment, the chemical
composition of geologic layers, atmospheres, and oceans. “Order emerges out
of chaos, the robot would notice, only at certain critical points in the flow of
­matter and energy,” and so the question for the robot historian might well be
how certain ­factors cohere within self-­organizing pro­cesses to tip them over
into evolutionary progression.91
Borrowing from Gilles Deleuze, DeLanda calls this autopoetic coherence
the “machinic phylum,” or the set of self-­organizing princi­ples and pro­cesses
that share deep mathematical similarities.92 For DeLanda’s putative robot
historian, the notion of a machinic phylum that blurs distinctions between
organic and inorganic life would be deeply appealing: it would suggest an in-
herent yet emergent coherence to the existence of “artificial” intelligence that
is not outside or alien to “nature.” Given how indebted computation is to war,
any account of how robot intelligence emerged would have to center military
technologies: “The moment autonomous weapons begin to select their own
targets, the moment the responsibility of establishing w ­ hether a h
­ uman is
friend or foe is given to the machine, we w ­ ill have crossed a threshold and
a new era ­will have begun for the machinic phylum.”93 In the three d ­ ecades
since DeLanda’s book, autonomous systems have proliferated, evolved, and
mutated in startling ways. In this chapter, I have shown how targeting tech-
nologies such as Agile Condor operate on the cusp of autonomy, producing
potential targets within a situation of ­imagined machinic precision. Yet ­there
are already autonomous weapons systems that significantly predate the new
typologies built on artificial neural networks and other predictive analytics.
Missile defense shields such as Israel’s Iron Dome operate on predefined
rules to knock out incoming attacks in response to sensor data. Packer and
Reeves point to aerial weapons systems “programmed with a range of po-
tential target criteria” that allows them to “slip between offensive and defen-
sive modes, loitering in an engagement zone u ­ ntil an appropriate target can

74 Chapter One
be discovered and automatically engaged.”94 Like all revolutions, then, the
seemingly sudden arrival of killer robots—­heralded by viral videos of danc-
ing Boston Dynamics humanoids and swarming slaughterbots—­has deeper
historical roots. Many of the most autonomous systems t­ oday are not found
on killer drones, but in huge guns mounted on naval vessels or on mobile
artillery platforms designed for surface-­to-­air defense.
While much of the history of early computing flowed from the labs of
darpa and other military agencies to the corporate world, rapid advance-
ment of machine intelligence now largely takes place at Google/Alphabet,
Amazon, Microsoft, Alibaba, Facebook/Meta, Palantir, and the countless
startups striving to join or be bought by the tech ­giants, or at university
labs, many underwritten by the tech industry.95 ai systems are built to be
transposable from one situation to another, such that machine vision and
navigation techniques developed for autonomous passenger vehicles can be
readily adapted to military contexts. With the infamous Predator already
mothballed and the Reaper slated to be decommissioned, remote warfare
is increasingly characterized by a far more diverse range of vehicles, plat-
forms, and systems. In the swift 2020 war between Armenia and Azerbaijan,
for example, the latter’s autonomous and semiautonomous drones proved
decisive, demonstrating the increasingly accessibility of ­these technologies
for military actors and signaling the capacity of homegrown automated sys-
tems to shift the calculus of war. In Ukraine’s ­resistance to the 2022 ­Russian
invasion, creative applications of consumer off-­the-­shelf drones augmented
the use of large-­scale weaponized drones and loitering munitions. At the
same time, an arms race for swarming drone technologies is underway, with
India trumpeting a field test of seventy-­five swarming drones in 2021 and
the darpa offset program showcasing mixed ground and aerial swarms
in 2019, stoking fears of a new genre of weapons of mass destruction. While
this diversification means that drones designed for an ever-­widening array
of mission types and milieus can be readily found, increasingly critical ques-
tions concern software systems, data collection and analy­sis, and the opera-
tive pro­cesses that enable identifying friends and foes, and targeting t­ hose
deemed threats. Like DeLanda’s robot historian, we are now confronted with
the prob­lem of tracing the emergence of such systems, but even more acutely
with the necessity of constructing the means to witness the autonomous vio­
lence they ­will—­and already do—­produce.
Reflecting on the necessity for research to understand war in ontological
terms, Caroline Holmqvist calls for greater attention to “what it means to
be a h­ uman being living the condition of war.”96 Without diminishing the

Witnessing Vio­lence 75
significance of this question, in the face of increasingly autonomous martial
systems and operations an inseparable concern is what it means to be non-
human in the condition of war. Or, to inflect this slightly differently, making
war sensible for h ­ umans means being able to ask how autonomous warfare
systems shape and are ­shaped by the world-­making and knowledge-­forming
interplay of ­humans and nonhumans alike. Like the Reaper or Agile Condor,
such systems are witnessing machines, but also what must be witnessed. I
want instead to ask how nonhuman witnessing invites an alternative ap-
proach to questions of ­human accountability, responsibility, and intelligibility
in the operation of autonomous war. But the challenge of pursuing martial
empiricism into the realm of emergent military technologies is that so much
remains in the virtual space of speculation and proposition. We can only seek
to move with the machinic turbulence of uncertain becomings that are still
very much in the p ­ rocess of (self-)­organizing into the autonomous, machinic
vio­lence of the f­ uture.
Within critical discussions of autonomous weapon systems, focus often
centers on the role of ­human actors within the system. As with so much
debate about ai more generally, prob­lems are framed around the account-
ability of systems to ­human oversight. In military parlance, this is typically
understood by the position of the h ­ uman in relation to the “loop” of decision
making that runs from sensing to targeting to firing. If a ­human is in-­the-­
loop, they have a deciding role on ­whether an action ­will be taken; on-­the-­loop
they have active oversight and the immediate capacity to intervene; off-­the-­
loop, the system runs autonomously without direct oversight. Prominent
critics of lethal autonomy, such as the roboticist Noel Sharkey, have proposed
more graduated categories for defining autonomy that center the agency of
­human actors, with the aim of delineating high degrees of autonomy that
should be prevented from being strapped to lethal weapons.97 But while t­ hese
are impor­tant distinctions that support the international ­legal push to ban
lethal autonomous weapons systems, they operate within a larger tendency
­toward the excision of the ­human from military systems. Military precision,
logistics, ­organization, and speed all depend on what Packer and Reeves call
“a preventive humanectomy” that promises to reduce friction and boost effi-
cacy by eliminating the weak point in data pro­cessing regimes.98 An ultimate
end of the militarization of violent mediation is thus the elimination of the
­human within technological systems to anything other than a potential target
for vio­lence. Within such systems, the capacity for the ­human to witness war
narrows to the sharp, brutal end of vio­lence, almost certainly launched from
a significant geo­graph­i­cal distance.

76 Chapter One
Witnessing this becoming-­target becomes impossible from within the hu-
manist frame, both b ­ ecause the ­human is excised and b­ ecause technoscien-
tific military systems, particularly ­those underpinned by complex algorithms
or artificial neural networks, are themselves inscrutable to ­humans. Prob­lems
of black-­boxed pro­cesses and partiality within knowledge production and
decision making are not unique to algorithms. Rather, as Amoore points
out, algorithms help “illuminate the already pre­sent prob­lem of locating a
clearsighted account in a knowable ­human subject.”99 Knowledge of both self
and other is always partial, yet ­these limitations of knowledge are buttressed
by culture, politics, ethics, and sociality. Witnessing functions to bridge this
lack, proffering a relationality grounded in the necessity of building shared
knowledges, ways of living, and forms of connection. Reflecting on the feed-
back loops, datafied ­human associations and actions, and back propagation
mechanisms of machine learning systems in both surgical robots and weapon-
ized drones, Amoore points out the “­human in the loop” is an elusive figure:
“The ­human with a definite article, the ­human, stands in for a more plural
and indefinite life, where h ­ umans who are already multiple generate emer-
gent effects in communion with algorithms.”100 Unlike the h ­ uman witness,
nonhuman witnessing transects t­ hese dynamics by refusing the distinctions
that underpin and separate out the h ­ uman and the machine. Against the
notion that a reasoning h ­ uman might provide both an ethical decision and
a witnessing account of autonomously executed vio­lence, nonhuman wit-
nessing insists on the incapacity of e­ ither h ­ uman or computer to account
for itself or the other. By starting with entangled relationalities, nonhuman
witnessing addresses violent mediation as an autonomous ­process that nev-
ertheless must be understood in relation to the ­human—­and the ­human must
be grasped in its complicity with and ­resistance to such violent mediations.
My claim is not that understanding certain machinic pro­cesses as nonhu-
man witnessing would magically “reveal” or “expose” something new about
­those pro­cesses. Rather, my contention is that the recognition of nonhuman
witnessing requires new critical understandings of the relations between ele­
ments within systems of autonomous vio­lence, and in d ­ oing so insists that we
resist an uncritical return to the figure of the autonomous liberal subject as
the antidote.101 If nonhuman witnessing takes place within autonomous mili-
tary systems through the registering of violent or potentially violent events by
sensors, their transformation into actionable data through machine vision,
and their determination as killable according to a computational matrix
of preemptive predictions, then the nonhuman witnessing of autonomous
military systems must reckon with the violent mediations of witnessing itself.

Witnessing Vio­lence 77
Within autonomous systems, ­those violent mediations are always directed
­toward the ­future. Or, rather, they depend on accumulated data from the past
to produce machinic predictions about the ­future.
Predictive analytics are thus always about the production of ­futures, or
the preemptive demarcation of certain virtuals as more or less on the verge
of becoming ­actual. “Threat is from the ­future,” writes Brian Massumi. “It is
what comes next. Its eventual location and ultimate extent are undefined.
Its nature is open-­ended. It is not that it is not: it is not in a way that is never
over.”102 This is the logic of preemption, where, as Andrejevic points out, “the
imminent threat becomes the lens through which a range of risks comes to
be viewed by t­ hose with the tools for responding to them.”103 Autonomous
military systems, ­whether weaponized or merely analytic, produce threat in
order to master it and in ­doing so collapse the ­future into the pre­sent through
the violent mediation of limitless potentiality into actionable probability.
Such systems are ontopowerful ­because they seek to intervene in becoming
itself, in the emergence of events from the temporal unfolding of existence
within time. While the claim of such systems is for security (of the state and
its citizens) and accuracy (in reducing the loss of life of t­ hose becoming-­
targets), this masks a necropo­liti­cal imperative: the automated determination
of death as a mechanism for the production of power. Lethal autonomous
weapons systems show how technoscientific necropolitics continually pushes
power to the edge of perception, which functionally merges with the limits of
[103.158.136.64] Project MUSE (2024-02-06 19:57 GMT) University of Edinburgh

operability. If the ultimate injunction of witnessing in war is to account for

the infliction of vio­lence, then witnessing automated killing must necessarily
entail the nonhuman.
In considering how vio­lence and perception are bound together in war,
Lucy Suchman poses the question: “Just what are the par­tic­u­lar apparatuses
of recognition that comprise con­temporary military discourses and tech-
nologies? How does the current ‘threat’ become recognizable, as specifically
situated persons, embodied and emplaced?”104 I would also ask, how is vio­
lence at work within the apparatus itself, in the p ­ rocess of making operative
images of persons, places, and animals? And what relations are forged, trans-
formed, or destroyed in the operation of the system? By pursuing the specific
pro­cesses of media technologies of increasingly autonomous war, I have
sought in this chapter to show how answering t­ hese questions depends upon
an openness to the transductive relations between h ­ uman and nonhuman,
organic and inorganic, technical and embodied. Just as the h ­ uman witness
might testify to what they have seen, however partial, and seek to render into

78 Chapter One
language the thickness of experience, however incompletely, so too might
nonhuman witnessing entail rendering sensible, however inadequately, the
violent mediations of datafication, preemption, and operationalism.
Once again, this pursuit of nonhuman witnessing returns us—­seemingly
inevitably—to the h ­ uman. In a fierce critique of the sociopo­liti­cal implica-
tions of algorithmic vio­lence, Peter Asaro writes that in an age of autonomous
weapons we need to ask: “What w ­ ill it mean to be ­human? What kind of
society ­will ­these systems be defending?”105 Questions of geopo­liti­cal power,
of regional and global balances and arms races, are not enough. Algorithmic
warfare leverages the globalized economy, infrastructures, and mobilities
that gird con­temporary technocapitalism, which means that t­ hese questions
of how we reckon with its knowledge machines and knowledge claims are
not solely the preserve of military strategists or critical theorists. The necro-­
and ontopolitics of algorithmic war and con­temporary state vio­lence share a
voracious need for embodied targets, h ­ uman or other­wise, and autonomous
war must be returned to questions of life in material and martial terms, as
well as conceptual ones. Bound up with this task is also an understanding of
the ­human and machinic l­ abor involved in such systems, a question which I
­will take up in the next chapter.
The point is not to grant the p ­ olitical subjectivity of the h­ uman witness
to algorithms or killer robots or semiautonomous drones, or to relegate the
­human from a central role in the witnessing of war. Recognizing the agency
of nonhuman entities does not equate to granting them citizenship, but non-
human witnessing aims to bring them into the space of p ­ olitical contestation
with their agency intact. Speculating on the f­ uture consequences of autono-
mous weapons for the status of the h ­ uman, Grove asks: “What w ­ ill a close
encounter with nonhuman intelligence do to force a ‘persisting us’ to rethink
the use to which we have put machines in the pursuit of what we ourselves
have been unwilling to do?”106 Another way to pose this question is to ask
what ethicopo­liti­cal status we might afford to self-­aware machinic encounters
with the world? How w ­ ill we think about the forms of knowledge they gener-
ate and the testimonies of unjust use they might compose? In returning to the
­human, then, the task at hand is to retain the nonhuman agencies, knowledges,
and relations excavated ­here, alongside an embodied, situated, and contingent
humanity. In the next chapter, I pursue this challenge in response to the
machine learning algorithms that are increasingly deployed as techniques
of power by states and corporations—­but that can also provide openings for
­resistance to ­those very institutions.

Witnessing Vio­lence 79
chapter
two

witnessing
algorithms
launched in august 2020, the latest edition of the venerable Microsoft
Flight Simulator video game series offered an open-­ended experience of a
world made suddenly inaccessible by covid-19. Unlike its p ­ redecessors, ms
Flight Simulator 2020 makes the entire planet its gameplay environment.
In the hyperbole typical of much of the media coverage, New York Times
tech columnist Farhad Manjoo proclaimed that Microsoft had “created a
virtual repre­sen­ta­tion of Earth so realistic that nearly all sense of abstraction
falls away.”1 As a technical achievement, Flight Simulator is certainly impres-
sive. Combining data from OpenStreetMaps and Bing Maps via the Azure
artificial intelligence cloud computing platform, Microsoft created an algo-
rithmic system to assign and render photorealistic 3d imagery of skyscrapers,
homes, trees, oceans, mountains, and so on. This imaging of the world is
not, however, photographic but datalogical: generated algorithmically by a
machine learning model fed vast amounts of map, satellite, photogrammetric,
and other data. It is a machinic imagining of the textures of the world. Like
Google Earth, it is a datalogical attempt at solving the fundamental prob­lem
that plagues the unusable map from Borges’s short story “On Exactitude in
Science,” which in the effort to precisely represent ­every detail of an empire
grows to the same size as the territory. Rather than indexing its cartography
to the world perceived by h ­ uman mapmakers, Flight Simulator generates
what its algorithms believe the world to be. Players quickly found numer-
ous strange glitches: a corporate office tower in place of the Washington
Monument, a mashup of vegetation and buildings in the Norwegian town
of Bergen, obelisks in place of palm trees. Far from a utopian rendering of a
world made beautiful yet knowable, Kyle Chayka writes in Slate that Flight
Simulator reminds us that “an automated, unchecked p ­ rocess is warping the
(virtual) world around us, leading to ­these weird errors and aberrations.”2
Even as Flight Simulator seemed to offer a new algorithmic means of witness-
ing in won­der at the world, its glitches remind us of the necessity of witnessing
­those same algorithmic systems. If algorithms are themselves witnessing, mak-
ing knowledge, and forging worlds of their own design, what might it mean to
witness their workings?
The world-­making capacity of the algorithm is not readily apparent in its
more common definitions: a step-­by-­step instruction of how to solve a task;
a ­recipe; a form of programmed logic; an automated filtering mechanism.
­These commonplace accounts fail to get to the heart of t­ hings, the operative
pro­cessing made pos­si­ble by the “if . . . then” procedure of the algorithm and
its potentially harmful outcomes.3 In princi­ple, algorithms are abstract pro­
cesses, which means they are not dependent on a specific computer language
for their validity. But in practice, algorithms are typically encoded in distinct
computer languages and ecosystems. More than this, though, they are also
inescapably codes in the sense that they unlock certain translations, opera-
tions, or transformations of data.4 We might even think of them as magic
in the sense that the incantation of the algorithm by the software within
which it is packaged enables action to be performed. Like codes and magic,
algorithms conceal their own operations: they remain mysterious, including
to their makers. This inscrutability is particularly the case with the machine
learning algorithms that have become the principal means by which power
is now enacted, maintained, and reproduced in the digital domain.
Machine learning is a technique for the statistical analy­sis of huge quanti-
ties of data. A machine learner is an algorithmic system in which computer
code learns from data to produce a model that can be deployed on more data.
Machine learning produces models by using algorithmic techniques to look
for patterns in huge amounts of data, then applying ­those patterns to the
data to become increasingly discerning: able to recognize, differentiate, and
discriminate between ele­ments within the database. Machine learning pow-
ers every­thing from inbox filtering to Netflix recommendations and it feeds
on the data produced through our interactions with t­ hose systems. Machine

Witnessing Algorithms 81
learning systems and the companies that promote them almost always seek
to obscure both the “­free l­ abor” of user interactions and the low-­paid l­ abor of
digital pieceworkers on platforms such as Mechanical Turk in an effort to sell
the technical prowess of their “ai” inventions. Machine learning uses layers
of neural networks—­arrays of computational nodes that work collaboratively
to build relations between bits of data—to make predictions about the data.
With OpenAI’s ChatGPT, this manifests as the statistical production of text
based on what the model anticipates to be the desired answer to a query.
In military operations, it might mean identifying and prioritizing distinct
threats in a crowded conflict zone. Rather than following a defined sequence
of steps, machine learning models act recursively to build relational functions
that can be applied ever-­more accurately and efficiently, provided the learner
is trained and optimized appropriately.5
But this technicity is not purely technical. As Adrian Mackenzie points
out, ­there are no machine learning systems without h ­ uman coders and
­humans are also needed to tag objects in the datasets for the supervised
training by which many machines learn.6 In so-­called unsupervised learning,
algorithms develop their own data tags, but h ­ uman effort is still constantly
required to tweak, select, optimize, and monitor training. Jathan Sadowski
calls this “Potemkin ai,” or artificial intelligence that is actually only thinly
computational and largely driven by h ­ uman ­labor.7 On top of the obscured
­human l­ abor, Sy Taffel shows how computational systems also elide massive
ecological costs of powering and cooling data centers, not to mention mining
rare and common metals or shipping equipment across the globe.8 To bring
machine learning into the language of this book, its models and algorithms
are not alien, purely technical agents wholly separate from the h ­ uman, but
rather enmeshed with the h ­ uman and with the more-­than-­human world.
How machine learners make knowledge ­matters ­because they are increas-
ingly pivotal to con­temporary finance, logistics, science, governance, national
security, and culture, yet they remain hard to scrutinize, building blocks in
what Frank Pasquale calls the “black box society.”9
Despite their technical veneer, algorithms are ­shaped and bound by as-
sumptions and values about the world, drawn from the datasets upon which
they are built, the biases of their architects, and the instrumental objectives
of the institutions that use them. Th ­ ese assumptions and values might be
as straightforward as ­whether to order library books by alphabet or cata­log
number, or as outrageously discriminatory as Facebook allowing housing
advertisers to exclude users from target audiences using zip codes and other
proxies for race, class, and religion. Given the colonial entanglements of

82 Chapter Two
modern science, regimes of classification, and the statistical techniques that
underpin con­temporary data science and machine learning, the constitutive
vio­lence of many such systems should come as no surprise. Algorithmic
vio­lence, ­whether in the form of digital redlining or autonomous weap-
ons, is an ethicopo­liti­cal prob­lem much more than a technical one.10 As Ed
Finn points out, the algorithm is a crucial site of critical inquiry ­because it
is “the object at the intersection of computational space, cultural systems
and ­human cognition.”11 Traceable back to the cybernetic era of computa-
tional research that followed World War II, algorithms w ­ ere at the center of
a radical transformation that substituted rationality for reason. Within two
­decades of the war, as Orit Halpern argues, “the centrality of reason as a tool
to model h ­ uman be­hav­ior, subjectivity, and society had been replaced with
a new set of discourses and methods that made ‘algorithm’ and ‘love’ speak-
able in the same sentence and that explic­itly correlated psychotic perspective
with analytic logic.”12
Now deployed across almost ­every field of ­human endeavor and inquiry,
algorithms bridge the gap between computation, culture, and thought—­but
they are not reducible to any of t­ hose domains. According to Taina Bucher,
algorithms are “entangled, multiple, and eventful and, therefore, ­things that
cannot be understood as being power­ful in one way only.”13 Consequently,
“algorithmic systems embody an ensemble of strategies, where power is im-
manent in the field of action and situation in question.”14 Research by Safiya
Noble and o ­ thers into the oppressive biases of Google and Facebook shows
how supposedly objective systems are inseparable from racism, sexism, and
other socially produced and reproduced structures of domination.15 Gen-
erative ai tools such as Dall-­E 2 or Midjourney are no exception, evidenced
by the efforts of their architects to engineer user inquiries rather than resolve
the impossible prob­lem of the under­lying data.16 As Nick Seaver argues,
“algorithms are not singular technical objects that enter into many dif­fer­ent
cultural interactions, but are rather unstable objects, culturally enacted by the
practices ­people use to engage with them.”17 Much like a poem, algorithms
are tricky objects to know and often cannot even reveal their own workings.18
Critical research thus attends less to what an algorithm is and more to what
it does.19
In pursuing nonhuman witnessing of, by, and through algorithms, my
focus is on their operative, extractive, and generative qualities, rather than
their computational mechanics. Through a series of investigations into dis-
tinct machine learning systems, I argue that algorithms can engage in a
perceptual ­process that constitutes nonhuman witnessing, elevating mere

Witnessing Algorithms 83
observation into an ethicopo­liti­cal plane. In drone warfare, an algorithm
might “see” certain activity, “decide” it is threatening, and “recommend” the
prosecution of vio­lence. My contention is that such algorithmic registering
and translating of worldly phenomena constitutes witnessing ­because it does
so to violent ends and caries the most immediate traces of that vio­lence. Fa-
cial recognition software is a tool for producing evidence through machinic
witnessing, yet both the data that feeds such systems and the unknowable
neural dynamics that animate them make it so dangerous that facial recog-
nition has been described as akin to plutonium.20 Algorithmic witnessing,
then, often takes place through the enactment of vio­lence, with the algo-
rithm as both witness and perpetrator. At the same time, such algorithms
are themselves entities that must be witnessed—­yet by their entangled nature
they resist being broken into consistent ele­ments that can then be rendered
knowable.
This chapter grapples with the competing dynamics of the doubled mean-
ing of its title: algorithms that do witnessing and the witnessing of algorithms
(and what they do). Or, to put this differently, this chapter asks both how
algorithms might be agents of witnessing and how algorithms might be wit-
nessed? Rather than look for machinic relations to events that might be
analogous to ­human witnessing, this chapter seeks out intensive sites within
human-­nonhuman assemblages where machinic affect—­technical yet con-
tingent, potential rather than predetermined—­enables forms of encounter
that generate a relation of responsibility between event and algorithm. ­Doing
so requires the bracketing of any ethical imperative to witnessing: algorithmic
witnessing can only ever be grasped within the milieu of the algorithm, an
agency that can only be ascribed ethics or morality through anthropomor-
phic gymnastics. Delving into the machinic affects of witnessing algorithms
­will require us to depart further still from the narrow humanistic conception
of witnessing and to insist on the separation of witnessing from testimony.
If algorithmic technologies are now crucial knowledge machines, yoked
to power, and the infliction of state vio­lence, then asking how witnessing
reckons with them and takes place through them requires attending to how
computational pro­cesses generate their machinic relations, and how t­ hose
relations sustain the power of ­those systems.
Even as the image increasingly overwhelms the word as the dominant
form of communication, the expansion of technologies that identify and
­organize images means that a new form of aggregated, relational perception
is taking hold. Writing on the aggregation of huge numbers of images into
datasets analyzed by machine learning systems, Adrian Mackenzie and Anna

84 Chapter Two
Munster understand ­these relational pro­cesses as “generative technical forces
of experience.”21 They propose the concept of “platform seeing” to describe
an operative mode of perception “produced through the distributed events
and technocultural pro­cesses performed by, on and as image collections are
engaged by deep learning assemblages.”22 In their account, “seeing” is not the
act of a singular entity but rather something that takes place across a g­ reat
many ­human, material, and computational agents. Images become subject
to a host of functions: precisely formatted for input to models; labeled, pro­
cessed, and used to configure small neural networks onboard smartphones;
moved from the devices of consumers to platforms and their data centers and
back. Through t­ hese and other functions, images transformed from ­bearers
of indexical relations to ele­ments within operational (image) collections.23
Consequently, the relations between images within the dataset, including the
relations of ele­ments within images to ele­ments within ­others, become more
impor­tant than the images themselves.
Platform seeing is thus the “making operative of the visual by platforms
themselves.”24 This invisual mode of perception takes place outside the do-
main of repre­sen­ta­tion: images no longer take their meaning from ­things in
the world but rather in relation to the ele­ments and edges of other images.
Crucially, this “operativity cannot be seen by an observing ‘subject’ but rather
is enacted via observation events distributed throughout and across devices,
hardware, ­human agents and artificial networked architectures such as deep
learning networks.”25 Despite the absence of a h ­ uman subject, t­ hese pro­
cesses still constitute something called “seeing” precisely b
­ ecause they remain
within the perceptual domain of recognizing and differentiating images. In
this chapter, I make a parallel argument about witnessing: that even with-
out a witnessing “subject” in the unitary humanist sense, witnessing occurs
within and through algorithmic systems. Such witnessing necessarily exists
on a continuum with perceiving and cannot be neatly distinguished from
it. Dif­fer­ent contexts, media technics, and h ­ uman entanglements produce
distinct intensive fields of relation that shift perceiving into the modality of
witnessing. Not all ­human perception entails witnessing, and so neither does
all perception by the nonhuman agencies of algorithms.
While witnessing rarely figures in discussions of algorithms and artificial
intelligence, terms that appear in witnessing discourses abound: truth, rec-
ognition, memory, transparency, ethics. This is not to suggest an inherent
synchronicity between witnessing and the algorithmic, but rather to point
out that the perception required for both to operate possesses a purposive
dimension. As Amoore writes, “A defining ethical prob­lem of the algorithm

Witnessing Algorithms 85
concerns not primarily the power to see, to collect, or to survey a vast data
landscape, but the power to perceive and distill something for action.”26 In
much the same way, witnessing is not reducible to seeing, but is a perceptual
encounter that produces an injunction to action through its configuring
of a par­tic­u­lar scene and its coalescing of that scene’s relational dynamics.
Like algorithms, witnessing makes truth claims about the world as well,
and is also prone to oversight, misapprehension, and misstatement.27 Like
algorithms, witnessing is prone to falsity, ­whether deliberate or accidental.
Their distributed, multiple, contingent, and operative existence means that
algorithms cannot be known or accounted, and yet neither can the ­human.
It is only ever ­humans, plural, who can give account, and ­doing so is always
incomplete. This is why, for Amoore, “algorithms do not bring new prob­lems
of blackboxed opacity and partiality, but they illuminate the already pre­sent
prob­lem of locating a clearsighted account in a knowable ­human subject.”28
Neither ­human nor algorithm can give an account of itself that is complete
or transparent. An ethics for algorithmic worlds cannot “seek the grounds
of a unified I ” but must instead “dwell uncertainly with the difficulty of a
distributed and composite form of being.”29 This chapter pursues the ques-
tion of what distributed, opaque, and decentered witnessing might look like
within technics of the algorithm—­and how such a contingent and multiple
domain might itself be witnessed.
Crucial to that task is tracing what I call machinic affect, or the intensive
relations that bind technical systems to one another and ­humans to technical
systems. By machinic affect, I mean the capacity to affect and be affected that
occurs within, through, and in contact with nonhuman technics. In keeping
with Félix Guattari’s expansive conception, my own use of “machinic” is not
restricted to the mechanistic but rather refers to the pro­cessual assemblage
of ele­ments, objects, concepts, imaginaries, materialities, and so on that
form “machines” through their distinct yet transversal relations. Guattari’s
machines are organic and inorganic, technical and social, material and ab-
stract.30 Machinic affect is not so much indifferent to the flesh as it is promis-
cuous in its adhesive and intensifying properties, such that the corporeality
of the ­human does not default to center stage.31 Excavating machinic affect
from technical assemblages requires attending to the distinctiveness of indi-
vidual technical objects as they assem­ble, attenuate, modulate, amplify, and
terminate technical and nontechnical relations. In the context of witnessing,
machinic affect can be applied to understand the relations forged between
witness and event when mediated through screens. But more importantly

86 Chapter Two
and generatively, machinic affect offers an analytic for making vis­i­ble the
other­wise obscured machinic relations of complex technical systems and
especially learning algorithms.
Machinic affect describes the dynamic intensities of technical systems. As
such, machinic affect is autonomous intensity: owned neither by one body
nor another, but constituting and constituted by them, ­whether ­human or
non. Pursuing machinic affect within the media-­specificities of algorithmic
systems, I am interested in how the pro­cessual empiricism of what Massumi
calls the “virtual” illuminates the relational dynamics of machine learning.
For Massumi, the virtual describes the immanence of potentiality, its passage
from futurity through experience and into pastness. The virtual is what might
arise and what might have been. It is not opposed to the ­actual, but its under-
side. Affect is “precisely this two-­sidedness, the simultaneous participation of
the virtual in the a­ ctual and the a­ ctual in the virtual, as one arises from and
returns to the other.”32 This chapter is about the necessity of witnessing how
algorithms, particularly machine learning ones, oscillate between actualizing
the virtual and virtualizing the ­actual.
If algorithmic systems are about taming potential into probability in the
name of emergent ordering of worldly phenomena, we can understand them
in Massumi’s terms as ontopowerful: as technological pro­cesses for the mas-
tery of becoming.33 Machine learning systems are constituted by unreason—­
even madness—­through looping recursivities.34 This nonlinearity, too, finds
much in common with Massumi’s recognition that “intensity would seem
to be associated with nonlinear pro­cesses: resonation and feedback that
momentarily suspend the linear pro­gress of the narrative pre­sent from past
to ­future.”35 As well as disassembling and distributing the subject, witnessing
algorithms requires dismantling and dispersing the event in time as it is taken
up and worked upon by algorithmic agencies. This chapter thus excavates
the distinctive dynamics of nonhuman witnessing across four instances of
algorithmic world-­making: the false witnessing of deepfakes; the animat-
ing of evidence in Forensic Architecture’s machine learning investigations;
military imaginings of archival and real-­time pro­cessing of full motion video
imagery from loitering drones; and the witnessing of machine learning pro­
cesses in aesthetic interventions into algorithmic systems. Operating with
dif­fer­ent learning models and data sources and within very varied contexts,
­these examples show the dangers of algorithmic witnessing and the necessity
of witnessing algorithms, but they also suggest the potential of such systems
to work against state and corporate vio­lence.

Witnessing Algorithms 87
bearing false testimony: deepfakes

The synthetic media that would become known as “deepfakes” first surfaced
to mainstream attention with a December 2017 article by Samantha Cole in
Vice Media’s tech site Motherboard about a pornographic video that appeared
to feature the actress Gal Gadot having sex with her stepbrother (figure 2.1).
As Cole reported on the tech site, the video was a fake, the clever but decid-
edly imperfect creation of a Reddit user with some basic machine learning
skills and open-­access tools downloaded from the code repository GitHub.36
Fake and face-­swapped pornography are not new phenomena: cgi porn
is widely available, while photoshopped porn images have been around as
long as the internet and altered nude photo­graphs since the early twentieth
­century at least. The difference in the Gadot video was the application of
deep learning techniques to automatically swap one face with another. That
technique gave the Redditor his h ­ andle and the new genre a name: deepfakes.
“With hundreds of face images, I can easily generate millions of distorted
images to train the network,” deepfakes told Cole. “­After that if I feed the
network someone ­else’s face, the network ­will think it’s just another distorted
image and try to make it look like the training face.”37 With so many high-­
quality images on which to train the system available online, celebrities like
Gadot are easy targets. But that same ease could readily apply to politicians,
and to voice as well as video. Arriving amid a rising tide of distrust in systems
and institutions, deepfakes seemed to herald a new threat, undermining
demo­cratic pro­cesses and cybersecurity and facilitating misinformation and
revenge porn. A cottage industry of deepfake creation and detection sprung
up in response. Deepfakes seemed to enable the bearing of algorithmic false
witness—­a prob­lem only complicated by the arrival of more user-friendly
artificial image and video generation tools in the years since.
While t­ here are several techniques that can be used to generate deepfakes,
the most effective are produced through a form of deep learning neural net-
work called “generative adversarial networks,” or gans. While image recog-
nition algorithms are typically trained using convolutional neural networks
(cnns) that slide filters across images to learn their spatial properties, gans
work by pitting two algorithms against each other in a game of true and
false (figure 2.2). First proposed by ai researchers from Google Brain in a
2014 paper, the premise of gans is ­simple enough: one neural network (the
generator) learns to create images that it then feeds to another network (the
discriminator), which decides if the image is “fake” or “real” compared to its
own training dataset.38 ­Those results are then fed back into the generator, so

88 Chapter Two
figure 2.1. Still from Gal Gadot deepfake porn

that it can learn from the assessment of the discriminator. What makes the
technique power­ful is that both networks are learning at the same time, with
the discriminator learning just enough to get ahead of the generator each
time the quality of its fakes catches up. To produce the Gal Gadot deepfake
with a gan, the generator would be fed the pornographic video while the
discriminator learned from real photos of Gadot. As the generator modified
its video using several image-­blurring and blending techniques, proxim-
ity to what the discriminator was learning about Gadot would yield better

figure 2.2. Diagram of general adversarial network structure

Witnessing Algorithms 89
and better results ­until the discriminator could no longer identify the fake
images as fake at all. If the gan was then trained on other video and image
sets, it would get even better at its task. In this way, gans can become highly
accomplished at swapping any face for any other. Versions of this technique
can be applied to specific parts of the face, too, such as the lips, or to audio,
enabling the falsification of someone’s voice to match a script, as in the widely
reported Obama lip-­sync demonstration.39
While computer science papers have focused on deepfake creation and
detection, the humanities and social sciences has begun to address a wider
set of questions.40 The most attentive examinations of deepfakes have oc-
curred within porn studies, where the gendered nature of the technology in
practice—­more than 99 ­percent of documented deepfakes feature ­women
face swapped into pornographic videos—­has been documented and exam-
ined in a range of contexts, from revenge porn to communities of practice
to the emergence of “designer” porn.41 ­Legal scholarship within the United
States has addressed how deepfakes create tension between rights to ­free speech
and privacy, as well as how they pre­sent a potential crisis for the verifi-
cation of evidence presented to court.42 Pos­si­ble impacts for cybersecu-
rity and information warfare are articulated in more apocalyptic terms.43
But deepfakes also point to the vibrancy of everyday data cultures, and
the experimental, open-­source approach to ai and automation literacies
taking place on GitHub, YouTube, and Reddit.44 “Deepfakes are complex
epistemic t­ hings,” observes Rebecca Uliasz, which “testify to the ongoing
socio-­technical value we place on visual accuracy which manifests in our
continued investment in imagistic realism as truthful.”45 As such, deepfakes
can be understood as yet another technological blow to shared epistemic
frameworks, further undermining certainty in image authenticity for both
journalists and publics.46
For witness, a New York–­based ­human rights ­organization that equips
citizens and activists with video tools and resources, deepfakes and related
forms of synthetic media are an urgent danger b ­ ecause they can amplify or
microtarget the kinds of media disinformation and incitement that spark
massacres, assaults, and ­political instability. In a report on synthetic media,
witness distills dozens of scenarios into five key prob­lem areas: real­ity edits,
credible doppelgängers, news remixing, floods of falsehood, and plausible
deniability, in which claims of deepfaking allow bad faith actors to deny
having said or done what a video shows.47 Deepfakes lead to two interrelated
epistemic challenges: “the inability to show that something real is real, and

90 Chapter Two
then the ability to fake something as if it was real.”48 For witness, the in-
ability to prove that something real is real pre­sents the more serious dilemma
­because it suggests an existential peril for the mediated pro­cesses that are so
essential to con­temporary shared realities. This algorithmic false witness risks
placing all mediated witnessing into question. Deepfake false witnessing cuts
and intensifies preexisting risks to individuals and communities by catalyzing
uncertainties within con­temporary media ecologies. Over the last few years,
witness has worked with Partnership on ai to develop guidelines for appro-
priate use of synthetic media in ­human rights contexts. ­These are necessary
and impor­tant practical steps, but the epistemic challenges of deepfakes and
related media forms persist.
By threatening the legitimacy of the image, deepfakes destabilize the very
foundations of media witnessing as a shared means of producing an agreed
real­ity. Deepfakes emerged in a media witnessing ecol­ogy in which power has
shifted from the authority of legacy media to the immediacy of smartphone
and other user-­generated content.49 As the necessity of grounding truth
claims becomes more urgent, deepfakes heighten the fallibility of witnessing
in, by, and through media. Th ­ ese are fake images that make truth claims,
even as they undermine the possibility of common epistemic ground.50 In
places with declining trust in government or with significant government
instability and insecurity, deepfakes have the potential to incite vio­lence and
violate ­human rights. Weaponized deepfakes assembled on the fly from so-
cial media rec­ords are one nasty possibility for the ­future of what Tom Sear
calls “xenowar.”51 If neither still nor moving images can be trusted to bear
the indexical relationship to the world that their authority depends upon,
the potential for any mediated witness to be false threatens to pry open the
fractures already r­ unning through any sense of shared real­ity. With their
emphasis on altering or swapping ­faces, deepfakes are affect machines even
more than cognitive deceptions. Machinic affect ­here takes a very recogniz-
able form in the micromovements of faciality described by Deleuze in his
account of the affection image in cinema and by Silvan Tomkins in his theory
of nine discrete relational affects manifested on the face.52 Face, voice, and
gesture are among the most crucial embodied qualities of bearing witness:
deepfakes seek to synthesize both fake and real to affect the viewer. Created
through the intensive interplay of machinic relations, deepfakes are also af-
fect engines when loose in the wild. As false witnessing algorithms, deepfakes
exemplify the inextricability of h ­ uman and non in witnessing assemblages
within t­ oday’s deeply computational world.

Witnessing Algorithms 91
Deepfakes are among the most unsettling instances of the shifting re-
lationship between image and data. “An image that is computational is no
longer strictly concerned with mimesis, nor even with signification,” writes
Steven F. Anderson. “Computational images may serve as interfaces, carriers,
or surface renderings, the real importance of which are their under­lying pro­
cesses or data structures.”53 Deepfakes are syntheses of recorded and gener-
ated images made pos­si­ble by the encoded nature of both. At the level of code
itself, neither bears any more material relation to the world beyond computa-
tion than the other. Even as the image reaches its zenith in visual culture, the
transfiguration into code that made its domination pos­si­ble contains within
it the collapse of the authority granted to the image by its seemingly indexi-
cal relation to the world. Ironically enough given their origins in diy porn
communities, deepfakes speak to how “the once voy­eur­is­tic gaze of cinema
has given way to power relations articulated through computational systems
rather than through ocular regimes predicated on reflected light and bodies
in space.”54 The false witnessing of deepfakes suggests that contestations over
the meaning of images is moving away from signification and into genera-
tion. For deepfakes and imagery produced by Stable Diffusion or Dall-­E 2,
contestation ceases to be about what the video image means and comes to be
about the ­process of its generation.
This movement from semiosis to p ­ rocess means that the false witnessing
of deepfakes must be contested at an ontological level. While early iterations
had a tendency for glitching and an unsettling uncanny valley-­like quality,
advances in the deep learning pro­cesses of gans now mean that h ­ umans
can typically detect deepfakes about half the time, or at the same rate as ran-
dom chance. Deepfake detection tools that draw on the same kind of deep
learning neural networks have become increasingly impor­tant, but their
emergence and growing accuracy has led to an arms race between forgers
and detectors.55 This formation of a new adversarial, nonhuman, and ma-
chinic relationship between witness and interrogator points to yet another
site in which critical debates about culture, politics, ethics, and knowledge
play out without the ­human in the driver’s seat. A potentially endless game
of deception and unmasking awaits in which witnessing itself becomes the
ground of contestation between adversarial machine learning systems and
where social and ­political life become the field upon which the consequences
of that strug­gle play out. Ironically enough, algorithmic false witnessing
heightens computation’s claim as both figure and ground for how knowledge
is produced and contested.

92 Chapter Two
between evidence and witnessing: forensic
architecture and open-­s ource machine learning

Synthetic media are everywhere, not just in deepfakes. Digital images and
objects that appear to index something in the world but do nothing of the
sort have their roots in video games and online worlds like Second Life.
With the growing appetite for niche machine learning training sets and ar-
tificial environments for testing autonomous machines, synthetic media are
increasingly central to the development of algorithmic systems that make
meaningful decisions or undertake actions in physical environments. Syn-
thetic media are swift to produce and can be tagged as part of the production
­process, which reduces costs, delays, and inaccuracies from using ­people to
tag images or other data.
Microsoft AirSim is a prime example, an environment created in Epic’s
Unreal Engine that can be used to test autonomous vehicles, drones, and
other devices that depend on computer vision for navigation.56 Artificial
environments are useful testing grounds b ­ ecause they are so precisely ma-
nipulable: trees can be bent to a specific wind f­ actor, light adjusted, surface
­resistance altered. They are also faster and cheaper places to test and refine
navigation software prior to expensive material prototyping and real-­world
testing. In machine learning, building synthetic training sets is now an es-
tablished practice, particularly in instances of ­limited data availability or lack
of data diversity. For example, the com­pany Synthesis.ai produces synthetic
images of nonwhite ­people to train vari­ous kinds of recognition algorithms.
Synthetic media are valuable in contexts such as armed conflict, where im-
ages might be too few to produce a large enough corpus and too classified
to be released to e­ ither digital pieceworkers for tagging or private sector
developers to train algorithms.
But what happens when synthetic media are marshaled to do the activist
work of witnessing state and corporate vio­lence? What are we to make of
the proposition that truths about the world might be produced via algo-
rithms trained almost exclusively on synthetic data? This section sketches
answers to t­ hese questions through an engagement with ­Triple Chaser, an
investigative aesthetic proj­ect from the UK-­based research agency Forensic
Architecture. Founded in 2010 by architect and academic Eyal Weizman and
located at Goldsmiths, University of London, Forensic Architecture invents
investigative techniques using spatial, architectural, and situated methods.
Using aesthetic practice to produce actionable forensic evidence, their work

Witnessing Algorithms 93
appears in galleries, courtrooms, and communities. In recent years, they
have begun to use machine learning and synthetic media to overcome a
lack of publicly available images on which to train their machine learning
models and to multiply by several ­orders of magnitude the effectiveness of
images collected by activists. In contrast to the false witnessing of deepfakes,
­these techniques show how algorithms can do the work of a more resistant
and generative witnessing, translated into open-­source tools for activists via
well-­documented GitHub pages.
Presented at the 2019 Whitney Biennial in New York, ­Triple Chaser com-
bines photographic images and video with synthetic media to develop a
dataset for a deep learning neural network able to recognize tear gas canisters
used against civilians around the world. It responds to the controversy that
engulfed the biennial following revelations that tear gas manufactured by Sa-
fariland, a com­pany owned by Whitney trustee Warren B. Kanders, was used
against protestors at the US-­Mexican border. Public demonstrations and artist
protests erupted, leading to significant negative press coverage across 2018 and
2019. Rather than withdraw, Forensic Architecture submitted an investigative
piece that sought to demonstrate the potential for machine learning to func-
tion as an activist tool. Produced in concert with artist and filmmaker Laura
Poitras, ­Triple Chaser was presented as an eleven-­minute video installation.
Framed by a placard explaining the controversy and Forensic Architecture’s
decision to remain in the exhibition, viewers entered a severe, dark room to
watch a tightly focused account of Safariland, the prob­lem of identifying tear
gas manufacturers, the technical pro­cesses employed by the research agency,
and its further applications. Despite initial intransigence, the withdrawal of
eight artists in July 2019 pushed Kanders to resign as vice chairman of the
museum and, l­ater, announce that Safariland would sell off its chemicals
division that produces tear gas and other antidissent weapons. Meanwhile,
Forensic Architecture began to make its codes and image sets available for
open-­source download and began applying the same techniques to other
cases, uploading its Mtriage tool and Model Zoo synthetic media database to
the code repository GitHub. A truth-­seeking tool trained on synthetic data,
­Triple Chaser reveals how machinic affects oscillate between witnessing and
evidence.
In keeping with the established ethos of Forensic Architecture, ­Triple
Chaser demonstrates how forensics—­a practice heavi­ly associated with both
policing and the law—­can be turned against the very state agencies that
typically deploy its gaze. As Pugliese points out, “Embedded in the concept
of forensic is a combination of rhetorical, performative, and narratological

94 Chapter Two
techniques” that can be deployed outside courts of law.57 For Weizman, the
fora of forensics is critical: it brings evidence into the domain of contestation
in which politics happens. In his agency’s counterforensic investigation into
Safariland, tear gas deployed by police and security agencies becomes the
subject of interrogation and re-­presentation to the public.58 In this making
public, distinctions and overlaps can be traced between dif­fer­ent modes of
knowledge-­making and address: the production of evidence, the speaking
of testimony, and the witnessing of the audience. But how might we under-
stand the role of the machine learning algorithm itself? And how are we to
conceptualize this synthetic evidence?
Weizman describes the practice of forensic architecture as composing
“evidence assemblages” from “dif­fer­ent structures, infrastructures, objects,
environments, actors and incidents.”59 ­There is an inherent tension between
testimony and evidence that counterforensics as a resistant and activist
practice seeks to harness by making the material speak in its own terms. As
method, forensic architecture seeks a kind of “synthesis between testimony
and evidence” that takes up the lessons of the forensic turn in ­human rights
investigations to see testimony itself as a material practice as well as a lin-
guistic one.60 Barely detectable traces of vio­lence can be marshaled through
the forensic ­process to become material witnesses, or evidentiary entities. But
evidence cannot speak for itself: it depends on the h ­ uman witness. Evidence
and testimony are closely linked notions, not least b ­ ecause both demarcate an
object: speech spoken, m ­ atter marked. Testimony can, of course, be entered
into evidence. But something more fundamental is at work in ­Triple Chaser.
Its machine learning model ­doesn’t simply register or represent. It is opera-
tive, generating relations between objects in the world and the ­parameters
of its data. Its technical assemblage precedes both evidence and testimony.
It engages in nonhuman witnessing. ­Triple Chaser brings the registering of
violations of h­ uman rights into an agential domain in which the work of wit-
nessing is necessarily inseparable from the nonhuman, ­whether in the form
of code, data, or computation.
As development commenced, ­Triple Chaser faced a challenge: Forensic
Architecture was only able to source a small percentage of the thousands
of images needed to train a machine learning algorithm to recognize the
tear gas canister produced by Safariland. They ­were, however, able to source
detailed video footage of depleted canisters from activists and even obtained
some material fragments. Borrowing from strategies used by Microsoft,
Nvidia, and ­others, this video data could be modeled in environments built in
the Unreal gaming engine, and then scripted to output thousands of canister

Witnessing Algorithms 95
images against backgrounds ranging from abstract patterns to simulated
real-­world contexts (figure 2.3). Tagging of t­ hese natively digital objects also
sidestepped the l­ abor and error of manual tagging, allowing a training set to
be swiftly built from images created with their metadata attached (figure 2.4).
Using several dif­fer­ent machine learning techniques (including transfer
learning, combining synthetic and real images, and reverse discriminators),
investigators ­were able to train a neural network to identify Safariland tear
gas canisters from a partial image with a high degree of accuracy and with
weighted probabilities. ­These synthetic evidence assemblages then taught the
algorithm to witness.
Like most image recognition systems, ­Triple Chaser deploys a convolu-
tional neural network, or cnn, which learns how to spatially analyze the
pixels of an image. Trained on tagged datasets, cnns slide—­convolve—­a
series of filters across the surface of an image to produce activation maps that
allow the algorithm to iteratively learn about the spatial arrangements of pix-
els, which can be repeated across large sets of images. ­These activation maps
are passed from one convolution layer to the next, with vari­ous techniques
applied to increase accuracy and prevent the spatial scale of the system from
growing out of control. Exactly what happens within each convolutional
layer remains in the algorithmic unknown: it cannot be distilled into repre­
sen­ta­tional form but rather eludes cognition.61 Machine learning pro­cesses
thus exhibit a kind of autonomic, affective capacity to form relations between
objects and build schemas for action from the modulation and mapping of
­those relations: machinic affect. Relations between ele­ments vary in intensity,
with the p­ rocess of learning both producing and identifying intensities that
are autonomous from the ele­ments themselves. It is precisely this that cannot
be “visualized” or “cognized.” Intensive relations assem­ble ele­ments into new
aggregations; bodies affect and are affected by other bodies. Amoore writes
that algorithms must be understood as “entities whose par­tic­u­lar form of
experimental and adventurous rationality incorporates unreason in an in-
tractable and productive knot.”62 Reflecting on economic self-­interest and the
false grounds of rational choice, Massumi points out that “rationalities are ap-
paratuses of capture of affectivity.”63 Machine learning works in concomitant
ways. Th
­ ere is an autonomic quality to such algorithmic knowledge-­making,
more affective than cognitive. This machinic registering of relations accu-
mulates to make legible other­wise unknown connections between sensory
data, and it does so with the potential (if not intention) for that registering
to make p ­ olitical claims: to function as a kind of witnessing of what might
other­wise go undetected.

96 Chapter Two
figure 2.3. Four variations of synthetic media from ­Triple Chaser, Forensic
Architecture, 2019. Courtesy of Forensic Architecture.

figure 2.4. Applying weathering and wear effects to synthetic cannisters, Forensic
Architecture, 2021. Courtesy of Forensic Architecture.
Underpinning the proj­ect is the proposition that social media and other
image platforms contain within them markers of vio­lence that can and
should be revealed. For the machine learning algorithm of ­Triple Chaser, the
events to which it becomes responsible are themselves computational: ma-
chinic encounters with the imaged mediation of tear gas canisters launched
at protesters, refugees, and mi­grants. But their computational nature does
not exclude them from witnessing. With so much of the world now e­ ither
emergent within or subject to computational systems, the reverse holds true:
the domain of computation and the events that compose it must be brought
within the frame of witnessing. While the standing of such counterforensic
algorithms in the courtroom might—­for now—­demand an expert ­human
witness to vouch for their accuracy and explain their pro­cesses, witnessing
itself has already taken place long before testimony occurs before the law.
Comparisons can be drawn to the analog photo­graph, which gradually be-
came a vital mode of witnessing and testimony, not least in contexts of war
and vio­lence.64 Yet, despite its solidity, the photo­graph is an imperfect wit-
ness. Much that m ­ atters resides in what it obscures, or what fails to enter the
frame, as in the nonhuman witnessing of Aleppo’s aftermaths that I examined
in the last chapter. With the photo­graph giving way to the digital image and
the digital image to the generative algorithm, the ambit of witnessing must
expand. As power is increasingly exercised through and even produced by
algorithmic systems, modes of knowledge-­making and contestation predi-
cated on an ocular era must be updated for an age of more overt and complex
machinic affect-­ability. Forensic Architecture’s work is also a potent reminder
that nonhuman witnessing is a m ­ atter for galleries and activist politics as
much as the courts, providing the aesthetic means for the h ­ uman to compre-
hend its constitutive entanglement with the non. Even if the law resists the
displacement of the ­human, art does not.
As ­Triple Chaser demonstrates, algorithmic witnessing trou­bles both
relations between witness and evidence and t­ hose between witnessing
and event. This machine learning system trained to witness via synthetic
datasets suggests that the linear temporal relation in which evidence—­the
photo­graph, the fragment of tear gas canister—is interpreted by the h ­ uman
witness cannot or need not hold. Through their capacities for recognition and
discrimination, nonhuman agencies of the machinic system enact the witness-
ing that turns the trace of events into evidence. Witnessing is, in this sense,
a relational diagram that makes pos­si­ble the composition of relations that in
turn assem­ble into objects that can be experienced. If witnessing precedes

98 Chapter Two
both evidence and witness, then witnessing forges the witness rather than the
figure of the witness granting witnessing its legitimacy and standing.
While this pro­cessual refiguring of witnessing has ramifications for non-
human agencies and contexts beyond the algorithmic, Forensic Architecture’s
movement into this space suggests the strategic potential for an alternative
politics of machine learning. In the four years since the release of ­Triple
Chaser, Forensic Architecture has extended their use of machine learning
to deal with identifying R­ ussian tanks in Ukraine and other investigations.
While I firmly believe that skepticism ­toward the emancipatory and resistant
potential for machine learning and algorithmic systems more generally
is warranted, t­ here is also a strategic imperative to do more to ask how
such systems can work for ­people rather than against them. With its tools,
techniques, and synthetic media databases all made open source, Forensic
Architecture aims to de­moc­ra­tize the production of evidence through the
proliferation of algorithmic witnessing that works on behalf of ngos, activ-
ists, and oppressed ­peoples, and against the technopo­liti­cal state. This inves-
tigative commons becomes an intensive field for nonhuman witnessing, in
which the entangled agencies of machines and h ­ umans work to register and
make addressable other­wise elusive vio­lence.

unwitnessed: proj­e ct maven and limitless data

In June 2018, word spread inside Google that the com­pany was partnering
with the US Department of Defense (DoD) to apply its artificial intelligence
expertise to the identification of objects in drone footage. A week ­later, the
same news broke on the tech site Gizmodo. Within days, Google had with-
drawn its engagement and released a set of princi­ples for ai development
that precluded working on weapons systems, although with plenty of wiggle
room for other defense and national security applications.65 The controversy
marked a new notoriety for Proj­ect Maven, the code name for the Algorith-
mic Warfare Cross-­Functional Team (awcft) created in April 2017 by order
of the Deputy Defense Secretary Robert Work. Its stated aim was to “turn
the enormous volume of data available to DoD into actionable intelligence”
with an initial focus on providing “computer vision algorithms for object
detection, classification, and alerts” in full-­motion video from drone sys-
tems.66 The awcft had a mandate to “consolidate existing algorithm-­based
technology initiatives related to mission areas of the Defense Intelligence

Witnessing Algorithms 99
Enterprise, including all initiatives that develop, employ, or field artificial
intelligence, automation, machine learning, deep learning, and computer
vision algorithms.”67 Not only would the team seek partnerships with Silicon
Valley, it would also adopt tech industry development techniques, such as
iterative and parallel prototyping, data labeling, end-­user testing, and algo-
rithm training.68 In a reversal of the Pentagon’s typical hierarchical, drawn
out, and multiyear technological development pro­cesses, Proj­ect Maven
would be agile. It would fail often and learn quickly; move fast and break
­things—­but with weapons systems.
Military secrecy makes even an approximation of the scale of data requir-
ing analy­sis impossible to determine. Media reports suggest that the propor-
tion of drone sensor data currently analyzed by h ­ umans represents a tiny
fraction. An article in Wired cites DoD officials claiming that 99 ­percent of
all drone video has not been reviewed.69 Proj­ect Maven boss General John
Shanahan is quoted as saying that twenty analysts working twenty-­four hours
a day are able to successful analyze—­exploit, in military lingo—­around 6 to
12 ­percent of imagery from wide-­area motion sensors such as the argus-­i s
per­sis­tent surveillance platform discussed in chapter 1. Proj­ect Maven aimed
to bring ai analy­sis to the full-­motion video data from the drone platforms
­doing much of the surveillance work against isis in Iraq and Syria: the mq-1c
Gray ­Eagle and the mq-9 Reaper. By February 2017, DoD had de­cided that
deep learning algorithms should ultimately be able to perform at near h ­ uman
levels but recognized that to do so meant working at scale. In its initial scop-
ing, Proj­ect Maven was intended to enable several autonomous functions,
including identifying thirty-­eight dif­fer­ent classes of objects, reverse image
search, counts within bounded boxes and over time, and selective object
tracking. It would integrate with Google Earth, ArcGIS, and other geographic
information systems (gis). Building datasets able to train machine learning
systems would require h ­ uman tagging of huge amounts of data. According
to media reports, Proj­ect Maven outsourced much of this to the piecework
platform Figure Eight (formerly CrowdFlower), providing unclassified and
nonviolent images with instructions to draw and label boxes around vari­ous
objects. Combined with classified imagery tagged by internal analysts, this
data could train the convolutional neural network algorithms to identify
and classify objects within video feeds, using iterative training and testing
techniques honed in the tech industry.
Stored as ones and zeroes demarcating the position and color of pixels
and accompanied by crucial metadata that makes them legible to the com-
putational system, ­these images are optical only in potential. ­Unless called

100 Chapter Two

up by a ­human analyst for display on screen, the optical, communicative, and
repre­sen­ta­tional modality of such images remain potential only. Full-­motion
video (fmv) of the December 2013 drone strike on a wedding pro­cession
in Yemen carries no connotative or denotative meaning for the algorithm,
despite its horrifying toll on the families and communities that lost a dozen
lives.70 The event’s significance is obtained purely through its relations of
similarity and difference to the sets of attributes invisually perceived by the
learning algorithm. All such images are si­mul­ta­neously virtual and a­ ctual along
several dimensions at once: virtual code carry­ing the seeds of ­actual optical
imagery; ­actual correlation in the unidentified scatter of virtual arrange-
ments of pixels; virtual events crowding into the actualizing tendencies of
the learning algorithm. Flagged as significant—­a truck moving too swiftly; a
cluster of bodies on a roof—­virtual and ­actual coalesce to pull the sequence
to prominence. We might name this recognition: the algorithm ­doing its job
of observing and discriminating. But the algorithm does more than recognize,
and we know from Oliver that recognition alone is not sufficient to produce
witnessing. Such algorithms forge a relation of responsibility, rendering a set
of relational attributes actionable within the field of possibilities produced by
the rules of engagement and other framing structures of war. By producing
claims to know the world that demand response, even if that response is to
pull a trigger that kills, algorithmic witnessing within the drone apparatus
does something more than mere observation. Or, rather, within a certain
[103.158.136.64] Project MUSE (2024-02-06 19:57 GMT) University of Edinburgh

confluence of machinic affective dynamics, the drone video algorithm gener-

ates a field of human-­nonhuman relations that becomes witnessing.
Applying ­these same princi­ples to activities more complex than object
identification—­assembling machine learning tools that can determine that
a par­tic­u­lar confluence of objects and attributes constitute a target—­both
heightens the stakes of nonhuman witnessing and introduces new prob­lems
into the technical pro­cesses themselves. In the signature strikes undertaken
by the US military, a narrow set of data points—­most of them drawn from
cellphone signal interceptions—­provides the basis for algorithmic analy­sis.
As Amoore writes, “When a random forest algorithm sentences someone to
death by drone strike, the infinite (gestures, connections, potentials) makes
itself finite (optimal output, selector, score), and the horizon of potentials
is reduced to one condensed output signal.”71 This violent mediation is pro-
duced by and through machinic affects: “A random forest algorithm ­will
never know a terrorist in the sense of acting with clearsighted knowledge,
but it mobilizes proxies, attaches clusters of attributes, and infers be­hav­iors
to target and to act regardless.”72 Intensities of relation spark the algorithm

Witnessing Algorithms 101

into response: action necessitated by clusters of machinic intensities coming
together to stir a determination that carries with it an ethical weight.
From the perspective of the machine learning system, shifting the empha-
sis of analy­sis from cellphone metadata to video imagery is largely a ­matter
of complexity and access to large arrays of the gpus necessary for computing
imagery at scale, which is not to say that such systems ­will work accurately,
limit vio­lence, or reduce civilian harm. In much the same way that deepfake
tools can be trained on audio as well as video, the distinctions are largely a
­matter of input data. Yet while the awcft has access to countless hours of
mundane footage that can provide a base dataset for tasks like identifying
vehicles and buildings, fmv of threats that might warrant lethal action or
even tracking by drone systems seemingly remain too scarce, too ambiguous,
or too like other imagery. ­There simply ­isn’t enough video for the machines
to learn effectively. Using similar techniques to Forensic Architecture, the
awcft reportedly generated artificial environments to produce training
data for threat detection systems.73 While details of that training data remain
classified and inaccessible, the introduction of synthetic data into a target-­
detection system has a rather dif­fer­ent valence from its use by Weizman and
his collective. We know that both war and policing, its domestic corollary,
depend upon and reproduce existing sociocultural codings, particularly
­those around race, gender, and class. In a country like A ­ fghanistan, where
men are frequently armed, the baseline designation of “military-­aged males”
predisposes the system to see activities such as the jirga or council as in-
cipient threats. Just such a prefiguring contributed to the unconscionable
drone strikes against just such a gathering at Datta Khel, a village in North
Waziristan, that killed forty-­four ­people in 2011. If training data is synthe-
sized within existing frames of war, what structures of domination and their
attended biases, misconceptions, and fantasies are coded into such training
materials? How might predictive tools gear ­toward identifying certain
­peoples and activities as threats? Nonhuman witnessing within the algorith-
mic systems might bear false witness in far more subtle and ingrained ways
than deepfakes can manage.
Drone warfare and drone policing alike are necessarily racializing: they
encode and produce racialized subjects as threats, with threat and race inti-
mately bound up with each other.74 Race is coded into the drone system all
too readily. This is ­because, as Ruha Benjamin writes, “race itself is a kind of
technology—­one designed to separate, stratify, and sanctify the many forms
of injustice experienced by members of racialized groups, but one that ­people
routinely reimagine and redeploy to their own ends.”75 In the algorithmic

102 Chapter Two

shift from visual to nonvisual regimes of classification, Thao Phan and Scott
Wark argue that “race emerges as an epiphenomenon of automated algo-
rithmic pro­cesses of classifying and sorting operating through proxies and
abstractions,” refiguring “racial formations as data formation.”76 According
to Lauren Wilcox, drone warfare, primarily deployed in the Islamic world,
“si­mul­ta­neously produces bodies in order to destroy them, while insisting on
the legitimacy of this vio­lence through gendered and racialized assumptions
about who is a threat.”77 Identifying specific bodies as threats necessitates
preconditioning them as threats within the system, which means determin-
ing which bodies should be coded for exposure, to borrow a phrase from
Benjamin. Generating training sets from synthetic events staged in 3d envi-
ronments begins with a set of decisions about who and what to include, what
­people and places should look like, how p ­ eople w
­ ill act, and what constitutes
and defines the relations between places, ­people, and actions. Both the pro-
duction of data and the iterative development procedures used by Proj­ect
Maven mean that the nonhuman witnessing of its algorithms is meshed with
the ­human. Far from removing ­human partiality, such pro­cesses embed the
discursive, affective, and fantastic logics of war in all their racializing and
gendering dimensions into the algorithm at e­ very stage of its design, train-
ing, and operation. Nonhuman witnessing in the context of drone warfare is
thus not a move t­ oward impartiality or the diminishment of the h ­ uman, but
rather the technical concretization in code of predetermined meanings that
are inescapably colonial and racist.
Proj­ect Maven and its ai ilk train the martial gaze on the unwitnessed
events of life in the age of drone warfare. Yet this witnessing is not analogous
to the ­human: it is fractured and distributed within the system, a techno-­
affective witnessing composed of machinic intensities. Signature strike, threat
detection, and targeting algorithms are not witnessing subjects in the hu-
manist sense, but witnessing assemblages distributed across the nonhuman,
invisual perception of machine learning systems. This invisual perception
necessitates the exclusion of much that is captured by the drone’s optical
and multispectral sensors: an infinitude of moments both major and minor
necessarily go unwitnessed. Or, rather, nonhuman witnessing within drone
algorithms always entails unreasoned and psychotically rational judgments
about what ­matters. Nonhuman witnessing in its algorithmic, war fighting
form must necessarily fail to note forms of vio­lence (martial, environmen-
tal, interspecies, interpersonal) that do not figure in the criteria for tagging
imagery and reinforcing machine learning. Such imagery does not go “unseen”
as such, but rather its seeing fails to register. Algorithmic witnessing of this

Witnessing Algorithms 103

kind is necessarily narrow, not more efficacious or richer than the h
­ uman
but stunted and strange.

witnessing the algorithm: beyond the black box

Safiya Noble’s Algorithms of Oppression opens with a now famous anecdote

about searching for “black girls” on Google and finding porn sites, then
searching for “white girls” and finding young girls at play. Throughout her
book, Noble shows what has since become well understood: algorithmic
systems repeat, entrench, and even sharpen the racism, misogyny, homopho-
bia, and other normative biases already pre­sent in the cultures from which
they arise. Faced with the technical, l­egal, and commercial black boxing of
algorithms, critical scholars such as Noble have rightly focused on the institu-
tions, structures, and applications through which data is collected, computed,
and instrumentalized by government and corporate entities. Noble’s work is
part of a growing body of critical practices (scholarly, artistic, and activist)
concerned with the reproduction of ­inequality in algorithmic systems that
has had a significant impact on public debate. In the United States, Joy
Boulminwi’s activist research and poetry exposed the biases of facial rec-
ognition, while V ­ irginia Eubanks’s ethnographic investigation revealed the
inequalities exacerbated by the infiltration of algorithms into social welfare
systems. In Australia, collective advocacy spearheaded by journalist Asher
Wolf and ­others forced the federal government to abandon its automated
“RoboDebt” welfare debt collection tool. ­These, and many other interven-
tions, have increased public understanding of the existence and effects of
coded bias, and forced the tech industry to take steps to redress its harms:
creating ethics boards, inviting critical research, appointing bias engineers,
and seeking to diversify their workforces. But such gestures are often mere fig
leaves, swiftly sidelined or rolled back when their presence becomes uncom-
fortable, evidenced most prominently by the dismissal of Timnit Gebru from
Google’s Ethical ai team for her refusal to withdraw an academic paper from
publication that raised ethical and ecological concerns about large language
learning models.78 Achieving what Lina Dencik calls “data justice” requires
more radical change and tech companies and their clients have been far less
inclined to ask ­whether certain computational systems should be built at all.79
Nor have they been willing to peel away the ­legal and commercial wrapping
on their algorithmic black boxes, which are tightly guarded as data becomes
a con­temporary form of capital.80 And even if ­those boxes ­were more open,

104 Chapter Two

the question of w ­ hether deep learning algorithms would reveal themselves in
a comprehensible way remains fraught. How, then, to witness the algorithm?
One answer to this question is in the witnessing of their effects: the increas-
ingly common irruptions of outrage in response to injustices perpetrated by
algorithmic systems. Or we might think of glitches in the algorithmic that
make vis­i­ble their computational construction, such as the strange fault
lines and blurs that can be found on Google Earth, the misrenderings of the
world-­making, machine learning network responsible for Microsoft Flight
Simulator, or the hallucinations of Microsoft’s chatbot version of its Bing
search engine. Another strategy has been to make infrastructures themselves
vis­i­ble. If we cannot see into the algorithm itself in a meaningful way, then
might its infrastructural assemblages be worthy of attention? Trevor Paglen’s
eerie photo­graphs of the National Security Agency Building and of milita-
rized data centers across the United States are one such proj­ect. Another is
John Gerrard’s “The Farm” (2015), which used aerial photography of a Google
data center in Oklahoma as the reference for a finely detailed simulation of
the same center. Exhibited as a high definition projection, Gerrard’s work
stays on the outside of the data center itself but re­creates its digital world
through the construction of a computational simulation using the Unigine
gaming engine.81 Venturing speculatively within the data center itself, Kynan
Tan’s “Polymorphism” (2016) digitally re­creates its interior materialities, such
as the nonhuman movements of automated tape back-up systems as they
robotically traverse arrays of server racks. When first exhibited, Tan used
power­ful subwoofers to sonically simulate the noise of the data center, a sen-
sory engagement with the imposing infrastructures, resource intensity, scale,
and speed of computation without making specific operations legible. In ­these
and other such works, the algorithmic is witnessed not through its code but
through its infrastructures: the hard, imposing materialities that undergird
and make pos­si­ble the purportedly ephemeral clouds of global computation.
Within the ai research community, this prob­lem of invisibility and inac-
cessibility is well recognized. For example, the computer science association
acm now runs FAccT, an annual interdisciplinary conference on fairness, ac-
countability, and transparency in ai.82 San Francisco com­pany OpenAI, now
widely known for its ChatGPT platform, provides one response to the issue
of ai black boxing via its Two-­Minute Papers channel on YouTube, which
pre­sents the learning undertaken by neural networks in two-­minute anima-
tions that show both what and how learning occurs over time, using newly
published papers as the primary source. With more than 1.1 million subscrib-
ers and many videos viewed several million times, the channel represents

Witnessing Algorithms 105

a remarkably effective approach to making vis­i­ble the learning pro­cesses
of ai systems. In one video, an OpenAI neural network learns to play hide
and seek, using boxes, ramps, walls, and game rules to succeed (figure 2.5).
As the environments change and the algorithm learns, strange happenings
occur: agents run up ramps to jump on boxes, use glitches in the simulation’s
physics to fly through the air, and throw objects off the screen. Both the limits
and possibilities of machine learning are immediately evident. Algorithms
become agencies that seem comprehensible ­because their workings can be
broken down into episodic form and cutely animated. Despite their pro-
pagandistic intent, t­ hese videos hint at possibilities for witnessing through
making machine learning vis­i­ble—­even if OpenAI has become increasingly
secretive about how its ChatGPT uses the gpt series of large language models.
Placing this dynamic within contexts of l­abor, logistics, and warfare,
Kynan Tan’s Computer Learns Automation (2020) slows down the machine
learning p­ rocess within three simulated environments and in d ­ oing so al-
lows a ­human audience to become cowitness to the nonhuman witnessing
of ai training. Computer Learns Automation is composed of three separate
training environments and agents: “RideShare,” in which a vehicle learns to
navigate an urban environment to pickup and drop-­off fares; “Robot Arm,”
in which an automated device learns to pick up boxes from one con­vey­or

figure 2.5. Still from “OpenAI Plays Hide and Seek . . . and Breaks the Game!,”
OpenAI, 2019

106 Chapter Two

­ elt and place them on another; and “Drone Strike,” in which a targeting
b
reticule learns to move across terrain, identify h ­ uman targets, and launch
accurate missile strikes against them. Exhibited at the Adelaide Biennial 2020
at the Art Gallery of South Australia, Tan’s work slowed the neural network’s
learning ­process down with the intention that all three agents develop reli-
able capability in their tasks during one hundred days on display. On three
high-­definition screens, visitors to the gallery could watch thousands upon
thousands of learning cycles as the car, robot arm, and crosshairs moved
across the screen. On the lower left of each screen, data readouts list training
time, training steps, total episodes, current and highest scores, and so on,
as well as information more specific to each agent, such as boxes moved or
casualties inflicted. This set-up allows viewers to watch in real ­time as the
machine learns to navigate and act on its environment, seeing what it sees
as it learns. Situated in identified contexts rather than playing hide and seek
or undertaking an abstract task, relations between l­abor, death, and value
production are visually tied to the learning of the machine.
Computer Learns Automation was built in the Unity 3d engine using its
native ml-­Agents package, which provides a set of trainable algorithms that
can be linked to the simulated environment so that it can send and receive
data. This allows the algorithm to learn and act in the environment si­mul­
ta­neously, which in turn makes it pos­si­ble to watch the ­process of the ma-
chine’s learning unfold. Slowed down, each learning cycle of the system can
be a viewed in a legible, computational real ­time. On display in Tan’s work
is an algorithm known as “proximal policy optimization,” a reinforcement
learning neural network technique developed by OpenAI. ­Earlier “policy”
deep learning algorithms used analy­sis of an environment to select between
pos­si­ble options for an agent, but had a tendency to be overly influenced by
choices around how much or l­ittle the policy should adjust in response to
stimuli: “Too small, and pro­gress is hopelessly slow,” notes OpenAI, “too large
and the signal is overwhelmed by the noise, or one might see catastrophic
drops in p ­ erformance.”83 Proximal policy optimization, or ppo, corrects
this tendency by feeding updates back to the policy at each step, aiming to
produce just enough reward to ensure the network learns quickly but ­doesn’t
rush down a false path. Tan’s “Rideshare” vehicle learns to find, collect, and
deliver fares by attempting an action, receiving a reward—or not—­and then
updating its policy to reflect that information. Driving forward, avoiding a
building, not hitting pedestrians—­actions such as this accumulate through
cycles of training to teach the algorithm to achieve an objective. ppo requires
a lot of sample cycles to be effective, but it balances the network’s dueling

Witnessing Algorithms 107

desires to explore options and to exploit what it knows w ­ ill reap rewards.
Computer Learns Automation reveals the stuttering, iterative, inhuman, and
imperceptible graduations through which this mode of deep learning takes
place.
In contrast to the crisp aesthetics that characterize much of Tan’s digital
work, Computer Learns Automation has an operative, pro­cessual quality. Its
simulated environments are just real enough: decodable by the viewer in the
domain of signification while appropriate to context for the machine. Yet
while the environments, agents, and actions are recognizable to the visitor to
the gallery, their semantic significance within the machinic network eludes
comprehension. Just what is happening in each of the three learning simula-
tions is uncomfortable, even unsettling. Machinic affects course through the
visual field yet are themselves what Mackenzie and Munster call “invisual”:
concerned with the composition of relations that are not themselves visual
at all but rather the perceptual upwell of operations deep within the hidden
layers of the ppo network as it explores and exploits according to technical
logics that retain a certain unassailable incomprehension. As the targeting
reticule of “Drone Strike” inches its way uncertainly across the mountainous
terrain, learning to find, fix, and finish the small collection of pixels that in-
dicates a ­human body, its movements are not ­those of the intentional h ­ uman
operator but rather of a machine motivated by an autonomic system for
which the notion of a meaningful goal is itself without meaning—or, rather,
without correlative meaning for the ­human viewer (figure 2.6). The machinic
resides in its emerging relational awareness, its becoming operative. Tan invites
us into the disjunctive space where unthinking yet agential, operative, and
transformative machine learning systems intersect with bodily vio­lence, ­labor
exploitation, and automated logistics.
The ­human audience can only ever become cowitness, a status reinforced
by the practical impossibility of watching all three environments learn over
one hundred days. Yet this partial cowitnessing is crucial to pulling the non-
human witnessing already at work in the visualization of the invisual learning
­process. With casualty counts, f­ounder wealth, and shareholder value tick-
ing up as the machine learns in its slowed down real time, Computer Learns
Automation reminds us of the necessity of nonhuman witnessing bridging
the seeming divide between technical systems, material conditions, and the
politics of technocultures. Tan insists on pursuing the always incomplete task
of witnessing algorithms in their elemental states, in their becoming increas-
ingly, brutally, and efficiently operative through the intensification and accu-
mulation of machinic affects within the invisual domain of computer vision.

108 Chapter Two

figure 2.6. Still from Computer Learns Automation, Kynan Tan, 2020.
Courtesy of the artist.

witnessing algorithms

Across this chapter, witnessing algorithms has emerged as a polysemic con-

cept. In the narrowest sense, witnessing algorithms can be understood quite
straightforwardly as algorithms that enable witnessing. In this vein we might
think of social media algorithms that bring certain kinds of news reportage
to the fore or algorithmic systems such as ms Flight Simulator’s that enable
an entirely simulated witnessing of the world. Yet witnessing algorithms are
also engaged in witnessing on their own behalf: a registering of happenings
in material and machinic worlds to which the algorithm obtains its own ver-
sion of responsibility. Algorithmic responsibility is not identical to that of
its (multiple, contestable) ­human equivalent, but rather describes the emer-
gence of relations that cohere and produce a necessity for action, however
that might come to be within the technicity of the algorithm. Consider how
images of the ­Triple Chaser tear gas canister come to ­matter in the machine
learning algorithms of Forensic Architecture, or the par­tic­u­lar movements
of bodies and vehicles in computer vision designed for drone applications
­under the aegis of the Algorithmic Warfare Cross-­Functional Team. Exactly
how ­those neural networks raise certain phenomena to significance but not

Witnessing Algorithms 109

­ thers—­how worldly happenings trigger machinic interest or not—­eludes
o
complete knowing, yet as the algorithm learns it is s­ haped by its (un)wit-
nessing. At the same time, algorithms are themselves entities that must be
witnessed, both in their effects and in their operations. As Kynan Tan’s Com-
puter Learns Automation teaches us, at issue ­here is the unknowable machinic
affects of the learning networks themselves. Witnessing the algorithmic is
not another demand to open the proverbial black box, but rather a call for
attending to algorithmic agencies in their emergence and on their own terms.
With algorithms swiftly becoming preeminent knowledge instruments
of governance, commerce, culture, science, and social life, how we reckon
with the technopolitics of their identification and formation of relations be-
tween worldly phenomena is an urgent question. As the hype of generative
ai reaches fever pitch, the ­political stakes of this task only heighten. Within
critical algorithm studies and data justice movements and activism, crucial
new lines of inquiry continue to multiply, not least in conjunction with the
enduring ­political proj­ects of resisting settler colonialism, struggling for
racial justice, and fighting for a meaningful response to the climate crisis.
This chapter has sought to think in sympathy with t­ hose inquiries, proj­ects,
and goals, asking how algorithmic technologies witness the world and how
they in turn might be witnessed. My approach h ­ ere follows Amoore’s call for
a cloud ethics that “does not belong to an episteme of accountability, trans-
parency, and legibility, but on the contrary begins with the opacity, partiality,
and illegibility of all forms of giving an account, ­human and algorithmic.”84
Following Édouard Glissant, such an opacity provides the ground for un-
derstanding knowledge and politics as emerging from irreducible difference
rather than being founded upon its erasure.85 It is the condition that under-
pins relation. Pluriversal politics necessitates just such an opacity, as well as
its accompanying partiality and illegibility, b ­ ecause a world of many worlds
requires the impossibility of transparency, other­wise any world becomes leg-
ible to any other, and in d ­ oing so ceases to possess vitality on its own terms.
In the domain of algorithms, analytically separating witnessing within the
event from the testimony that takes place ­after is urgent ­because so much
of witnessing algorithms is about registering events, machinic or other­wise,
as pro­cesses of emergent relations (of knowledge, power, vio­lence, control
and aesthetic and p ­ olitical potential)—­rather than explaining, narrating, or
communicating them.
Perhaps more than any other domain addressed in this book, the algorith-
mic cuts across world-­ending catastrophes old and new. Algorithms are infra-
structural, institutional, and embedded in technocultural milieus that are,

110 Chapter Two

in turn, inseparable from the production of race, gender, and class. Rooted
in settler-­colonial practices of categorization and control, financialization,
climate modeling, and war, as much as in social media and internet search,
algorithms are yet another site of deep entanglement between h ­ uman and
nonhuman. Nonhuman witnessing of, in, and through algorithmic pro­cesses
is about finding ­those links and recognizing that it no longer, if it ever did,
makes sense to think of the h ­ uman witness outside the nonhuman technici-
ties of our media, our archives, and their agential materialities. Nonhuman
witnessing arises from a field of relations between the h ­ uman and nonhu-
man, relations that are as ecological as they are technical—­and it is to the
ecological that this book now turns.

Witnessing Algorithms 111

chapter
three

witnessing
ecologies
in the communiqué from the Fiftieth Pacific Islands Forum, held in Tuvalu
in August 2019, leaders from the region “reaffirmed climate change as the single
greatest threat to the livelihoods, security, and wellbeing of the p­ eoples of the
Pacific,” but ­stopped short of calling for significant and immediate action. In
the Kainaki II Declaration that accompanied the formal communiqué, coun-
tries are called to “reflect” on transitioning away from coal rather than banning
its use, “meet or exceed” national emissions reductions rather than creating new
and more ambitious ones, and continue “efforts ­towards” meeting international
climate-­funding promises rather than demanding urgent and ambitious
commitments.1 According to media reports, Australia successfully stymied
efforts for a much bolder declaration, reducing Prime Minister Akilisi Po-
hiva of Tonga to tears and prompting leaders from Fiji, Vanuatu, and Tuvalu
to make heated remarks about their more power­ful neighbor. “We came
together in a nation that risks disappearing to the seas, but unfortunately we
settled for the communiqué,” said Fiji’s prime minister, Frank Bainimarama.
“Watered-­down climate language has real consequences—­like water-­logged
homes, schools, communities, and ancestral burial grounds.”2 The ire of
Bainimarama and o ­ thers was directed primarily at then Australian Prime
Minister Scott Morrison, a man who once triumphantly brandished a lump
of coal in Parliament, only reluctantly accepted the science of climate change,
and stalled pro­gress to limit emissions and develop renewable energies at
­every opportunity, achieving the ignominious distinction of Australia rank-
ing dead last among 170 states analyzed in a 2021 un report on climate ac-
tion.3 At the forum, it seemed Morrison wanted ­every dollar Australia spent
in the Pacific to be recognized, but refused to commit to any action that
might slow the rising seas threatening to swallow Tuvalu and other islands.
Much can be said of events such as this and the warped politics of climate
change, the enduring inequalities that underpin the failure to act by wealthy
nations, and the histories of colonialism, clientelism, and militarism that
shape the pre­sent Pacific. Just as the Marshall Islands and other nations in
the Pacific w ­ ere crucial sites for nuclear testing throughout the Cold War,
so too are they now the canaries in the mineshaft of climate change. Indeed,
Elizabeth DeLoughrey points out that “climate science and nuclear weapons
testing have an intimate relationship,” as the tools and techniques for un-
derstanding the atmosphere developed for war w ­ ere applied to establishing
carbon baselines and monitoring their change.4 Climate crisis is thus sutured
to “catastrophic ruptures to social and ecological systems” that “have already
been experienced through the violent pro­cesses of empire” and continue in
the ongoing, unnamed imperialism of regional geopolitics.5 Climate is itself
increasingly a military prob­lem, securitized by planners in ways that have
­little regard for the wellbeing of populations most subject to it.6 When the
Islander leaders of the Pacific juxtapose Australia’s domestic energy pricing
concerns with the erasure of life, culture, and community, it makes clear
that trauma is not registered as an individual experience but as an ecological
phenomenon. Pleas for an a­ cceptance of shared responsibility in the face of
drowning depends on shared witnessing, on opening onto impossible loss,
grief, and ecological trauma.
Among the most widely known evocations of the drowning islands of the
Pacific Ocean are the poems of Kathy Jetñil-­Kijiner, a Marshallese spoken word
­performance artist and writer. Her poem “Tell Them” includes t­ hese lines:

tell them about the ­water

how we have seen it rising
flooding across our cemeteries
gushing over the sea walls
and crashing against our homes
tell them what it’s like
to see the entire ocean__level___with the land7

Witnessing Ecologies 113

DeLoughrey observes that the poem “employs allegory to figure the island
as a world in ecological crisis, depicts an active, nonhuman ocean agent, and
articulates the imperative to both witness and testify to a dynamic, changing
Earth.”8 Allegory, she argues, is one of the most power­ful forms of cultural
narration of climate crisis precisely b ­ ecause it bridges ruptures in knowl-
edge, experience, and culture through “its ability to represent both historical
and scalar relations” by animating “universalizing tropes such as planet,
species, nature, and the h­ uman into narrative—­and thereby into space and
time.” While the poem can certainly be read in an allegorical mode, ­there is
9

something ­else at work ­here beyond recognition that “the Marshallese are
both h­ umans and nonhumans.”10 The repeated refrain of the poem’s m ­ iddle
stanzas—­“tell them we are . . .”—­intermingles ­human and non, “hollow
hulls” and “wood shavings,” “sweet harmonies” and “styrofoam cups of
koolaid red.”11 Distinctions of status slip away between “skies uncluttered”
and “dusty rubber slippers” as space, place, object, speech, and gesture
become entangled with the “we” of the poem. H ­ ere is a complex ecol­ogy,
one rendered sensible—­able to be felt—­through the rhythm and rhe­toric of
the poem but not reducible to language. “We are” might also be an assertion
of ontology, of shared being-in and becoming-­with the world that is slowly
being drowned. “Tell Them” is an allegorical call for climate justice, but
also an address to the nonhuman entanglements already rupturing in the
refracting wakes of catastrophic pasts and f­ utures. Its witnessing demands
not only response-­ability on the part of the state system, but also that the
rich ecologies of the Marshall Islands be recognized as response-­able and
address-­able. If this poem—­and ­others like it—­are calls to witness and acts
of testimony, then their mode of witnessing is nonhuman, animated by the
inextricable entanglements of being, land, living, community, ocean, and
culture.
This chapter coheres around the proposition that one way that ecologi-
cal trauma—­complex, mutable, resilient, ephemeral, material, moving,
unsettling—­comes into focus is through aesthetic works that undertake the
tentative, always incomplete proj­ect of nonhuman witnessing. In pursu-
ing this proposition, I attend to artistic and literary works that examine
ecological trauma through scale as a site of p ­ olitical contention and in
the existential rupturing of nuclear weapons. Violent mediations and ma-
chinic affects animate the martial and cap­i­tal­ist operations, events, and
technologies that concern this chapter, but ­here I shift my inquiry from
what animates assemblages of catastrophic vio­lence to pursue traumatic

114 Chapter Three

aftermaths for more-­than-­human ecologies. Ecological trauma is not con-
ceptually ­dependent on the terms that have occupied the first two chapters
of this book, but, rather, as I show through this chapter, provides a way of
thinking with and responding to crises and vio­lence that far exceed the
­human sensorium and the prosthetic techniques we have built to enhance
or supersede it.
The chapter begins with an exploration of ecologies in media and cultural
theory, before pursuing scale as a relational problematic in the face of eco-
logical vio­lence. From t­ here, I develop the concept of ecological trauma in
earnest, defining it as a rupturing of relations that ­ripples through the ongo-
ing composition of more-­than-­human ecologies. To explore how nonhuman
witnessing offers a response, I then turn to the vio­lence and trauma of
nuclear testing on First Nations lands in Australia, and the fiercely resistant
art practice of the Kokatha and Nukunu artist Yhonnie Scarce. Rather than
pursue the witnessing of ecologies through scientific frameworks, I aim to
trou­ble the dominance of such forms of knowledge, and particularly the
aerial imaging through which the planet becomes media. While such tech-
nologies are critical to the formation of scientific knowledge about climate
crisis and ecological vio­lence more generally, the growing entrenchment of
remote-­sensing visualizations are what Lorraine Daston calls an epistemic
image, “one made with the intent not only of depicting the object of scientific
inquiry but also of replacing it.”12 This makes probing their limits a necessary
task. Satellite programs such as Landsat certainly fall within the rubric of
nonhuman witnessing, but what they register and make legible strug­gles to
escape the epistemic and technical frames within which it takes place.
Less bound by such strictures, aesthetic interventions make nonhu-
man witnessing sensible and graspable in ways that bring the constraining
frames, structures, politics, and vio­lences of the technoscientific state to
the fore. Continuing empire, mutating colonialism, nuclear testing, endur-
ing irradiation, rising ­waters, mined out lands: catastrophic ­futures are al-
ready ­here and have been for a long time, but that h ­ asn’t s­ topped Man from
charging headlong and ever deeper into oblivion. How, then, to unknot the
forms of knowing and being that make up the ­human of Man, the figure who
proudly waves coal in national parliaments and is unmoved by ­those who
weep for what is and w­ ill be lost? How to reckon with the ecological entangle-
ments wrought by industry and war? How to witness within and through
ecologies of the ­human and more-­than? How to open up a communicative
commons that grants standing to ­human and nonhuman?

Witnessing Ecologies 115

witnessing ecologies

This chapter expands the focus of the book to think with more-­than-­human
ecologies that encompass land, sky, and ­water, rather than remaining within
the technocultural domains that have been its principal preoccupation. As
with its p ­ redecessors, this chapter travels with the doubling movement of
its title: the witnessing of ecologies and ecologies of witnessing. Understood
as complex systems of interacting and interdependent parts, ecologies are
constituted by relations between ele­ments. ­Whether wrought in the split-­
second fission of a nuclear bomb or the drawn-­out temporality of radioac-
tive contamination, ecological vio­lence strikes at the relational composition
of ecologies themselves. Uprooting a verdant tree to clear the way for a new
road is not ecologically violent simply b ­ ecause the tree itself is lost, but
rather ­because its removal tears at the fabric of the ecol­ogy within which it is
webbed. As Cubitt writes: “Ecologies are not networks connecting previously
separate ­things: ­Every ele­ment of an ecol­ogy mediates ­every other. Life medi-
ates nutrients and sunlight, storing, changing, growing, passing, mutating,
returning.”13 Media theorist Matthew Fuller makes the point that the word
ecol­ogy “is one of the most expressive language currently has to indicate
the massive and dynamic interrelation of pro­cesses and objects, beings and
­things, patterns and ­matter.”14 But ecologies can also be brutal, particularly
once we extend their conceptual reach into the violent. “Geopolitics, enacted
through global war, is itself a form of life that pursues a savage ecol­ogy,” Grove
insists, “radically antagonistic to survival as a collective rather than discrimi-
natory goal.”15 Ecologies are not inherently moral, but are rather inescapably
­political on a planet ­shaped by Man.
Conceiving of media as ecological and ecologies as medial provides a
conceptual apparatus through which to examine, in the context of ecological
vio­lence and its attendant traumas, the communicative mode I am calling
“nonhuman witnessing.” As this chapter argues, nonhuman witnessing can
become a reparative response to ecological trauma, the state of wounded
survival that follows in the wake of ecological vio­lence. But it also responds
to a deeper historical rupture between ­human and nonhuman, a cleaving of
Man from Nature that is rooted in Platonic and Aristotelian thought and thus
inherent to the ascendance of Anthropos, even before its violent intensifica-
tion as Wynter’s Man, which I discussed in the introduction and ­will return
to in the coda. “The more ­humans defined themselves over against nature,”
Cubitt observes, “the more they defined nature over against themselves, in
this way formalizing and enforcing the split between the natu­ral environment

116 Chapter Three

and humanity, which in the ­process became a nonnatural, religious, or ratio-
nal quality.”16 As Haraway and other feminist p ­ hilosophers of technoscience
have shown, the recomposition of nature and culture as entangled with each
other is an urgent p ­ olitical task. Haraway coined the term “natureculture” to
enact this intertwined coexistence, insisting that no concept of ecol­ogy could
exclude the ­human, or vice versa. The stakes of this intervention are signifi-
cant, not least for science, which must be understood as operating in agential
entanglement with “nature” rather than observing it from a distance.17 This
chapter commits itself to this task by asking how thinking-­feeling forms in
response to ecological vio­lence.
Aesthetic works figure more prominently in this chapter, as I read artists
and writers who find the means for evoking and establishing communica-
tive relations between ­human and nonhuman, even working to dissolve
such distinctions. The works examined h ­ ere pursue what Cubitt calls an
“ecological aesthetics and politics” that makes pos­si­ble “communication of
and through difference.”18 Nonhuman witnessing of ecological trauma is not
confined to the overt aesthetic production that occupies much of this chap-
ter, yet ­there is always an aesthetics of nonhuman witnessing, in the sense
that aesthetics is inseparable from sensing and its registration. Geological
formations witness the passage of deep time, the arrival and departure of ice
ages, the life and death of forests, and the passage of animal life. Nonhuman
witnessing describes the fixing of fossils rec­ords: captured in mud or peat
or sand, bodies shape the earthy m ­ atter that spills over and claims them, an
ecological affectivity in the transformation of materialities as they fix into
enduring form. Archaeology and geology seek out this witnessing, still in
­process on the planetary scale of time, only to reduce it to “evidence” that
can be ordered into disciplines of knowledge rooted in whiteness, extraction,
and colonialism.19 Biochemistry, biophysics, the quantum mechanisms of the
universe itself: all entail relational dynamics that register change as sensation,
as ele­ments in relations of mediation, becoming through encounter and in
time. But while pursuing this radical empiricism of nonhuman witnessing
to its most elemental would be a worthy if quixotic proj­ect, my concern ­here
is with its concrescence into modes that register more coherently with the
­human sensorium.
With the first satellite images and, ­later, the Earthrise (1968) and Blue
Marble (1972) photo­graphs taken by astronauts, atmospheric sensing has
held out the tantalizing promise of making nonhuman scales, perspective,
and spectrums accessible and sensible. Despite Cold War rivalry and govern-
ment investment in military spending driving the rise of technoscience, the

Witnessing Ecologies 117

extraplanetary view swiftly produced a utopian politics. Most famously,
Stewart Brand’s ­Whole Earth Cata­log played a key role in the emerging Cali-
fornian cyberculture, which in turn spurred the rise of Silicon Valley and helped
create the ideological and material conditions for con­temporary algorithmic
enclosure.20 Militarism and military investment w ­ ere never far from the sur-
face, as the foundational role played by darpa in the creation of the internet
attests, but Brand saw the view from space as si­mul­ta­neously triumphant and
humbling, a testament to ­human achievement but also to the necessity of liv-
ing together on this small blue dot in the expanse of space. Suffused with the
overdetermined figure of Man described by Wynter, Brand’s vision promoted
a universalizing whiteness manifested most potently through the absence of
race and class.21 Imagining the fusion of cyberculture with business as the
means for si­mul­ta­neously acquiring both wealth and liberation would play
no small part in the emergence in our own time of billionaires with space
shut­tles and dreams of colonizing Mars. But Silicon Valley ideology was not
the only offshoot of the capacity to capture the planetary through sensing
apparatuses.
One of the earliest applications of electronic digital computers a­ fter their
emergence in World War II was weather forecasting, an effort led by the
Manhattan Proj­ect mathematician John von Neumann. Even in ­those early
days, von Neumann sought what he called “the infinite forecast,” the capac-
ity to simulate climatic circulations over a long enough period to grasp its
fundamental princi­ples.22 But the origins of climatic computing can also be
found in both nuclear fallout monitoring regimes and the Cold War effort
to render the world computable by early warning nuclear strike systems.
Model simulations required new spatial and atmospheric data, collected
through a range of sensors fitted to planes, boards, floats, and satellites.23 The
latter proved valuable, with the tiros, Nimbus, and Earth Radiation B ­ udget
Experiment satellites enabling the emergent planetary regime to engage in
vertical mediation between the terrestrial and the stratospheric.24 ­These
geospatial satellites and the growing array of sensors sent into orbit with
them produced a new capacity for earth imaging across the breadth of the
electromagnetic spectrum. “Earth imaging,” writes Russill, “now depends on
light recorded from sites that are uninhabitable or inaccessible to h ­ umans, at
wavelengths we cannot perceive directly, travelling at speeds and in quantities
we cannot ­handle.”25
No longer could ­human perception claim a privileged status when it came
to making sense of environments: becoming knowable—to science but also
to militaries, governments, ngos, and even publics—­meant being registered

118 Chapter Three

by remote-­sensing systems collecting data inaccessible to the h ­ uman senso-
rium. Tracking environmental change, w
26
­ hether from natu­ral phenomena
or industrial depredations or the vio­lence of war, no longer operated on any-
thing like a ­human scale. On the ground, knowledge maintained its currency
within par­tic­u­lar contexts, as the rise of humanitarian witnessing attests, but
earth imaging ensured that h ­ uman senses w ­ ere “no longer the grounds for
authoritative depictions of environmental change.”27 Remote sensing is thus
a form of nonhuman witnessing—­one that illustrates precisely the way the
­human is exceeded by emergent technics and by the complexity and scale
of ecologies, and yet also returned to through address and the injunction to
respond. Earth imaging not only made ecological volatility sensible, but it
also conjoined war and environment through satellite-­sensing apparatuses
and the pro­cesses of violent mediation much like t­ hose discussed in chap-
ter 1. This desire to exert control from the atmosphere extends to the desire
to control atmosphere itself via what Furuhata calls “climatic media,” which
ranges from planetary geoengineering to the mundanity of green­houses and
air-­conditioning.28 Environments themselves become subject to the potential
vio­lence of mediated control. Violent mediation is thus as much a part of
climate science and ecological vio­lence as warfare, at work in many of the in-
formational and communicative pro­cesses that facilitate extractive industries
and ­those that enable climate monitoring and regeneration.
As earth imaging has become more accessible, its role within ­human
rights and ecological monitoring has grown as a means of bringing state and
corporate slow vio­lence into the zone of the sensible and ­political. Dutch
nongovernment ­organization Pax for Peace, for instance, employs remote
sensing in its analy­sis of Syria to demonstrate the interrelations of war and
environmental damage. Remote-­sensing data allows Pax to identify damage
over time and to find critical sites within the ecol­ogy, such as makeshift oil re-
fining, and map their location and impacts. But Pax also uses on-­the-­ground
sources, recognizing that earth imaging can elide crucial information and
contextual complexity. As Weizman has pointed out and I discussed in the
previous chapter, the resolution of publicly available earth imagery is often
­limited so that the ­human body fits within its bound­aries, thus placing the
body—­the locus of ­human rights—­below the “threshold of detectability.”29
But this level of resolution fits well with climate-­monitoring regimes and
their computational architectures, including initiatives such as the Micro-
soft Planetary Computer that promises to harness the compute power and
machine learning capabilities of big tech to ecological research and action.
As Delf Rothe puts it, “Visual technologies such as satellite remote sensing

Witnessing Ecologies 119

play a crucial role in the ontological politics of environmental security” and
have “considerably influenced the epistemological horizon of environmental
security thinking.”30 Earth imaging as nonhuman witnessing is thus already
bound up with the savage ecol­ogy of con­temporary geopolitics. A crucial
question is w­ hether it can be other­wise.
Remote sensors are communication technologies, but their traffic is one
way: they bring the nonhuman into the domain of the sensible but provide no
means of address to the nonhuman in response. But as Jennifer Gabrys has
documented, remote sensors also form part of a media ecol­ogy that seeks
to make the earth itself programmable: “Sensing is then not just a ­process
of generating information but also a way of informing experience.”31 Against
large-­scale efforts to make the planet computational for the purposes of
climate monitoring and military targeting alike, Gabrys attends to what she
calls “citizen sensing.” Th­ ese are collaborative, grassroots proj­ects that em-
ploy a diy ethos and enter into a pro­cessual, dynamic relationship between
technics and environment and which, in the terms of this book, could be
understood as an insistence on making nonhuman witnessing ­political. But
returning to the work of Gabrys and o ­ thers through the frame of nonhuman
witnessing is not my purpose ­here. As this chapter unfolds, my interest is less
on remote sensing and earth imaging as modalities of nonhuman witnessing
and much more on how ecologies can be witnessed and how witnessing takes
place ecologically.
In the manifestations with which this chapter dwells, nonhuman witnessing
mediates between what Félix Guattari calls the three ecologies: the environ-
ment, social relations, and ­human subjectivity.32 New modalities for such me-
diation are vital for Guattari: “Now more than ever, nature cannot be separated
from culture; in order to comprehend the interactions between ecosystems,
the mechanosphere and the social and individual Universes of reference, we
must learn to think ‘transversally.’ ”33 As an “assemblage of enunciation” that
conjoins machines, p ­ eople, animals, environments, and objects, nonhuman
witnessing generates transversal relations with the potential for strengthen-
ing into enduring bonds. Mediation animates t­ hese transversal relations,
enabling communicative flows that d ­ on’t just carry information but render
aesthetics sensible by a multiplicity of agencies, h ­ umans among them. As
Fuller points out, “The stakes [Guattari] assigns to media are rightly perceived
as being profoundly p ­ olitical or ethico-­aesthetic at all scales.”34 This question
of scale recurs throughout this chapter: scales of time and space and their
collapse, scales of perspective and intensity, scales of intimacy and vio­lence.

120 Chapter Three

scale as a site of witnessing

With its roots in the Latin scala, meaning ladder or stairs, scale refers to de-
fined relations of space, time, or quantity between one ­thing and another. A
musical scale sets out the relations between one note and a series of o ­ thers,
while a cartographic scale defines the ratio of distances on a map to ­those
on the earth. As Timothy Clark writes, scale “enables a calibrated and useful
extrapolation between dimensions.”35 Scale, then, is one means of making
instrumental and practical sense across difference, a means of managing rela-
tions between one ­thing and another. Scale helps anchor perception in worlds
that extend beyond the perceptual reach of the h ­ uman sensorium; it enables
one to conceive of entities far bigger or smaller, say, than can be contained
within the ­human visual field. This is one of the promises of remote sensing:
not only to extend perception to atmospheric or underwater viewpoints, but
also to enable sensing at spatial and temporal scales that exceed the ­human.
As Fuller points out, “A ‘scale’ is something that operates at one level in what
might be thought of as an infinite zoom, ­were a camera to be built that
could be sensitized to ele­ments as diverse as practices, institutions, atomi-
cal structures, weather patterns, linguistic formations, protocols, transport
infrastructures, a glance.”36 High-­resolution satellite imagery thus not only
enables breadth of perception but also depth through the capacity to zoom
imagery down to the half meter and even smaller. Scale is an epistemologi-
cal tool, a means of o­ rganizing the world and its causal relations. It does not
inhere in any given entity but is an imposed relationality between one t­ hing
and another. At the same time, “a scale provides a certain perspectival optic
by which dimensions of relationality and other scales may be ‘read.” ’37 This
means scale can be intensely p ­ olitical b
­ ecause it constructs relations between
entities and pro­cesses and, in d­ oing so, can become bound up with questions
of agency.
Defining our pre­sent geologic era as the Anthropocene, argues the post-
colonial historian Dipesh Chakrabarty, shifts the scale at which h ­ uman agency
operates: “To call ­human beings geological agents is to scale up our imagi-
nation of the h­ uman . . . to attribute to us a force on the same scale as that
released at other times when t­ here has been a mass extinction of species.”38
But climate change is not only about happenings at the scale of the planet or
even the capacity of the ­human to have effects at the planetary scale. Rather,
Clark argues that it involves “an implosion of scales, implicating seemingly
trivial or small actions with enormous stakes” even as disciplinary, ideological,

Witnessing Ecologies 121

religious, and other bound­aries collapse into one another or delimit knowl-
edge in damaging ways.39 Approaching the prob­lem from a dif­fer­ent a­ ngle,
Derek Woods argues that scale itself should be the site of critique, suggesting
that ­doing so makes clear that “the subject of the Anthropocene is not the
­human species but modern terraforming assemblages.”40 Consequently, as
Alaimo argues, ­those most responsible for the climate crisis need to engage
in “scale shifting that is intrepidly—­even psychedelically—­empathetic, rather
than safely ensconced.”41 If t­ here is an emergent “structure of feeling” around
climate change, it must surely be a generalized anxiety bound up with ur-
gency, disbelief, and futility—­with scale is at its core.42
At stake in ­these and other such interventions is the capacity to overcome
scale as a prob­lem for knowing and communicating climate change. How,
then, might scale itself become subject to politics? How might scale not sim-
ply be communicated but witnessed? That is, how might scale be registered
as a site of necessary ­political and ethical engagement? How might scale, its
effects and its collapse, be grasped as a ­matter of practical world-­making and
repairing?

witnessing scale

The nonhuman witnessing of scale opens onto embodied engagements

with the strange disjunctures of climate change. Th ­ ese disjunctures include
its unfolding into a ­future in which every­one currently alive is dead, and its
weird geographies, its planetary scope and localized effects, its collapse of
distinctions between apparently discrete systems and spaces. All this de-
mands “rethinking perception as unfixed, nonlinear, embodied, and mobile,”
as Zylinska writes in relation to nonhuman photography.43 Scales connect
the ­human and nonhuman in complex, inextricable ways: they bind entities
through relation, yet do so transversally, rather than through any explic­itly
causal interrelation. Scale is a site of nonhuman witnessing b ­ ecause it is a
manifestation, even a technique, for the registration of relations that are not
at all obvious, or that defy ­human experience, or that insist upon incommen-
surability. Witnessing scale, ­whether of time or space or anything e­ lse, means
making p ­ olitical and contestable its structures, assumptions, effects, histories,
and technicities. Technoscientific views from above are a critical convergence
of all t­ hese ­things, not least ­because the view from above also coalesces war,
data, and climate in multiple ways.

122 Chapter Three

Thanks to the global touring of a major exhibition and its accompanying
documentary, the Canadian photographer Edward Burtynsky, along with his
longtime collaborators Jennifer Baichal and Nicholas de Pencier, has played
a significant role in popularizing the term Anthropocene and produced
some of the best-­known interventions into the view from above. Shot using
high-­resolution digital cameras, Burtynsky’s oeuvre documents d ­ ecades
spent traveling to places where the markers of h ­ uman activity on the plan-
etary surface are devastatingly evident. His photo­graphs are arresting, even
disturbingly beautiful, finding in open-­cut mines, polluted deltas, and de-
forested landscapes an aesthetic of shadowed contours, strange colorations,
geometric fractures, surreal surfaces. While some of his work operates at an
immersive ­human scale, almost all his photo­graphs since the late 2000s are
aerial. Typically photographed from a light plane, the images splay out with
just enough perspective so that salt pans in Gujarat, India, or lithium mines
in the Atacama salt flats, Chile, seem to extend in­def­initely beyond what the
camera reveals. In “Salt Pan #18,” asymmetrical polygons of land stagger
away from the bottom of the image in long lines, while in “Clear Cut #3,” the
curling marks of clear-­felled palm oil plantations in Malaysia curve off e­ very
edge of the image. A series from the Morenci Mine in Arizona, USA, renders
the landscape alien: vivid oranges and purples, vivisected by curves and lines
carved by ­immense vehicles that emerge slowly from the image, barely dis-
tinguishable on the monumental prints on gallery walls (figure 3.1). Point of
view and framing together render the images difficult to position definitively:
spatial scales feel monumental but resist enumeration, content escapes form
even as the aerial view seems to offer the possibility of revelation.
Unsettling, even destabilizing, Burtynsky’s photo­graphs are generators
of affective disembodiment, of being temporally shoved out of the h ­ uman
sensorium and placed in relation to the scale of the h ­ uman as geologic agent.
­Human and nonhuman fold into each other, perturbing spatialities of scale
by presenting the planetary terraforming of anthropogenic devastation at the
limits of the h
­ uman.44 What we witness in ­these works is thus the affectivity
of the geological and geometrical, the problematics of scale. ­These images
attest not only to catastrophic ­human intervention in the natu­ral world, but
also to the tension between ­human capacity and aesthetics, between the
technical and the beautiful. ­Here, nonhuman witnessing exceeds what resides
in the visual field: relations of scale are themselves intensive, forceful, and
embodied in the most radical sense of folding the ­human into the nonhu-
man spatialities of climate crisis and its causal agents of extractive industry.

Witnessing Ecologies 123

figure 3.1. “Morenci Mine #2,” Clifton, Arizona, Edward Burtynsky, 2012.
[103.158.136.64] Project MUSE (2024-02-06 19:57 GMT) University of Edinburgh

© Edward Burtynsky, courtesy Sundaram Tagore Galleries, Singapore / Nicholas

Metivier Gallery, Toronto.

Against the scientific aesthetics of satellite imagery, which are typically

framed by indexical legends and produced at specifically determined scales,
Burtynsky’s photography situates the view from above as more contingent,
as embodied in the aesthetic relations engendered by the image. So, while his
photo­graphs buy into the capacity of the view from above to reveal, they re-
sist entry into the epistemic category of the technoscientific view that drives
knowledge production in war and science alike.
Part of what makes climate change such a fundamental ­political challenge
is that it is si­mul­ta­neously an ontological and epistemological prob­lem. Mov-
ing up, down, and between scales, climate change confuses systems of gover-
nance and knowledge. Clark calls this the “derangement of scale,” in which
“received concepts of agency, rationality and responsibility are being strained
or even begin to fall apart in a bewildering generalizing of the ­political that
can make even filling a k­ ettle as public an act as voting.”45 For all the derange-

124 Chapter Three

ments of scale produced by the climate crisis, its defining scalar feature might
well be the collapse of scales, their folding into one another such that scale
itself proves at once illusory and determinate, ephemeral and material. This
ambivalent relationship between the h ­ uman and nonhuman eye, between
creative practices of witnessing and scientific documentation of the world, is
evident in a 2018 work by the Australian media artist Grayson Cooke.
Shown at major venues across the country, “Open Air” combines the
paintings and pro­cesses of artist Emma Walker with satellite images of Aus-
tralia from the Landsat “Digital Earth Australia” program, set to the 2013
­album Open by the cult Australian band the Necks. The work plays with
the visual registers of the aerial view, troubling the mediated materialities of
land and art. ­Running just over an hour, it brings together motion-­controlled
aerial photography of Walker’s abstract paintings with time-­lapse images
from the Landsat archive, which has been returning to image the same point
on the planet ­every sixteen days for more than forty years (figure 3.2). In the
video of Walker’s paintings-­in-­progress, the camera scans surfaces slowly:
intensely immediate, close to rough wood, cracked paint, flowing pigment,
and heat applied to paint. At times, the Landsat images cut sharply from one
to the other, at o
­ thers they dissolve slowly, rich reds and blues sliding into
one another, clouds and their shadows just barely separable. In some arrest-
ing sequences, the screen splits and mirrors, or satellite and photographic
image overlay one another, collapsing together disparate scales, materialities,
and topographies.

figure 3.2. Still from “Open Air,” Grayson Cooke collaboration with
Emma Walker, 2018. Courtesy of the artist.

Witnessing Ecologies 125

Launched by nasa in 1972 as the Earth Resources Technology Satellite
and renamed in 1975, Landsat is the longest continuous program of satellite
imaging of the planet, with its imagery used in every­thing from agriculture
to conservation to surveillance. The two currently active satellites, Landsat 8
and Landsat 9, rec­ord blue, green, and red light from the visual spectrum, but
can also sense in the infrared spectrum invisible to h ­ uman eyes. Their data
is freely available but produced by US government funding and so operates
within the ambit of US strategic priorities, with a vis­i­ble spectrum spatial
resolution of 30m and closer to 100m for infrared. In “Open Air,” Cooke’s lay-
ering of multiple scales enables continuities between the orbital satellite and
the macro video lens to bring into the terrain of the perceivable the climatic
and geologic pro­cesses that might other­wise evade the ­human. Without
narration or context, the Landsat images are more affective than repre­sen­ta­
tional: viewing them is not about decoding their content, but rather feeling
through the strangeness of watching change from above.
Despite their high-­definition clarity and our capacity to “read” them, t­ hese
images are only secondarily repre­sen­ta­tional: rather, they are testimonies
to nonhuman mediations, to vital pro­cesses of change in form, space, and
time. Scale collapses, eliding distinctions between pigment and pixel, paint-
ing and planet. Witnessing ­here is pure aesthetics: a registering of relations,
an enfolding of materiality and mediation. As the instrumental soundtrack
ebbs and flows, the artifice of the painting—­the ways it is not land—­become
both more evident and less significant. Its mediation makes it mutable; pro­
cesses (viscous dissolutions, searing wood fibers, bubbling and cracking
coatings) supersede the ­thing itself. So too in the Landsat images, where
their archival mattering as objects of scientific research falls away in the
meditative movement between images. Scale is pre­sent but cannot hold—­and
what is witnessed in the dissolution is that collapse in the nonhuman pro­
cesses of materialities fluxing in form. While Cooke works directly with the
remote-­sensing epistemic image, its indexical and informational functions
are systematically eradicated, made materially aesthetic by the movement
between satellite image and digital capture of paint, wood, dirt, and sand. As
with the h ­ uman experience of the Anthropocene and its climatic vio­lence,
claims to know are unanchored from their spatial references, made strange
and intimately nonhuman.
If the spatial scales of climate captured by Cooke are dizzying, their tem-
poral counter­parts can be weird and estranging. Time is the site of one of
its most confounding paradoxes: the urgent need for action now to confront
something that exists as an affective fact of catastrophic futurity. But Man is

126 Chapter Three

not used to thinking in time horizons that exceed Himself and yet demand
radical and immediate transformation in the pre­sent. Derangements of the
sensorium become estrangements of senescence. In a photo­graph document-
ing “Boiling Milk” (1999), a p
­ erformance that took place one morning in 1999
near Krafla, Iceland, the small pan of milk in Ilana Halperin’s gloved hands
barely touches the surface of the sulphur spring (figure 3.3). Crouched by the
side of the hundred-­degree lake, her arm extended from a narrow spit of land,
the artist seems almost a supplicant, her gesture one of ritual offering. Her
bright red raincoat pulled tight to the curve of her back, face emerging from

figure 3.3. “Boiling Milk,” Krafla, Iceland, Ilana Halperin, 1999

Witnessing Ecologies 127

the hood to focus intently on the task at hand, she stands out sharply against
the background blues and greys of the w ­ ater, mist and sky that dissolve into
one another as she waits for the milk to boil. By bringing together the swift
domestic act of boiling milk with the deep time of geothermal reactions,
David Farrier writes that Halperin “summoned an extraordinary confluence
of dif­fer­ent scales,” such that in “what appears to be a fleeting, even ­humble
exchange between ­human and geologic temporal ­orders, a deeply Anthropo-
cenic sensibility emerges.”46 Halperin describes her own work as examining
“geologic intimacy,” which Farrier sees as one form of the poetics of thick
time. Yet ­there is something ­else at work ­here, too, a witnessing of temporal
disjuncture, of nonhuman indifference to the scientific insistence on dating
and ­measurement.
Consider the embodiment of the encounter: wrapped protectively and
bent carefully at the knee, Halperin curls ­toward the hot ­water, only the skin
of her face exposed to the heat and stink of sulphureous w ­ aters. It is not only
the milk that feels the deep time of geothermal heat as its proteins coagulate
and separate, but Halperin herself. She is witness to the encounter, but also
entangled within it. In just a few short minutes, the composition of the milk
changes from cool to hot, beginning to steam like the air of the lake itself.
Located on the Mid-­Atlantic Ridge, where the North American and Eurasian
tectonic plates pull slowly apart, Krafla makes the slow drift of planetary
geologic change accessible to the ­human sensorium. In “Boiling Milk,” the
transfer of energies takes place across radically incommensurate time scales,
the millennial inching apart of the plates producing volcanic activity that
heats the lake and in turn Halperin and her milk. Milk—­perishable, biologi-
cal, life sustaining—­takes into itself the heat of infinitesimal geologic move-
ment. In this transfer of energy, scales collide but do not collapse. The milk
transforms, becoming other than it was through Halperin’s ritual gesture. But
this witnessing is not happenstance; it is deliberately enacted and carefully
framed, mediated by the photo­graph and its title into an image testimony of
the potential intimacy of time that far exceeds the h ­ uman.47 Reflecting on
such temporalities, Zylinska points to the emerging significance of photog-
raphy ­after the ­human, a phrase that refers not only to “the straightforward
material disappearance or conceptual overcoming of the ­human at some
point in the ­future . . . but also to the pre­sent imagining of that disappearance
as a prominent visual trope in art photography and other cultural practices.”48
Tracing the way in which photography a­ fter the h ­ uman confronts deep tem-
poral and spatial scales, as well as prob­lems such as extinction, is an essential
­political, ethical, and artistic question. In “Boiling Milk,” the ungraspability of

128 Chapter Three

deep time manifests in the dissolving background, the indistinct materialities
of the environment itself. And yet the body of the artist is not diminished or
made fragile but is attentive to its relations. Her body holds itself carefully,
her attention is a mode of care for the moment itself, for this seemingly
­simple event of holding a pan of milk above heat from beneath the earth’s
crust. Viewing this image, one can witness not only temporal scale, but also
an ethics of care t­ oward what that scale does, how it can be a site of connec-
tion and bring shared intimacy between ­human and nonhuman.
But the witnessing of temporal scale can also be radically unearthly. In
the final pages of Cixin Liu’s epic Three-Body Prob­lem trilogy, the impossible
scale of the life of the universe itself enters into a strange relation with two of
the novel’s h ­ uman protagonists, Cheng Xin and Guan Yifan. Suffice to say,
the details of how this speculative fiction progresses from the midst of the
Cultural Revolution to an interstellar f­ uture are beyond reckoning with h ­ ere.
But by the third book’s end—­spoiler alert!—­Cheng Xin and Yifan are in
a small space shut­tle orbiting a distant planet at the speed of light when they
are caught by the rippling wave left behind by a light-speed engine that curves
space to propel ships forward. Its rupturing of spacetime slows the speed of
light itself to a crawl, such that—as the laws of relativity require—­time passes
incredibly slowly for the two of them relative to the universe outside. Forced
to use hibernation technology to survive through the slow reboot of their
shut­tle’s computers, sixteen days pass for the pair while the planet experi-
ences more than e­ ighteen million years. Using ground-­penetrating remote
sensors, they are able to find a message left for them through the eons and a
doorway that leads into an artificial universe. It is a closed ecol­ogy of a single
cubic kilo­meter, suspended outside of time and from its vantage the two ­will
be able to watch our universe collapse into singularity and be reborn in a
new Big Bang. But the loss of mass from thousands of such micro-­universes
risks reversing the crunch of the g­ rand universe and instead expanding it into
endless, de­energized lifelessness. Rather than contribute to such existential
senescence, Cheng Xin and Guan Yifan give up their existence outside of
time and the promise of the birth of a new universe to live the death of our
own. They leave b ­ ehind a computational rec­ord of h ­ uman existence, and
a small globe containing two fish, w ­ ater, and a tiny artificial sun. To our
universe, they return the mass and its incipient energy that has made their
existence outside of time pos­si­ble.
It is, even by the standards of science fiction, an almost preposterous
projection, a conceit of ­human galactic endurance that belies our seeming
incapacity to do anything but destroy the richness of life on this planet. And

Witnessing Ecologies 129

yet it chases ­after something profound: a speculative pursuit of an infinite
relation between the h ­ uman and the nonhuman vastness of the universe.
Against the total mass of an expanding and contracting universe, what is
­human life, memory, existence? As an exercise of thought, it proposes that
the witnessing of humanity as a species depends inevitably and inextricably
on the nonhuman. On the one hand, this is an obvious claim—­what e­ lse
might be the other of such witnessing, that to which address is made and
response implored? But it is also a proposition that, for all its technowizardry
and speculative gymnastics, returns the ­human unavoidably to the question
of its relations to the milieu through which it moves and lives, w ­ hether at
the scale of the universe, solar system, planet, ecol­ogy, community, or self.
Witnessing the ­human at the limit point of the existence of the universe itself
means insisting on an offering to the ultimate nonhuman, life rendered down
to the necessarily flawed remainder yet insisting that some memory endure
into the emergence of a new space-­time.
Reflecting on the Abrahamic tradition of testimony, Peters writes that
“testifying has the structure of repentance: retroactively caring about what we
­were once careless of.”49 Already, testimony serves this function of repentance
in the Anthropocene: it marks and acknowledges the failures of government,
publics, and individuals alike, as well as the small victories of collective action,
of reparative meaning-­making. Scale can be operative, as well as relational.
Anna Tsing argues that scalability served a crucial role in the accumula-
tion of capital and in the spread of extractive modes of production across
the planet. Scalability describes “the ability to expand—­and expand and
expand—­without rethinking basic ele­ments.”50 Tsing points out the scalabil-
ity of the plantation was essential to its proliferation as a model throughout
the Amer­i­cas, just as scalability remains a fundamental princi­ple of con­
temporary business from social media platforms to fracking operations.
As C. L. R. James and more contemporarily Chris Taylor, Caitlin Rosenthal,
and Katherine McKittrick, among ­others, have argued, the plantation is
the model for the factory and for neoliberal conceptions of scaling produc-
tion up and down.51 In approaching this weaponization of scale, Tsing calls
for an attentiveness to nonscalability, “to the work of contingency and failure”
and the workings of “scalability in action.”52 By attending to registrations of
scale, nonhuman witnessing offers another means of thinking the nonscal-
able and the operations of scale between that which scales and that which
does not. Nonhuman witnessing of scale, then, is not solely about how we
grasp the ungraspable, but also how we intervene in the ways that scale is
put to use. If ecological vio­lence operates across spatial and temporal scales,

130 Chapter Three

then so too is ecological trauma bound up with scale. Nonhuman witnessing
of scale, then, brings us to one of the core problematics of ecologic crisis: its
traumatic disjunctures, cascades, and contaminations.

ecological trauma

Rising panic in the West over the “end of the world” often fails to recognize
already existing experiences of ruined lifeworlds. Nor do enough planned
or ­imagined responses to the climate emergency give heed to the ontologies,
epistemologies, and practical knowledges of t­ hose p­ eople who lived far more
sustainable lives before and despite settler colonialism. Ecological catastro-
phe has already been experienced by First Nations: the anthropogenic end of
worlds is, all too terribly, nothing new. Through vio­lence to knowledge, land,
and ways of living, as Kyle Powis Whyte argues, “settler colonialism commits
environmental injustice through the violent disruption of ­human relation-
ships to the environment.”53 Felling forests to graze ­cattle and grow crops,
introducing invasive species, diverting rivers and flooding valleys, flattening
hills and bifurcating mountains with highways—­the list of such disruptions is
endless. Nor, of course, are such ecological traumas confined to the past. En-
vironmental destruction, loss of traditional forms of community, and death
itself all flow from resource extraction, weapons testing and war, plantation
agriculture, and other forms of what Rob Nixon calls the “slow vio­lence” of
late capitalism, inflicted on the poor, oppressed, and dispossessed.54
Ecological trauma describes the injurious and ongoing effects at the level
of experience of the rupturing of relations that compose ecologies as living
and changing assemblages of more-­than-­human entities and pro­cesses. All
traumas target relations, severing encounters or events from the flow of ex-
perience and lodging ­those fragments in bodies as they go on, affecting and
affected by the world as it unfolds. But ecological trauma can be understood
as trauma that results from the rupturing of the relations that compose an
ecol­ogy, rather than t­ hose that enmesh a body within its world. Located at
the relational-­compositional level of the ecol­ogy itself, ecological trauma
echoes collective cultural trauma, but is differentiated by its insistence on
nonhuman entities and the situatedness of all ecologies and their relations.
Like trauma more generally, ecological trauma is found not in the vio­lence
that enacts a rupturing of relations but in how that rupturing carries through
into the f­ uture. Contaminating the unfolding multiplicities of experience
that animate an ecol­ogy with the past, ecological trauma is also haunted by

Witnessing Ecologies 131

f­ utures forged by ecological vio­lence. Th ­ ese are f­ utures diffracted through
trauma: the threat of collapse, stagnation, and death.
Ecological trauma encompasses but also exceeds what I and o ­ thers have
elsewhere called “climate trauma,” the traumatic rupturing of relations that
resides in the impossibility of the individual subject reckoning with the scale
of the climate crisis in its totality.55 Timothy Morton has argued that global
warming must be understood as a hyperobject, “massively distributed in
time and space relative to ­humans.”56 As a hyperobject, global warming can
only ever be grasped through second-­order abstractions such as graphs or
in localized effects, such as the slow drowning of Pacific islands, but such
repre­sen­ta­tions can at best be synecdochic of the incomprehensible totality
of the climate crisis. The prob­lem is that the object-­ness of the hyperobject
takes ­precedence over its local manifestations, systemic origins, and mean-
ingful strategies for its amelioration. The theoretical maneuver that trans-
forms climate crisis into a hyperobject is itself a violent mediation, one that
strips away agency, complexity, and relationality even as it evokes ­those very
­things within its theoretical tool kit.57 Framing con­temporary crisis within
the hyperobject paradoxically reasserts the Anthropos, even as the h ­ uman
is disavowed by Morton’s insistence on the separateness and inaccessibility
of objects in general. While it is true that “climate change” constitutes an
abstraction that necessarily contains more than can be grasped, reifying the
planetary scale risks replicating the annihilation of experiences of ecological
vio­lence, loss, grief, and renewal that takes place in more intimate, varied,
dispersed, and uneven ways. The ecological trauma of our age might be bet-
ter grasped as both one and many; always ecological traumas, plural, even
when it seems other­wise.
What I am proposing is a radical empiricist approach to ecological trauma
that recognizes rupturing and vio­lence as pro­cessual phenomena. Just as a radi-
cal empiricist approach to experience recognizes that the pre­sent is always lost
between unfolding pasts, which carry with them lost ­futures, and the tug of
potential ­futures, so too in more-­than-­human ecologies is the pre­sent always
unavailable to its own experience. F ­ uture collapse bears down on wounded
ecologies in the pre­sent, bringing itself into being through the continuance of
vio­lence in the form of trauma. Consider, for instance, the fires that have be-
come the norm in California, Brazil, and Australia, and the way their loom-
ing ever-­presence affects life even beyond the devastating damage to animals,
habitats, and homes.58 As Massumi writes: “This is the figure of ­today’s threat:
the suddenly irrupting, locally self-­organizing, systemically self-­amplifying
threat of large-­scale disruption.”59 ­There is, in a very real sense, an affective

132 Chapter Three

injury in the now from that which has and yet has not arrived. Always in
dynamic flux, even if that flux is entropic, the pre­sent of an ecol­ogy is always
missed in much the same way as it is for ­humans.
This traumatic relation to ecological crisis is not simply about the appar-
ent futility of action in this era of late capitalism but also the exponential
complexity of the prob­lem. Ecologies are not isolated systems; any bound­
aries placed upon them are always artificial and temporary. We ­humans
distinguish one ecol­ogy from another to make them sensible and addressable
but ­doing so is always a tactical m ­ easure: ecologies fold into one another, at
once one and many. Against the reified hyperobject of catastrophic climate
change, we might instead conceive of an endless complex planetary ecol­ogy,
si­mul­ta­neously composed of an infinite array of other ecologies. Ecological
traumas have thus ­shaped and continue to shape lives, communities, cul-
tures, and ecologies. Among ­those traumas is one that resides at the atomic
­organization of existence: nuclear war. Nuclear explosions occur within a
fraction of a second but leave radiation that contaminates, mutates, and ends
life into unfolding deep-­time f­ utures. Nuclear weapons and their catastrophic
damage constitute a vital site for engagement with nonhuman witnessing as
both an other-­than-­human registration of change and an aesthetic proj­ect
of ­human and more-­than-­human commingling. It is to the testing of ­these
paradigmatic technologies of world ending that I now wish to turn.

witnessing the nuclear

Yankunytjatjara elder Lester Yami called it a “black mist,” a thick cloud en-
veloping Adnyamathanha country, part of a huge swathe of Aboriginal land
in South Australia used for nuclear testing by Britain from 1953 to 1963.60 He
described his experience to the 1984 Royal Commission into the tests: “A big
bang—­a noise like an explosion and ­later something come in the air . . . [it]
was coming from the south, black-­like smoke. I was thinking it might be a
dust storm, but it was quiet, just moving . . . through the trees and above that
again, you know. It was just rolling and moving quietly.”61
Personally authorized by Prime Minister Robert Menzies and conducted
in secret, British nuclear testing in Australia took place on the Montebello
Islands (in 1952 and 1956), at Emu Fields in South Australia where Lester
Yami encountered the black mist (1953), and, most infamously, just to the
south of Emu Fields at Maralinga (1956–1963).62 Emu Fields was a particu-
larly disastrous choice: difficult to access by vehicle and prone to violent dust

Witnessing Ecologies 133

storms, it significantly increased the risk of wider nuclear contamination
due to irradiated dust carried on the wind. But the worst damage was done
at Maralinga, where the British tested seven atomic bombs and conducted a
series of even more disastrous “minor tests.”
In 1956, “Operation Buffalo” tested Red Beard and Blue Danube, pluto-
nium warheads with a destructive equivalent to the weapon dropped on
Hiroshima, with the smaller “Operation Antler” conducted the year a­ fter
(figure 3.4).63 ­These ­were followed ­until 1962 by a series of so-­called minor
tests, in which plutonium was scattered around vari­ous trial sites and blown
up to analyze shock waves, safety m­ easures, and radiation effects—­with devas-
tating consequences for Country.64 Ineffectual clean-­ups ­were attempted, with
two desultory efforts by the British in the 1960s and more comprehensively by
the Australian government in 2000, although costs ­were soon cut and their
effectiveness has been contested.65 In 2021, a study undertaken by scientists

figure 3.4. British nuclear testing at Maralinga, archival image

134 Chapter Three

at Monash University showed plutonium in the soil at Maralinga.66 ­Because
very l­ ittle was done to protect local communities, many suffer from high rates
of cancers and disease.67
Lester Yami described the immediate effects in visceral terms: “I cannot
remember how long we ­were getting sick and sore eyes and watery eyes and
diarrhea . . . vomiting and skin rashes . . . purtju, sore on the skin . . . I could
not see with both eyes.”68 The entire ecology—­people, ­water, vegetation, ani-
mals, dirt, dust, geology—­were directly exposed to radioactive contaminants
during the blasts and fallout, embedding radioactive ele­ments within the
ecosystem, passing them through bodies and life cycles.69 ­Here, the recollec-
tions of Nyarri Morgan, a young man at the time of the tests, are instructive:
“We thought it was the spirit of our gods rising up to speak with us . . . then
we saw the spirit had made all the kangaroos fall down on the ground as a
gift to us of easy hunting so we took ­those kangaroos and we ate them and
­people ­were sick and then the spirit left. . . . The smoke went into our noses,
and other ­people still have that poison t­ oday.”70
Maralinga was formally returned to its Traditional ­Owners in 2009, but
Country and its ecologies, the rich relations that bind ­human and non, re-
main contaminated, wounded, and traumatized (figure 3.5). For First Na-
tions in the settler state of Australia, such wounding of Country constitutes
an existential vio­lence. Trawlwulwuy scholar Lauren Tynan writes: “Coun-
try inhabits all relationality and is used widely across Australia to describe
how all land is Aboriginal land, Aboriginal Country; Country is agentic
and encompasses every­thing from ants, memories, h ­ umans, fire, tides and
research.” Vio­lence to Country needs to be understood as something far
71

more injurious and rupturing than what might be denoted by damage to “en-
vironment” within Western epistemologies. As Tynan continues: “Country
sits at the heart of coming to know and understand relationality as it is the
web that connects h ­ umans to a system of Lore/Law and knowledge that can
never be human-­centric.” Country is thus radically at odds with what Aileen
Moreton-­Robinson calls the “possessive logics” of white settler sovereignty,
that claim land as property and thus render it always potentially subject to
extraction and vio­lence.72
While the British authorities made efforts to mitigate the effects on white
farmers, the Aboriginal inhabitants of the region w ­ ere almost entirely ne-
glected. Aboriginal culture, history, lifestyle, and ceremony ­were not con-
73

sidered impor­tant by civic and military authorities. “The nomadic nature

of the desert p ­ eople, their traditional lifestyle and seasonal journeys for
hunting and ceremonial purposes was poorly understood,” writes historian

Witnessing Ecologies 135

figure 3.5. Signage at the former nuclear test site at Maralinga

J. D. Mittman. “The country ­there was regarded as ‘empty wasteland,’ in line

with the l­egal doctrine of terra nullius, a Latin phrase meaning ‘empty land’
or ‘land belonging to no-­one.’”74 This ­legal doctrine of terra nullius articulated
the land as lacking property relations. First Nations ­peoples might live on the
land but had not undertaken improvements legible to the colonizers as demar-
cating possession. Through this doctrine, First Nations ­were dispossessed of
the land so that white settlers could possess it, a move that not only stripped
them of property ­under Crown law but also assigned them as belonging to
the state of nature.75 Just as terra nullius was retroactively applied to authorize
the theft of land ­under settler colonialism in what became known as Aus-
tralia, so too was it used to justify the new nuclear colonialism. For a British
Empire in disarray, Maralinga was an acceptable sacrifice zone, an empty
wasteland, populated by ­people still counted among the flora and fauna of
the nation, that could be readily transformed into an extraterritorial zone
of incision in which its inhabitants, having lived ­there for thousands of years,
­were suddenly rendered illegal.
This construction of a zone of absence and excision echoed that of the So-
viet ­Union in Kazakhstan and Siberia, and the French and the United States
in the Pacific. In this sense, Maralinga reproduced the nuclear colonialism
emerging across the United States and the Pacific.76 Unsurprisingly, the most
targeted ­were and continue to be Indigenous ­peoples and lands. Operating in
contexts of radical power asymmetries, nuclear colonialism depends on ma-
terial and discursive maneuvers, generating economic dependencies on the
one hand while constructing lands and ­people as permissible objects of

136 Chapter Three

vio­lence on the other.77 In the Marshall Islands, the United States detonated
thermonuclear weapons o ­ rders of magnitude more destructive than t­ hose at
Hiroshima and Nagasaki, leading to the displacement of p ­ eoples from their
traditional homes and to a horrific legacy of birth defects across the islands.
As the opening pages of this chapter made clear, ­those same islands are
now among the most at-­risk places on the planet for the rising sea levels of
global warming, in yet another tragic knot in the entangled history of nuclear
war and climate science. Perhaps unsurprisingly, drone histories occur ­here
too, with the Operation Kamikaze remotely pi­loted drone munition tests
conducted in the shadow of C ­ astle Bravo, at fifteen megatons the largest test
ever conducted by the United States. Nuclear colonial discourse i­magined
the Pacific as isolated islands and empty seas. But nuclear testing helped
mobilize a renewed Oceanic p ­ olitical activism that insists on a relational
­political ontology, founded on the connectedness of sea and islands, ­peoples
and fish.78 Aboriginal land in Australia was similarly subject to expropria-
tion and excision. Pitjantjatjara Anangu from Ooldea w ­ ere forcibly removed
from Country to a purpose-­built settlement in Yalata more than 150km to
the south. The test site was renamed Maralinga by the Australian authorities,
the word for “thunder” in the Garik language, chosen for its fit to the nuclear
vio­lence that would take place ­there.79 Maralinga was excised from the civil-
ian ­legal order and made inaccessible to its Traditional ­Owners, a redoubled
denial of sovereignty. Weapons testing began shortly ­after.
Detonated in the first test of Operation Buffalo at Maralinga, the bomb
known as Red Beard used nuclear fission, a ­process that exploits the desire
of unstable atoms to achieve an impossible equilibrium (figure 3.6). ­Under
the intense force of neutron bombardment or chemical explosion, an atom
of an unstable isotope—­a mix of Uranium-235 and Plutonium-239 in the
case of Red Beard—­splits in search of stability. As it splits, energy is released
but so too are smaller nuclei—­fission products—­that strike other unstable
atoms, changing their atomic structure. Two or more neutrons, subatomic par-
ticles within the nucleus held in check by electromagnetic force, get ejected.
Ejected neutrons disrupt other, already unstable isotopes. Uranium-235
becomes Uranium-236, an even more unstable isotope that splits again,
releasing more energy and more fission products. More atoms are struck
and split; more energy is released. This is the chain reaction that generates
the catastrophic explosion of the atom bomb, an urgent hunt for stability
that produces nothing but more splitting, more energy, more instability.80
Nuclear bombs are designed to explode well above the ground, maximizing
both the force and radius of the blast: this is what produces the spectacular

Witnessing Ecologies 137

figure 3.6. Explosion
of a Red Beard warhead
on September 27, 1956

mushroom cloud, which reached a height of almost twelve kilo­meters with

the first test at Maralinga. Heat is so intense that it turns bodies to ash, melts
metal and concrete, and, at Maralinga and Emu Fields, transformed the silicate
and sand of stretches of desert into a crust of glass.81
Glass becomes an aesthetic medium for the nonhuman witnessing of
nuclear vio­lence in the hands of the artist Yhonnie Scarce: brittle yet tough,
capturing light yet also diffracting and refracting it, rigid when cool yet
­shaped by the breath of the glassblower when molten. Belonging to the Ko-
katha and Nukunu ­peoples whose country forms part of the Maralinga ex-
cision zone and born in the military town of Woomera, Scarce’s personal
and familial history is bound to settler colonialism and to its entanglements
with militarism. Her work is intensely ­political: an unflinching critique of
past and pre­sent settler vio­lence, but also a cele­bration of the resilience and
endurance of Aboriginal p ­ eople and of Country in the face of the colonial
logic for elimination. Repeatedly returning to the interconnections between
the classification and dispossession of First Nations land and scientific and
military testing, Scarce’s work is unified by an unrelenting aesthetic of aus-
tere grayscale, battered found objects, stark medical equipment, and spare

138 Chapter Three

architectural structures and installations. While Scarce works with a range
of media, glassblowing is at the core of her practice, a visceral and difficult
art that demands much of breath and body. The subject of a major survey
exhibition at the Australian Centre for Con­temporary Art in Melbourne
and at the Institute of Modern Art in Brisbane in 2021, Scarce’s works are
held at the National Gallery of Australia, the National Gallery of Victoria,
Art Gallery of South Australia, and by other galleries and private collectors
around the world. While her work always addresses colonialism and its vio­
lence, the subject ­matter ranges from ­family history to medical experimenta-
tion to the legacies of nuclear testing. In three of her major works—­Thunder
Raining Poison (2015), Death Zephyr (2017), and Missile Park (2021)—­Scarce
reveals the potential for a glassy aesthetics of nonhuman witnessing to nu-
clear vio­lence.
Thunder Raining Poison (2015) and Death Zephyr (2017) are formally and
thematically similar works (figures 3.7 and 3.8). Both are composed of thou-
sands of blown glass yams suspended from the ceiling in arrays that reference
the atmospheric forms produced by nuclear tests at Maralinga. Th ­ ere are
critical differences between the two works: Thunder Raining Poison cap-
tures the instance of detonation, yams arranged in a teardrop formation and
mostly made of clear glass, interspersed with black and blue; Death Zephyr
examines the spread of contaminated particles in the aftermath of the blast,
its mix of black and clear yams hang in a swirling current across the gallery
space, a material enactment of Lester Yami’s “black mist” over Country. Una
Ray observes that “the yam and other ‘bush tucker’ plants such as the bush
banana and bush plum, along with their associated Tjukurrpa (Dreaming)
sites are impor­tant subjects for Aboriginal artists, particularly w
­ omen who
traditionally held the comprehensive knowledge of the regularly harvested
and managed bush gardens across Australia.”82 Composing yams into forms
of nuclear explosion and dispersal materializes the violation of life by such
vio­lence, its assault on the nonhuman that sustains the ­human. Not only is
sand blasted into nonlife, ­these works suggest, but also the means to sustain
life affectively and discursively signified in the alchemy of yams become in-
edible. Distended and distorted, the yams are reminders too of “eviscerated
organs or exhumed physical evidence in the prosecution of war crimes,” as
art critic and Bundjalung and Kullilli man Daniel Browning puts it.83 Both
Thunder Raining Poison and Death Zephyr are reminders of the limits of the
forensic practices of the state to understand the consequences of the tests
conducted on Aboriginal land precisely ­because life is ecol­ogy: yams, sand,
land, bush animals, p ­ eople, are bound together by Country, a relationality

Witnessing Ecologies 139

figure 3.7. Thunder Raining Poison, Yhonnie Scarce 2015. Courtesy of Yhonnie
Scarce and THIS IS NO FANTASY.
figure 3.8. Death Zephyr, Yhonnie Scarce, 2017. Courtesy of Yhonnie Scarce and
THIS IS NO FANTASY.

that is just as crucial to the ecol­ogy as the electromagnetic forces that hold
neutrons in check are to stability at the atomic scale.
Unstable isotopes are radioactive: they contain an un­balanced combina-
tion of neutrons and protons in their nucleus, which typically means too
many neutrons. By shedding extra energetic particles, t­ hese isotopes “decay”
into other particles, becoming more stable and less radioactive but releasing
nuclear radiation in the ­process. When a nuclear bomb is detonated, radio-
active particles are dispersed by the explosive force, attaching themselves
in turn to other particles. This is nuclear fallout: the irradiated particles of
weapon debris and dust that are carried on the wind, as Death Zephyr reminds
us, before they fall to earth. In their fall, they can attach and deform more
particles and the cells that make up life, such that stones, plants, animals, and
­people become carriers of contamination, nonhuman and doomed witnesses
to nuclear catastrophe.
Some of the most devastating effects at Maralinga w ­ ere not the bombs
themselves, but the “minor tests” involving the detonation of scattered pluto-
nium and other radiation “safety” experiments. Depending on the half-­lives
of the isotopes involved, radioactive contamination might be pre­sent for
minutes, days, or years.84 Radioactive contamination can have enduring ef-
fects: making soil and ­water poisonous, producing cancers and miscarriages,

Witnessing Ecologies 141

and deforming life at its most basic workings. Yet radioactive decay also con-
stitutes a material witness to the unthinkable force of the nuclear explosion, a
nonhuman registration of the impossible vio­lence that it produces—­and that
produces it. Radioactive contamination and its decay are themselves both
nonhuman witnesses, and what must be witnessed.
While foreign wars and colonial “exploration” are commemorated across
Australia, the vio­lence done to its original inhabitants remains po­liti­cally
contested and largely unmemorialized. Official memorials function as modes
of commonplace witnessing, generating the shared meanings through which
certain p ­ olitical knowledges and identities are reproduced and o ­ thers are
elided or erased. Missile Park, commissioned to accompany the 2021 survey
85

of Scarce’s work, offers a countermemorial to the vio­lence of nuclear testing

and settler colonialism. Blasted raw and roughly painted black, three corru-
gated metal sheds echo the structures common to Maralinga and Woomera,
but also to Australian settlement more generally (figure 3.9). One shed invites
entry into a near-­dark space, lit only by the gallery light that finds its way
through the gaps of corrugation, in which a s­ imple t­ able holds twenty bush
plums blown from black glass (figure 3.10). Gunditjmara and Torres Strait Is-
lander artist and curator Lisa Waup calls t­ hese sheds “containers of trauma.”86
­Here in the dark, in the hurried tin-­shed architecture of the expansive proj­ect
of Australian settlement, the bush plums attest to the memory of ecologies
of life cleaved by vio­lence. Ray describes the artist’s work as “a composition
between hand and breath, the alchemical, elemental p ­ rocess of glassblowing
neatly fuses the maker to her material and her métier to meaning.”87 Missile
Park coalesces ­these elementalities, finding in their ecological relations a
host of evocative tensions. Bush plums of black glass capture breath, that
most ephemeral and basic gesture of fleshy life, but they also hold in their
material memory the pollution of air, the violent contamination of the bomb.
They bear nonhuman witness to the scales of incommensurate temporali-
ties of the nuclear: the split second of the explosion, the long half-­life of
plutonium, the irradiated endurance of Country. H ­ ere is ecological trauma,
witnessed through an ecol­ogy of pro­cesses, objects, and milieus that only
come to ­matter through their relational composition.
­There is a dark irony in the witnessing capacities of glass, one that extends
beyond Scarce’s art and the silica seared into trinitite at Maralinga to the
failed use of vitrification in the latest attempts to clean-up the radiation at
the test sites.88 Developed by the US com­pany Bastelle, in-­situ vitrification,
or isv, uses electricity to immobilize plutonium and other unstable isotopes
in glass-­like blocks that can keep them safe for hundreds of thousands of

142 Chapter Three

figure 3.9. Missile Park exhibition installation, Yhonnie Scarce, 2021. Courtesy of
Yhonnie Scarce and THIS IS NO FANTASY.

figure 3.10. Missile Park interior, Yhonnie Scarce, 2021. Courtesy of Yhonnie
Scarce and THIS IS NO FANTASY.
years, at least in theory. In practice, isv was difficult to implement and not
always fit to task, the government department overseeing the p ­ rocess failed
to establish clear criteria—­and, surprising no one familiar with the history
of the state’s treatment of Indigenous ­peoples in Australia, once costs grew,
vitrification was abandoned in f­ avor of exhuming and burying the waste.
Glass as a failed medium of remediation testifies to the unyielding nature of
nuclear radiation, but also to the per­sis­tent coloniality of settler politics, to
the legacies of who counts as ­human and who does not. As a byproduct of
nuclear testing and as a failed mechanism for decontamination, vitrification
is a ­process of mediation: silica into glass, by way of the intense applications
of energy. Its vio­lence is not inherent, but contextual. Through the breath of
the glassblower, vitrified silica becomes intimate and lively: a rich ecol­ogy
of country, life, fruit, vegetable and yet still an ecol­ogy deeply wounded by
the vio­lence of war and settlement. Glass yams and bush plums distill the
ecological traumas of nuclear testing at Maralinga, the stuff of life rearranged
into the mushroom cloud and its dispersal and memorialized in the tin sheds
of the test sites.
What it means to witness such ecological trauma looks very dif­fer­ent
within the accepted bounds of historical witnessing, particularly in the of-
ficial form it took in the Royal Commission into British Nuclear Tests in
Australia, chaired by Jim McClelland. While the commission sought to ac-
count for the health impacts on Aboriginal ­people and heard the testimony
of Lester Yami and ­others, its principal focus was the irradiation of Austra-
lian ser­vicemen, the safety precautions implemented by the British, and the
nature of the agreement between Australia and its imperial overlord. From a
cultural standpoint, Yami’s black mist is surely its enduring figure, one that
finds a glassy counterpart in the art of Yhonnie Scarce. Nuclear activism and
public pressure in the 1980s did much to make Maralinga and Emu Fields
vis­i­ble to the wider Australian public, and in 2009 almost all the lands of the
excision ­were returned to their Traditional ­Owners. But the Royal Commis-
sion, the failed cleanups that followed, and the narrow inquiries from vari­ous
departments and committees function as stark reminders of the impossibility
of such organs of the settler state working against its fundamental invest-
ments in militarism and the denial of Indigenous sovereignty. Within such
confines, the capacities of witnessing are bound not only by the necessity of
speaking but also by ­legal norms and parliamentary terms of reference.
A more expansive witnessing must be sought elsewhere, in the poetry of
Indigenous writers such as Oodgeroo Noonuccal, Lionel Fogarty, and Natalie
Harkin, and even in the inventive research of scientists, who have exposed

144 Chapter Three

the radioactivity embedded into particles of the land. In a poem written for
the cata­logue of the 2021 survey of Scarce’s work, Harkin writes:

mine and refine this float of molten

landscape raw silica-­sand and
limestone sites slice and stirred
and hot-­shop forged we
witness excavation of targets and
melts a redaction of origins of
lives of lands
see what a breath can do89

An intimacy emerges in Harkin’s words that bears a certain resemblance to

that between Ilana Halperin and the deep time of volcanic heat in “Boiling
Milk.” Yet h ­ ere the relation is one of breath, an intensive yet ephemeral bond
between the geological, the nuclear, and the fragility and force of h ­ uman life,
the glassblower and her glass. As with all the aesthetic works examined in
this chapter, Scarce’s are of course instigated by h ­ uman subjects and wrought
by ­human hands. Yet they dance with the nonhuman in equally inseparable
ways, from their insistent materiality to the nuclear vio­lence they reference
to the settler histories of dehumanization and the nonhuman classification of
Aboriginal p ­ eoples. Glass suspended in air coalesces an instance of the most
radical ecological vio­lence pos­si­ble: a vio­lence torn from the fracturing of the
atomic structure of ­matter itself, a vio­lence that holds the potential not only
to erase the ­human but also to destroy all but the most defiant, hidden, and
per­sis­tent forms of life. And yet glass also captures the endurance of breath,
the variability of life, and that most fundamental of mediations: photons of
light passing through a medium. Life in all its relations, located in Country
and food, in air and stone and ­water, is not simply indexed or represented in
the art of Yhonnie Scarce: it is pro­cessually pre­sent, materialized as a witness-
ing ecol­ogy come to life out of the urgent need to witness ecological trauma
and its continued imbrication with the settler state and its militarisms.

wounding

In The Logic of Sense, Gilles Deleuze recognizes that futurity resides at the
heart of the event and its relation to h
­ uman expression. The event is “always
and at the same time something which has just happened and something
which is about to happen; never something which is happening.” 90 This

Witnessing Ecologies 145

simultaneous doubling and splitting of that which has just happened, or the
­actual, and that which may be about to, or the virtual, constitutes a kind of
rupturing: a wound. While this wound is not corporeal in the same way as a
cut or broken bone, it is nonetheless bound up with sensation, with the bodily
experience of the event—­and, crucially, with its separating into a distinct sym-
bolic ele­ment in the realm of pure expression. Or, to put this in more distinct
terms, the wound is the rupturing of virtual into ­actual, w ­ hether in experi-
ence, thought, or expression. Deleuze’s choice of the wound as a ­metaphor is
telling: it draws par­tic­u­lar attention to the vio­lence inherent in the limiting of
potential that occurs in any given ­thing becoming ­actual. To call this a wound-
ing suggests that all intersections of the virtual and the ­actual, all forms of
creation—­whether life-­living or art-­making, h ­ uman or other­wise—­are inex-
tricable from injury, from a cleaving of one t­ hing from another. H ­ ere, then,
is the dynamic of violent mediation at the level of expression itself, of the
coming into living of life: a transformative mediation from one state to an-
other that cannot but cut off, leave ­behind, exclude, or ruin, even as it makes
pos­si­ble the new and the other­wise from which the good might flourish.
This relation between life and expression circulates in Deleuze’s enigmatic
final essay, “Immanence: A Life . . . ,” in which he dwells on the two terms
of the title and their relation to each other.91 He shows how immanence
neither refers to an object nor belongs to a subject but is immanent only
to itself. The second term—­a life—­captures something at once instinctively
understood and yet very difficult to pin down precisely. It is life as an indefinite
­thing—­not this life or that, not my life or yours, but rather a life, indefinite and
potential, indeterminate yet somehow also composed of singularities. At once
the many and the one, to borrow from the pluralism of William James that
so influenced Deleuze.92 This multiplicity in the midst of singularity shares
much with what Mario Blaser and Marisol de la Cadena call a “world of
many worlds,” a pluralism that is not only experiential but po­liti­cally onto-
logical.93 Drawing from her deep anthropological engagement with Andean
Indigenous communities and politics, de la Cadena insists on the p ­ olitical vi-
brancy and active agency of earth-­beings, a necessarily truncated translation
of mountains, lakes, rivers, and other existences that, for Andeans, “blurred
the known distinction between ­humans and nature.”94 The pluralistic politics
that flows from recognizing the standing of earth-­beings is a question that
Iw­ ill return to in the coda of this book, but h ­ ere I want to draw a relation
between this pluralism and the pro­cessual emergence of actualities from
within the sheer stuff of existence. Only in wounding does a life become
the life of a subject or object, or even a milieu. “A life contains only virtual,”

146 Chapter Three

writes Deleuze. “It is made up of virtualities, events, singularities.”95 Writing
with Guattari, Deleuze describes the plane of immanence as “the plane of
Nature, although nature has nothing to do with it, since on this plane t­ here
is no distinction between the natu­ral and the artificial.”96 Such an approach
echoes the relationality that Tynan describes as “premised on a truth that ‘all
­things exist in relatedness’ and whilst this is a naturally occurring princi­ple of
many Indigenous worldviews, it is a princi­ple that is sustained and strength-
ened through practice.”97 Without eliding or erasing the differences in t­ hese
epistemological standpoints, what emerges across them is a recognition of
the mutuality and relationality that makes existence and experience pos­si­ble.
It is within this sense of the mutuality of all existence that the complex
and necessary nature of Deleuze’s wound becomes clear. The wound is not
simply to be suffered or endured; it is not an injury with moral overtones.
It is incarnated in life as a state of ­things, as corporeal, temporal, and ex-
periential, yet it leads into that indefinite, elusive plane of a life precisely
­because the wound is a “pure virtuality on the plane of immanence” even as
it is actualized in par­tic­u­lar bodies.98 Put differently, the wound is a kind of
passage, the means by which the plane of potential takes place, something
felt as loss but also as always newly opened. Deleuze writes: “My wound ex-
isted before me: not a transcendence of the wound as higher actuality, but its
immanence as a virtuality always within a milieu.”99 He might have chosen
another word—­cut or break, perhaps—­yet this choice of wound (une bles-
[103.158.136.64] Project MUSE (2024-02-06 19:57 GMT) University of Edinburgh

sure in French) m ­ atters. It is a reminder that embodiment is necessarily both

a rupture from what might have been and the becoming-­in-­the-­world of a
host of lively potentials. Naming this a wounding calls attention to an ethic
of care; it evokes both fragility and resilience. It suggests that how wound-
ing happens, what form and movement it takes, ­matters both for a life and
for life lived. This has consequences for conceptualizing and responding to
ecological traumas: it suggests that wounded ecologies are also a wounding of
the relation between experience and expression, between life and aesthetics,
between existence and becoming. As a response to ecological traumas, the
pro­cessual and unfinished nature of nonhuman witnessing seeks to make this
constitutive wounding sensible—­however fleetingly.
Wounding, unsurprisingly, also occupies a central figural position in the
study of trauma in the humanities. As literary theorist Cathy Caruth insists
in what has become a canonical formulation, trauma “is always the story of
a wound that cries out, that addresses us in the attempt to tell us of a real­ity
or truth that is not other­wise available.”100 This wound, like that of Deleuze,
occupies a doubled position: marking both body and mind yet unknown

Witnessing Ecologies 147

to ­either. Only in the belated arrival of a voice from within the wound—­a
return of the wounding in the form of trauma—­does knowing become pos­
si­ble through paradoxically testifying to its own impossibility. As in Deleuze,
the wound of trauma is not a ­metaphor but rather the living embodiment
of a relation of rupture between experience and expression. Trauma studies
shares with Deleuze this recognition that the rupturing of the wound occurs
at the most basic forms of relation: its rupturing ruptures the planes of ex-
perience and expression. More, that this rupturing is at once destructive and
creative—it closes off or eliminates potential even as it produces the a­ ctual.
Ecological trauma also possesses an affective dynamic, what Massumi
describes as the participation of the ­actual in the virtual and the virtual in the
­actual.101 In ecological trauma, as in all traumas, that enmeshment of virtual
and ­actual is radically constrained by the foreclosure of potential and mean-
ing. This foreclosure is characteristic of traumatic ruptures to experience:
the disjunctive wound becomes a discordant, damaging feedback loop.
Divorced from prob­lems of scale and the necessity of ­human subjectivity,
the wound in the virtualities of a life works at a remove from the prob­lem of
shifting between the personal and the collective that trou­bles so much work
on trauma in the humanities. Nonhuman witnessing addresses trauma at this
vital plane of existence b ­ ecause it resists the temptation to wait for trauma
to arrive in the ­human. It attends to traumatic ruptures within life itself and
their material entanglement of bodies of all kinds, from t­ hose of p ­ eople to
rocks, storms, and nonhumans animals.

ecologies of witnessing

Nuclear testing, catastrophic climate change, the ecological traumas of mili-

tarism, capitalism, and colonialism: ­these are planetary phenomena that
nevertheless have consequences that are at once intimately embodied and
collectively targeted. Ecological catastrophe is martial as well as capitalistic. It
is not an accident of history that required the Marshallese poet Kathy Jetñil-­
Kijiner to write of both rising w ­ aters and the deformations of life from
nuclear contamination, of empty flesh sacs born in place of babies. Nor
can the intensifying militarization of the oceans around China, a p ­ rocess
in which Australia is a small but active player, be extricated from the failure
of my country to act in response to the pleas and tears of Oceania’s leaders.
Witnessing ecologies—­both as a p ­ rocess of registering ecologies as sen-
sible assemblages and as the composition of witnessing through the formation

148 Chapter Three

of relations—­are not an antidote to ecological trauma, but they are an open-
ing t­ oward the potential for repair. Nonhuman witnessing as an ecological
relation generates transversal vectors between ele­ments within ecologies: wit-
nessing as ecological, produced in the dynamic relation of systems, but also
the witnessing of ecologies, through the aesthetic registration of their rela-
tionality as ­process in time and space. Attending to the nonhuman witnessing
of ecologies and ecological relations continually returns us to mediation at its
most fundamental: the transfer and translation of energies from one medium
to another, a ­process that can be harnessed by technoscientific instruments
but also far outstrips them. Technoscience ­doesn’t provide the only sensors
and communicators of significance, even if we understand ecologies in a nar-
rowly biological sense. Ecologies abound with sensing and sensors; animate
and inanimate bodies alike take part in the dance of mediation, interrupting
and modulating its flow. Some of t­ hese mediations are certainly violent in the
sense that this book has articulated. But seemingly destructive forces within
ecologies are not alien to life, as even a cursory knowledge of evolution, the
seasons, or the role of fire in the flourishing of certain plants would attest.
Ecological trauma, as a distinctive taking shape of that wounding to a life
described by Deleuze, is not a “natu­ral” phenomena, but one inextricable
from the vio­lence that h ­ umans do, and certain h
­ umans far more efficaciously,
deliberately, and comprehensively than ­others. But while Man, the Anthro-
pos, is surely responsible for the era that now bears his name, the forms and
practices of knowledge-­making that accompanied that transformation to
what counts as life itself need to be wrested away.
Nonhuman witnessing, pursued in this chapter through the technosci-
entific apparatuses of remote sensing and in the aesthetic interventions of
ecologically inflected art, makes pos­si­ble the witnessing of ecologies and
ecologies of witnessing that displace the h ­ uman yet neither disavow respon-
sibility nor refuse the address of wounded ecologies. In the next chapter, I
pursue this collective and more-­than-­human concept of trauma alongside
both the violent mediations and machinic affects of more technical systems
with the aim of showing how nonhuman witnessing helps elucidate more
quotidian experiences of vio­lence, loss, and absence in digital cultures. This
return to the h ­ uman provides the ground for the coda that follows, where
I take up the question of the politics of nonhuman witnessing more explic­
itly. Nonhuman witnessing is neither separate from nor prior to politics, but
rather contains within it the latent potential for ­political relations that might
other­wise fail to cohere, animate, or confront the ­human with the necessity
of response.

Witnessing Ecologies 149

chapter
four

witnessing
absence
first absence:
the execution of james foley

On August 19, 2014, the Islamic State in Iraq and Syria (isis) uploaded a
video titled “A Message to Amer­i­ca” that depicted the beheading of the
kidnapped journalist James Foley. Despite being swiftly pulled by YouTube,
the video and gruesome stills from it circulated on social media, news sites,
web forums, and shock galleries. Shot in crisp high definition, the video was
slickly produced and professionally edited. Deviating from the grainy foot-
age and awkward staging of executions filmed in A ­ fghanistan or Iraq in the
years a­ fter 9/11, it had a consciously con­temporary aesthetic. ­After a long
message addressed to President Obama, Foley appears on his knees, dressed
in orange. ­Behind him is a black-clad and masked executioner, around them
blasted desert and stark sky. The beheading itself lasts only ten seconds;
yet the moment of death is not shown. It occurs off-­camera, dis­appeared
in the digital cut. A knife saws, but t­ here is no blood. Th
­ ere is only the body,
the head. The cuts shown are staged, experts say. Death itself is absent, but
radically so—­despite not occurring on camera, it is everywhere in evidence.
Reflecting on the recurrent moment of the cut in photography, Kember and
Zylinska ask what might it mean to cut well, to cut in a way that entails a
vital, creative ethics.1 But what might it mean to cut poorly, to cut the clumsy
cut? For the digital cut to cut out the cutting of the body? In this gruesome
portrait of death without the moment of d ­ ying, t­ here is an absence within
an absence—­yet one that has a presence in the digital contagion of traumatic
affect. Perhaps the killing was botched, the blow of the sword too weak or at
the wrong ­angle.
Or perhaps the cut was too bloody, too grotesque. ­After all, the video’s pur-
pose was not only to incite shock, but also to recruit—to catch the disaffected,
the angry, the alienated and offer purpose through blood and vio­lence.2 This
is a video that aims to traumatize, but also to speak to and through trauma.
As such, it is perhaps best understood as an image of digital war that exempli-
fies, as Andrew Hoskins and Shona Illingworth write, “a shift in the trauma of
civilians from a memory of the past to a perpetual anticipation of the threat
of the ­future, subjecting increasing numbers of ­people to unending physical
and psychological incarceration in a traumatizing pre­sent.”3 To watch such a
­thing must be brutally visceral—­but I d ­ on’t know, I h ­ aven’t seen it. Like the
deferred moment of death itself, I held back from an active participation in
its affective economy and have encountered it only in stills and second­hand
accounts, mediations of a mediation. Yet my r­ esistance to seeing the video
does not prevent its forcefulness from making its mark: t­ here is an urgent
affectivity in its absence, even now.
Despite its wide circulation, the beheading of James Foley—­and l­ater of
Steven Sotloff and ­others—­produced a radical absence. Its absence resided
in the anxiety it engenders, the anxiety of potentially encountering the visual
force of war’s vio­lence. An errant click, the wrong news article, a social media
post that slips through the controls instituted by Twitter or Facebook—to
encounter ­these videos would be so easily done, a s­ imple digital stumble or
the caprice of an algorithm. Crowding virtuals of affect, accumulating po-
tentials on the verge of becoming ­actual: an affective-­traumatic atmosphere.
Brutal vio­lence had infected the everyday of the digital. Who could say how
or where it had proliferated? The my­thol­ogy of digital permanence, the no-
tion that w­ hatever words or images of ourselves find their way on online stay
­there, resonated with the video’s disappearance. It was always potentially
appearing, even when it never arrived. Already testimonial texts that bear
witness to p­ olitical murder, such videos circulate in search of co-­witnesses,
dependent on news values, browsing habits, and algorithmic recommenders.

Witnessing Absence 151

Fragmented terrains of media seemed suddenly not simply a prob­lem for
trust and accountability, but a risk to bodily integrity.
Once, broadcast networks might simply have colluded to conceal the
video, but its indefinite circulation encouraged hosting on the websites of
establishment media. If it was out t­ here, it should be h ­ ere, or so the think-
ing went. In the early 2000s, before social media as we know it and with the
digital ecol­ogy far less developed and vibrant, watching the video of another
execution—­the journalist Daniel Pearl—­had required sustained pursuit
through web forums and the glitchy ­predecessors to YouTube. Not so for
James Foley. Even with the object absent, secreted from viewing, its traumatic
affect still leaked, oozed, and pooled.4 Even in absence, t­ hese videos accumu-
lated affective force, so that not watching did not prevent encounter: what
was encountered was their looming lack of presence. Carriers of an affective
contagion, more than a stand-in or symbol of the pos­si­ble disturbances en-
gendered by the digital, the videos are traumatically affecting even without
being seen. Like the body of the terrorist a­ fter 9/11, their passing-by reshaped
the surfaces of fearful bodies.5 ­Here was terror, potentially: the lone wolf
video, stalking the algorithmic hinterlands of platform capitalism. Distant
war on the verge of becoming intimate, of demanding witness.
Weeks ­after, Sydney and Brisbane woke to media blasting stories of dawn
raids capturing suspected terrorists, footage of police storming ­houses,
and breathless excitement from politicians and pundits. Random public
beheadings w ­ ere planned, the prime minister of the day quickly claimed,
backed by anonymous leaks from the Australian Federal Police and displays
of a seized sword.6 What happened in the desert was in our midst, or so it
seemed. In the iconography of the sword and the references to beheading,
the raids resonated with the circulating videos, with mediated vio­lence al-
ways on the verge of encounter. They amplified fear, made manifest in bodily
sensation the possibility that distant vio­lence could appear on any screens,
anytime. It d ­ idn’t m
­ atter, h
­ ere in Australia, that the sword was revealed to
be plastic, that its ­owner was Shiite and thus anathema to isis.7 The very
connectedness of the con­temporary world, the ever-­presence of the digital,
sharpened into an affective threat: violent mediation made manifest in the
digital quotidian. To have this infiltrate the h ­ uman sensorium, to have been
confronted with radical absence in the digital’s capacity to transmit violent
and traumatic affect, was to shift one’s affective relation to the digital itself.
It was to risk being forced into a witnessing relation, one latent within the
nonhuman infrastructures of the digital systems yet invisible ­until it was
too late.

152 Chapter Four

radical absence

It is not only the beheading of a kidnapped journalist. Disappearances keep

appearing in the digital sphere: an airplane vanishes into the sky; friends
learn someone has died when Facebook “memorializes” their page; sacred
sites are destroyed by a mining g­ iant in search of iron ore. Each event is
dif­fer­ent in its particularity but shares an affective architecture: it is a mani-
festation of absence that is nonetheless vitally pre­sent. This radical absence
throws ­those who encounter it into a witnessing relation with the felt force
of disappearance. Radical absence occurs when this force surrounds some-
thing that cannot but fail to appear, yet in its nonappearance entangles the
­human with nonhuman infrastructures of mediation and circulation. Such
absences are not exceptional, but rather part and parcel of the digitality that
constantly and constitutively entangles everyday life. While encounters with
radical absence are not rare, coming into contact with their mediated traces
and resonances can possess an unexpected intensity. Encounters with radical
absence constitute both a witnessing of absence and the absence of witness-
ing: a paradox that is made manifest and material through the inescapable
presence of nonhuman agencies and infrastructures. Radical absence be-
comes pos­si­ble through the constitutive affectivity of digital mediations. As
Richard Grusin writes, “The affective ele­ments of our interactions with every-
day media technologies work both socially and po­liti­cally.”8 Radical absence
arises not from disconnection but from an abundance of connection, not
from a failure to witness but from the unceasing potential of witnessing to
take place and the repeated demand that we do so. It circulates unpredictably
and is experienced variably, yet once encountered it demands witnessing. It
is a forceful if fleeting veering of experience into disjunctive and disruptive
terrain. Radical absence is radical b ­ ecause of its intensity, not b
­ ecause of any
definitive rupture with prior forms of media and pro­cesses of mediation. It
makes perceptible the disjunctive pluriversality of cultural and p ­ olitical life.
As with the other chapters, this one oscillates along two trajectories: the
witnessing of absence and the absence of witnessing. It argues that radical
absence is crucial to witnessing what is not ­there, or fails to materialize, or
is destroyed, or has died: a necessarily nonhuman witnessing. What does it
mean for witnessing to understand traumatic mediations as bound up with
the absent presence of data infrastructures? In worlds of increasingly fluid
and uncertain distinctions between the ­human and the nonhuman, radical
absences occur with remarkable potency. Traumatically affecting, if not trau-
matizing, they have the capacity to produce intensely felt disjunctures. Such

Witnessing Absence 153

intensive disjunctures pull us into a witnessing relation, but this witnessing
­doesn’t simply happen through media. Radical absence entails both the event
of absence itself and the eventful ­process of its violent mediation. Radical
absence and its witnessing are thus inseparable from nonhuman pro­cesses
of signal flow, datafication, and algorithmic se­lection, but also the material
infrastructures of data centers and optical fiber cables. Radical absence frag-
ments time and segments space, distributing and dispersing the experiences
of both ­human and nonhuman entities.
Radical absence depends in the most fundamental sense on the sheer phys-
icality of ­those infrastructures, even as it obscures their presence b
­ ehind the
screen interfaces of social media and search engine platforms. Such infrastruc-
tures coproduce witnessing radical absence at the ontoepistemological level:
no ­matter how ­human its subject ­matter, witnessing radical absence depends
upon and veers inevitably into the nonhuman. Radical absence entails ma-
chinic affects, but its dependence on such infrastructures means that it is also
bound up with ecological trauma. Understanding the witnessing of radical
absence thus requires attending to this infrastructural layer of absent pres-
ence and considering how dis­appeared infrastructures might be made to (re)
appear within the nonhuman witnessing of all-­too-­human actions and events.
Radical absence trou­bles relations between the sensing and sense-­making that
defines aesthetics, producing a disjunctive intensity rather than an anesthetic
numbing. It is a formation of machinic affect that galvanizes violent media-
tion at the level of experience itself: a flashing up of traumatic rupture at the
interface of the ­human and the non in the infrastructures of digital life. Like
the art and activism that have percolated through this book so far, nonhu-
man entanglements offer some potential for radical absence to open spaces,
however minor, however fragile, for reparative politics—to find something of
the transformative in encounters with loss. To encounter radical absence is to
be thrown into a witnessing relation, but one that refuses fixity, that loses its
own substance: a witnessing inseparable from the machinic affect of digital
life, in all its nonhuman excess, patterning, and sensory dysphoria.

second absence: mh370

Less than an hour ­after take-­off on March 8, 2014, somewhere over the South
China Sea, Malaysian Airlines Flight 370 made its last contact with air traf-
fic control at 1:21 a.m. local time. Flying from Kuala Lumpur to Beijing was
meant to take less than six hours, but the Boeing 777 was only seen again

154 Chapter Four

in fleeting electronic fragments—­a handful of blips on military radar, six
satellite “handshakes”—­and even then only ­after the fact, data dredged from
vari­ous regional monitoring stations.9 As hours lengthened into days, the
disappearance became charged with an unsettling intensity. mh370 came
to be oddly, inescapably pre­sent in its absence. Broadcast media filled with
breaking news segments, expert panels, and frenzied banners; programming
was interrupted and redirected, folding into the online buzz of anxiety, spec-
ulation, and hope. On social media, the real-­time digital flow made pos­si­ble
a cofeeling of this absence, an attention to its emergence and coalescence
within the stream of enmeshed communications. Emerging too was grief for
the 239 missing passengers and crew, grief for their relatives and friends who
gave ­faces to loss, marking the absence of loved ones on their skin. Participa-
tory platforms from Twitter to Facebook to Reddit enabled p ­ eople across the
globe to track and even participate in the search, an affective engagement in
which countless microencounters modulated and amplified the experience of
the plane’s absence. Checking in on events and finding no revelation or reso-
lution was not a lonely task, but rather one assembled of new encounters with
continued disappearance, with the very failure of finding the plane. Theories
abounded. The ­pilot was a terrorist, his home-­flight simulator an object of
suspicion. Passengers ­were hijackers, a pair of ­Russians briefly became a locus
of interest. Amateur sleuths set to work; maps proliferated. Islands ­were pored
over for the space to land and hide a plane, disused runways ­were cata­loged,
fuel capacities and headwinds ­were calculated to define the limits of where
the flight might be.10 More theories: it had been shot down ­after straying into
US war games with Thailand, or by China, or b ­ ecause it came too close to a
secret American base in Diego Garcia. Dark ­matter leaking from within the
planet had produced an unseeable, untraceable vortex.11
Investments in such speculation ranged from the occasional to the obses-
sive, a desperate desire to give narrative to the dis­appeared plane. No doubt
for many it burst across their digital worlds and slipped away, leaving only
faint traces. Yet to encounter the plane’s disappearance was to be affected by
an absence that was so profoundly pre­sent that it became radical. The more
it persisted, the more the search widened, and the more theories grew—­the
more its absence could be felt intensely. The world became less known, tech-
nology failed to m­ easure up to the faith we invest in it. Oceans w ­ ere revealed
as vast realms about which h ­ umans know l­ittle, tides and currents without
accurate models, topographies without maps. The limitations of our capac-
ity to search and rescue became starkly evident, the smallness of the h ­ uman
confronted with the scale and force of the seas.12 Skies ­were suddenly less

Witnessing Absence 155

tracked and watched than we had ­imagined. Despite the seeming ubiquity
of atmospheric, terrestrial, and aquatic remote sensing, what Gabrys names
“program earth” could still produce catastrophic errors.13 When Malaysian
authorities declared the plane lost in the Indian Ocean on March 24, more
than two weeks ­after the disappearance, no wreckage had been found. Th ­ ose
satellite handshakes and radar blips had led to mathematical equations
theorizing the zone of the plane’s crash, some 1.8 million square miles of
ocean.14 ­Because its wreckage was never found, the accumulation of calcu-
lation declared the absence of plane and p ­ eople to be final. Even e­ ighteen
months ­later, when a flaperon from the plane’s wing was found on Ré­union
Island in the Indian Ocean, no one could say for sure that the plane had sim-
ply fallen into sea or from where the wreckage might have traveled. Undis-
covered, mh370’s black box flight recorder held onto its secrets. Autonomous
drone operations failed to find wreckage, producing instead a happenstance
cartographic archive derived from the data of sonar sensors.15
gps tracking, satellites, radars, transponders: our experience of the con­
temporary world is of endless interconnection. Smartphones know where
we have been and how often, electronic tags chart the movement of our cars
across cities, transport cards log our daily travels. Hollywood has taught
us that the technological eye is all-­seeing, that even the act of stepping off
the grid is itself marked and known. Expansive computation harnessed to
remote-­sensing apparatuses promised what Paul Edwards calls the closed
world of Cold War computation.16 Yet this apparatus was calibrated to
the concerns of capital, climate, and empire, to the monitoring of missile
launches, border zones, glacial erosions, coastal reef temperatures, agricul-
ture fertility, and mining. To encounter such a disappearance within and
through the digital—in Facebook posts and Twitter hashtags, snippets of
YouTube video and subreddits—­was to encounter a strange rupture. A fissure
in the seeming solidity of the technoscientific world. A plane had vanished:
How could this be in an age of transponders and satellites and ubiquitous
connectivity? How could something so familiar and material simply vanish?
Where w ­ ere the witnesses? Through ceaseless connection with the shared
experience of a dis­appeared object witnessing itself became ever more non-
human: it was as if something inescapably ­actual had slipped back into the
virtual, as if the concrete had dissolved into the affective. ­There was a kind
of trauma in this—not so g­ reat, of course, as that felt by the families of the
missing, but an affective trauma, a trauma produced by a breakdown in
the certainties of the con­temporary world. This breakdown ­ripples into the
mediated environments that Paddy Scannell calls the “invisible care structure

156 Chapter Four

that gives the conditions of ­things we can trust in the world and a world we
can take for granted.”17 A plane simply disappearing ruptures that trust; the
mediated circulation of its absence threatens just such a rupturing in the trust
that defines our phenomenological relations to media.
This is traumatic affect, digitally mediated, a prepersonal yet corporeal
contact with radical disjuncture. This traumatic affect is not static, and as
much about the ­future as the past. It might not direct our actions as such,
but its infiltration of our sensorium cannot quite be undone. To encounter
mh370 digitally was to do so in diffuse pulsings, in micromoments of media-
tion: the disappearance of a plane and its passengers held in the palm of the
hand, engaged with the fingertips. An absence so radical—so fundamental
yet urgent, so distinct from the everyday—­that it demanded witnessing,
even as all that remained to be witnessed was a space in the world where a
plane once flew. As submersible drones whirred through oceans, new worlds
unfurled to ­human knowing but the remnants of the plane remained lost. Its
last witnessing was an elemental one: the deep blue media of the sea itself.18

traumatic affect

Media are far more than surfaces on which trauma is inscribed. As Amit
Pinchevski argues, we can think of the “the traumatic as something that is
made manifest through media technological rendering,” rather than some-
thing that is simply represented in media.19 If radical absence begins with the
failure of the eyewitness to witness, an epistemological failure to translate
the registering of an event into knowable form, its continued existence as a
forceful absence on the plane of experience depends on more-­than-­human
pro­cesses of mediation. Mediation and trauma both share an uncertain re-
lationship between past and pre­sent, between presence and absence, and
between proximity and distance. As such, “media constitute the material con-
ditions for trauma to appear as something that cannot be fully approached
and yet somehow must be.”20At the level of ­process, technical media contain
within their own constitution the paradoxes that make trauma overwhelm-
ing: media are always entangled with experience, yet also insist on their
separateness. “Media ­matter,” writes Cubitt, “both in the sense of giving
material specificity to our descriptions of such abstract concepts as society
and environment, and in the sense of the active verb: mediation comes into
being as m
­ atter, its mattering constitutes the knowable, experienceable world,
making pos­si­ble all sensing and being sensed, knowing and being known.”21

Witnessing Absence 157

Digital media are also decidedly material, requiring huge amounts of w ­ ater,
space, and electricity to run and with catastrophic impacts on environments
and the animals and plants that inhabit them. While the other chapters in
this book oscillated in their proximity to the h ­ uman, h ­ ere I attend to quotid-
ian h­ uman life to examine its enmeshment with nonhuman systems at the
visceral, recompositional level of trauma. H ­ ere, violent mediations, machinic
affects, and ecological traumas cohere on everyday experiences of what Lau-
ren Berlant calls “crisis ordinariness,” or “traumas of the social that are lived
through collectively and that transform the sensorium to a heightened per-
ceptiveness about the unfolding of the historical.”22
In mediated encounters with crisis ordinariness, the factual and the fictional
can be blurred in complex and unpredictable ways, jumbling together the
urgent and the trivial, the enduring and the ephemeral, the intense and
the diffuse. Digital media can itself be unanchoring, displacing priorities,
destabilizing shared knowledges, and amplifying conspiracy and paranoia.23
How is one to cope with the media witnessing of 9/11, the ur-­trauma of the
screen, first on t­ elevision screen but then repeated across other media? Or,
more contemporaneously, images and video of police killings and beatings
of Black and First Nations p ­ eople, or of funeral pyres in India as covid-
19, vaccine apartheid, and government ineptitude took life a­ fter life? Th ­ ese
and countless other mediations make up what Mark Seltzer calls “wound
culture,” the “collective gathering around shock, trauma, and the wound.”24
This fascination is a complex one, caught up in movements of pulling away
as much as turning t­ oward, repulsion as much as attraction. Conceiving of
trauma as affective also entails a more fluid, interdependent understanding
of the social and the individual, and the dynamic role of mediation in their
relations. Taken together, ­these movements open more nuanced and variable
ways to understand what encounters with radical absence might mean at the
level of meaning-­making and of ­political possibility.
Traumatic affect, as I have argued with Meera Atkinson, describes the
“the mode, substance and dynamics of relation through which trauma is ex-
perienced, transmitted, conveyed, and represented.”25 Rather than a concept
that produces taxonomies and distinctions, that includes or excludes varying
experiences and repre­sen­ta­tions, traumatic affect is conceptually open and
fluid. It recognizes that encounters with trauma can be traumatically affect-
ing without being traumatizing in quite the same way that psychoanalysis or
clinical psychological imply. In other words, all trauma is affective but not
all experiences of traumatic affect produce trauma. Such traumatic affects
are not only—or not solely—­the affects constitutive of traumatic events, but

158 Chapter Four

also affects experienced in encounters with trauma. Affect, ­after all, is nei-
ther dependent upon nor delimited to the h ­ uman. As such, the intensity of
traumatic affect varies along a dynamic continuum, rather than occurring in
some static form, and its relational intensities are bound up with nonhuman
technics and milieus.
Thinking of mediated trauma in terms of its nonhuman relationality
opens onto new possibilities. As Pinchevski notes, “Media (re)produce the
traumatic by effecting its ungraspability affectively, by imparting impact in excess
of content, sensation in excess of sense.”26 An encounter with violent mediation
might itself be traumatic or traumatically affective, such as in the designation
of a racial category by Facebook’s advertising algorithms or the unexpected
encounter with footage of death or pain in a TikTok feed. Mediated encoun-
ters can also be intensely immediate: mediation is precisely what renders
them intensive. Traumatic affects can impart, at varying intensities, the force
of trauma without inscribing meaning. Traumatic affect can be understood
as one of the qualitative dynamics of crisis ordinariness. Thus, while this
chapter pursues traumatic affect within encounters with radical absence on
social media, its intensive relations of rupture and disjuncture form part of
what Andrew Murphie calls “a complex storm of feeling, of aspects of world
feeling each other in intense, unexpected and constantly mutating ways.”27
Witnessing radical absence, then, does not apply exclusively to social or even
digital media, but rather might be extended into the crises of war, ecol­ogy,
and data that have occupied this book so far.
While normative conceptions of trauma, like t­ hose of witnessing, tend
to insist on its exclusivity to the h
­ uman, understanding trauma and affect in
vitalist terms means opening space for the felt force of potential as it shapes
relations. Virtuality always precedes and exceeds the h ­ uman b­ ecause it is
never contained within the experience of any individual; it is always in and
of worlds, in relation, in media, and in the making. Recall that for Massumi,
affect is the shifting entanglements of “the virtual in the a­ ctual and the a­ ctual
in the virtual” in the unfolding of encounters.28 Massumi’s virtual does not
refer to the false or unreal, but rather to the crowding potentials of all ex-
perience: ­those ­things that are-­about-­to-be or might-­just-­have-­been that
are inextricable from the ­actual. Virtuality insists that nonhuman milieus,
technics, animals, and ecologies always co-­constitute the ­human as potential
and as a taking shape and becoming with world(s). Thinking trauma in terms
of affect thus enables a way of understanding its dynamics as fluid, moving,
changeable, multitudinous, and even contagious, while not relinquishing its
injurious and harmful ruptures of experience.29 As David Lapoujade writes,

Witnessing Absence 159

parsing William James, “Discontinuity always appears against the ground of
continuity,” such that even as experience fractures it remains held together
as one-­and-­many by relations of nonrelation.30 Conceived as always more
than one and always more-­than-­human, traumatic affect invites nonhuman
witnessing in response. In ­doing so, it offers a more generative mode for un-
derstanding relations across the diffuse circulation of crises of war, data, and
ecol­ogy in the cap­i­tal­ist quotidian of social media, search engines, and news
sites. Witnessing radical absence is both a response to the traumatic affects
of digital disappearance and an address to the absent presence of nonhuman
infrastructures.

third absence: digital death

In mid-2012, Jessa Moore logged onto Facebook and learned that her friend
Anthony Dowdell had killed himself. She and ­others began to post memories
and photos, to tag him at restaurants or bars. “Facebook became our memo-
rial,” she said. “We could leave messages for him and each other.” Facebook
became a site of shared mourning, but also a way to keep memories alive—­
even as it continually reminded Jessa of her friend’s absence from her life.31
Almost a d ­ ecade l­ater, Jessa’s experience is far from unique as I and many
­others can attest, but her story, told in a widely read article in the Huffington
Post, marks an early incursion into media discourse of death on Facebook
and its weird affects. Estimates suggest that upward of thirty million Face-
book profiles have outlived the ­people who created them, with around eight
thousand users passing away e­ very day. In 2019, Carl J. Öhman and David
Watson published a statistical projection of the accumulation of profiles from
deceased users, using country and age data scraped from the Facebook api
in conjunction with country mortality rates. Their findings suggest that up to
4.9 billion dead users could populate Facebook by 2100, leading the research-
ers to call for a new, scalable, and sustainable model for preserving the data
of the deceased.32 Already, a microindustry has emerged to manage digital
estates, wrapping up accounts, tracing assets, and passing on data.33 On Face-
book, friends or ­family access accounts and make them inactive, or provide a
death certificate to Facebook to have their account officially “Memorialized,”
transforming their profile into a commemoration to which existing Facebook
friends can post but remains other­wise unchanged.34
­Others are simply left in place, digital presences that bear no clear marker
of absent life, as if the user has simply stepped away from the computer. Yet

160 Chapter Four

unlike the cluttered desk or sink piled with dirty dishes, such a profile remains
open to further engagement. Like Jessa tagging her friend Anthony at a bar he
would have liked, or leaving messages on his wall, t­ hese pages have a strange
digital afterlife. This capacity to tag is particularly potent. As Paul Frosh
argues, tagging on social media “reveals itself as an existentially significant
technique for mediating the attachments of the body and the self in the face
of pos­si­ble unravelling and disintegration.”35 Tagging the dead marks an insis-
tence on retaining a certain presence in the world produced by digital media,
but it also points to the tenuous hold that the practice of tagging has on the
deep relational work of maintaining attachments. Yet stranger still and far less
vis­i­ble than tagging and memorializing are the archival remainders of dead
profiles, living on in Facebook’s inaccessible and undisclosed data centers.
Facebook’s archive is not ephemeral but material, as Mél Hogan shows, consti-
tuted by “the electricity that powers the machines, and a virtual ethersphere
that produces bigger rec­ords than the lived realities it rec­ords, as a politic
of preservation that is, on the one hand, successfully inhabited and, on the
other hand, dangerously reconfigured and protected as such.”36 Death on
Facebook as both event and archive, as enduring public profile and invisible
data cluster, is not apo­liti­cal but bound up with the politics of data centers
and big data aggregation, and with the practices of attention, engagement,
consumption, and preservation that shape digital sociality and selfhood.
Memorialized profiles manifest decreasing attention, a flurry of sorrowful
posts that fades with time, sparking at moments of remembrance or past joy.
­Those left untended intrude unwittingly: an algorithm suggests you get in
touch with a deceased friend, a reminder appears for their birthday, you are
tagged in a photo with them. A brief foray onto Google reveals the complex
multiplicity of experiences of digital death, from memorial pages for beloved
pets to ­services that send letters from the afterlife to start-­ups promising the
first stage of a transhumanist transcendence of death through a postlife
digital existence. On social media, death often intrudes unexpectedly, en-
countered in unfitting places: clicking on the profile of an acquaintance not
seen for years, checking notifications on a bus on the way to work, wondering
why your ­daughter ­hasn’t called home from college. Individual stories pre­sent
differing relations. Jessa Moore visits her friend’s page each day to leave him
messages; a ­widow wrestles with ­whether to leave her partner’s page active.
Death in the digital sphere is collaborative and delocalized, constituted by
friends and ­family, by the algorithms of the encoded world. It repeats the in-
tersubjectivity of digital identity, its formation not only through interpersonal
connections but also via the technics and constraints of digital platforms

Witnessing Absence 161

themselves. “It’s more for us than for him,” said Jessa. Witnessing absence
on Facebook is a relational practice, a means of forging communities of care
around shared knowing and grieving, but one funneled through the extrac-
tive data regime of the social media platform.
Social media make vis­i­ble and grant a kind of permanency to other­wise
transitory relations: schoolmates, old work colleagues, fellow backpackers,
or partygoers. Once, their deaths would have filtered into our lives sparingly,
or not at all. Social media relations d ­ on’t fade in quite the same way; even
when someone might have dis­appeared from your newsfeed, an event of
some magnitude or the foibles of the algorithm might push them to the fore,
shunting aside the absence afforded by space and time. When testimonies
to lost lives appear, they do so in the same flattened aesthetic as ­every other
item. A friend pours grief for a dead parent into the status box, hits post, and
their words appear alongside ­political rants, sports highlights, requests for
advice on buying a new barbeque. ­Unless the link is broken, Facebook keeps
connections active; more, it calls its users into action. Connect with Anushka,
wish Peter a happy birthday, like the photo Siobhan just shared of her new-
born child. To be thrown into shock or grief is no small ­thing, to encounter
in a digital presence a radical absence of life can produce a bodily response
of visceral intensity. Life is absent, yet also jarringly, movingly, or even thank-
fully pre­sent: its absence is radical in the sense that its traces are inescapable.
Photo­graphs, posts, comments, likes, events, check-­ins, games—so much
more cohesive, contained, and accessible than the material and ephemeral
remnants of the predigital.
On Facebook, the dead are radically absent in part b ­ ecause engagements
with them are so vis­i­ble: mourning is not only public, but enduring.37 Re-
sponding to death becomes a kind of public testimony: remembrances and
condolences are not fleeting or offered in private, but within the performa-
tive space of the social network. Eliding geography and producing intimacy
across distance, such grieving entails a kind of flattening of experience within
the bland corporate aesthetic and ethos of the platform. Facebook becomes
a constraining medium, giving a form to grief that limits or even denies
something of experience, a more intensified form of the delimiting of affect
that occurs in the writing of trauma.38 All griefs are rendered equal, or near
enough, and this can be traumatically affecting. Lines between rubberneck-
ing and mourning are increasingly blurred, such that expressing grief in pub-
lic can be undercut by the doubt of ­others. To lose a partner, a close friend, a
son or s­ ister, and then have countless o ­ thers lay claim to them can reproduce
loss as much as testify to life. Traumatic affect in digital death can be fleeting,

162 Chapter Four

a passing encounter with dis­appeared life. Or it can be unexpectedly intense,
an encounter with loss that throws one’s own life out of kilter, the actuality
of death intersecting its virtual counterpart. An affectivity of absence pro-
duced within and by the ever-­pressing potential of the digital, made material
through nonhuman infrastructures of data collection, storage, pro­cessing,
and distribution. Witnessing death as radical absence is all too h ­ uman, yet
this digital mediation ensures that it is inseparable from nonhuman technics,
algorithms, data, and affects.

radical absence (redux)

Screen-­based media are only one slice of the pervasive digital mediation of con­
temporary life, but their ubiquity means every­thing from homes and shopping
malls to buses and elevators has been infiltrated by the datalogical. To move
through such spaces is to have our attention demanded and diverted, with
digitized movement and sound calling us into a more temporal relation to
the visual and aural than the static imagery of the past allowed. This demand
for attention is also a bodily experience, from the adrenal surge that redirects
the body in gaming to the haptic signals of smartwatches. Augmented and
virtual real­ity hold the promise of interrupting our relation to the visual
field, layering data over what we see or replacing our immediate surrounds
entirely. Fantasies of neural link implants hint at a ­future of screen-­body
fusion. Even now, interpersonal interactions slip between online and off, or
take place si­mul­ta­neously in both domains.39 Smartphones and their ilk have
become what Bernard Stiegler calls “mnemotechnologies,” ­doing the work of
thinking, remembering, and pro­cessing our knowledge of the world.40 What
had been stable categories of causation no longer hold as relations between
objects, ­humans, and dif­fer­ent media become increasingly fluid and relative.41
Even if ­there are antecedents for the transformative effects of digital net-
works in the long h ­ uman history of mediations, such as the telegraph’s col-
lapsing of distance or cinema’s production of new modes of time, t­ here can
be ­little doubt that recent ­decades have seen an accelerated evolution in the
imbrication of media technologies and h ­ uman life. Nonhuman technics and
the h­ uman sensorium are increasingly enfolded; affects flow between the cor-
poreal and machinic, intensities surging across surfaces and substrates, mod-
ulating and shaping. In the words of Nigel Thrift, “­There is no stable ‘­human’
experience b ­ ecause the h
­ uman sensorium is constantly being ­re-­invented
as the body continually adds parts to itself; therefore, how and what is

Witnessing Absence 163

experienced as experience is itself variable.”42 This cyborg tendency is not new,
but it is accelerating and expanding. Drones and smartphones, for instance,
are far from the only emergent forms of machinic perception. Sensors creep
into ­running shoes and sidewalks, grocery aisles, and the skin of diabetics.
Machine vision enables autonomous systems to power cars, surgical robots,
and the logistics of ports. Learning algorithms analyze public health data and
execute trades at scales, speeds, and frequencies far beyond ­human cognitive
capacities. Chatbots generate text from the statistical analy­sis of patterns in
data trawled from the internet. More than simply shifting how experience
is composed, the ever-­presence of t­ hese technovital mediations changes the
ground of experience itself for the h ­ uman and for nonhuman entities of
all kinds. Media technologies, argues Marie-­Luise Angerer, play an “active
role in promoting a comprehensive relationality by setting and correlating
the rhythms of large and small units and inward and outward sensations.”43
Sensations are not only produced bodily, but through the folded together
mediations that make for a new fluidity of experience itself. As Brian Mas-
sumi and Erin Manning write, “The field of experience can be best described
as relational-­qualitative.”44 This is not to say that the ­human experience of the
con­temporary world is without grounding, that what was solid has melted
into the networked ether, but rather that the ground truths of experience
are more and more entangled in the systems and pro­cesses that produce the
conditions of crisis in which so much life is lived.
Radical absences are made pos­si­ble by this folding together of experience
and digital mediation. They are themselves not rigid phenomena: videos
circulate and then fade from view, flight paths are traced then slip away,
posts in remembrance grow infrequent. Space and time bend and flex in odd
ways. This combination of spatiality, temporality, and contingency m ­ atters.
Mary Ann Doane writes of t­ hese phenomena in the context of the cinematic
image, in which “chance and contingency have been assigned an impor­tant
ideological role, supporting the fascination with the apparently alternative
temporalities offered by the cinema.”45 Yet what occurs in the digital media-
tions of disappearance described ­here shows how the digital enables dif­fer­ent
experiences: chance, contingency, spatiality, and temporality are always and
inescapably constitutive of encounters with radical absence. Th ­ ere is a ten-
sion in this between their intensity in the encounter and its durability over
time and across space. Traumatic affects s­ haped by radical absence can be
modulated, amplified, diminished, deferred, interrupted, and truncated in their
passage across and between bodies. Radical absences are events rather than
stand-­alone phenomena, manifestations of encounter that are co-­composed,

164 Chapter Four

emergently assembled. Experiences of radical absence are neither singular
nor collective as such, but rather transindividual, “the collectivity at the heart
of all individuations, before and beyond any spectating into individuals.”46
Yet this emergence of the event—­the potential encounter with the behead-
ing video, the pervasive presence of a dis­appeared plane, the still-­living of
death on Facebook—is not solely dependent on the h ­ uman. Technologies
­matter, pro­cesses m ­ atter, mediations m­ atter—­they m ­ atter in the doubled
sense of carry­ing impor­tant information and making material. In this sense,
radical absence performs an indexical function: it is the trace of the dis­
appeared. Yet the indexicality of radical absence does more than point as
close as can be to that which is not ­there: what makes ­these absences radical
is the affectivity of encounter that materializes in lived experience that which
is indexed in the digital. Radical absence collapses the seemingly limitless na-
ture of media into its finitude, evoking the limits of what media can capture.
Radical absence reminds us of the vitality of digital media, that its materiality
does not reside solely in binary code but in how ­those codes work upon bod-
ies and in the data centers, cables, transmitters, and repeaters that make their
operation pos­si­ble. Indeed, it is in this conjunction of (non)­human(s) and
media(tion) that par­tic­u­lar encounters with digital disappearance become
radical. More than agency as such, but a p ­ rocess of assemblage: “The directed
intensity of a compositional movement that alters the field of experience.”47
Within encounters with radical absence, ­these compositional formations
move most intensively in the tension between what is not and what almost
might be, in the swarming of potential that withholds certain ­actuals. In
short, radical absences are never static—­closer to presences, yet not quite,
possessed of a force that is exactly not presence.
Nor are they solely traumatic in their affectivity: much more than trauma
circulates in the absences described ­here. Love, grief, fear, despair, alienation,
and other affective formations are often at work. Histories, states, and moods
of bodies m ­ atter too. To have a fear of flying and encounter a rising tide of
mh370 posts and comments in your social media stream gives a par­tic­u­lar
­angle or tenor to that encounter. Massumi calls this differential attunement:
“bodies in encounter are both completely absorbed in the felt transition,
but they are differently absorbed, coming at it asymmetrically, from dif­fer­ent
­angles, living a dif­fer­ent complexion of affecting-­being affected, transitioning
through the encounter to dif­fer­ent outcomes.”48 This differential attunement
means that the ways in which traumatic affect affects ­will vary: radical ab-
sences are not radical in quite the same way in e­ very encounter that (re)
constitutes them. Yet what they share is this traumatic affectivity—it is the

Witnessing Absence 165

qualitative relation that defines the encounter. Not traumatization as such,
but a disjunctive, rupturing affectivity, in which affect’s presignifying quality
limits the capacity of the event to become meaningful. It is this felt-­feeling
of a refusal to cohere into sense-­making that makes the potential encounter
with beheading videos a force of disruption that is experienced bodily.
Radical absences do not call the body into them, but rather call the body
alongside them, to encounter what has dis­appeared in the force of its ab-
sence: an all-­too-­human witnessing thrown into the domain of nonhuman
technics. Entanglement without overlapping, a shared composition that pro-
duces a kind of synchrony that is not sameness. “Entanglements,” Rey Chow
reminds us, are “the linkages and enmeshments that keep ­things apart; the
voidings and uncoverings that hold ­things together.”49 By calling attention
to the disjunctive ways in which we are entangled with disappearances that
manifest in the digital sphere, radical absence is an injunction to the necessity
of nonhuman witnessing, of witnessing that exceeds the h ­ uman, occurring
not in the event itself but in the affectivity of its aftermath and in the material
traces of its datafied afterlives. The strange time of digital media ­matters h ­ ere,
with its tension between liveness and belatedness, proximity and distance,
and the vagaries of algorithmic determinations of significance that pluck
events out of the past to bring them to attention. The time of witnessing in
digital media—­especially on social networks or in the preferences of Google’s
PageRank ordering—is subject to nonhuman contingencies, associations,
circulations, and relations. Machinic affects compose clusters of relation that
pull certain t­ hings to the fore, and then allow them to recede or dissolve.
Witnessing radical absence has no pre­sent as such, not even in the moment
of James Foley’s death, or the plane’s disappearance, or the memorialization
of a Facebook page, or the aftermath of the violent expansion of a mine.
Radical absence thus constitutes a structure of relation to the disjunc-
tive crises of the affective pre­sent, the pre­sent as it is affectively formed as
something that can be made sensible despite its discontinuities. It is a form
of attachment, a way in which subjectivities relate to contexts in modes rang-
ing from the aesthetic to the p ­ olitical to the occasional. Radical absence is
an attachment to that which has dis­appeared yet remains affectively pre­sent,
digitally manifested, preserved in infrastructure. While traumatic affect—­
not to mention circulations of grief, fear, disgust, and outrage—­animates its
disjunctive force, such affect is not the ­thing itself. Traumatic affect is not nec-
essarily contagious, but rather a form of relation conducive to affective con-
flagration: it is an affective structure formed within what Berlant calls “a crisis
culture borrowing trauma’s genres to describe what ­isn’t exceptional at all in

166 Chapter Four

the continuous production and breakdown of life.”50 This structure is one of
flows and vectors rather than scaffolds or walls. Radical absence stretches
the bound­aries of the subject, reconstituting selfhood within digital world-
ings s­ haped by the traumatic affects of intimately distant disappearances. It
percolates through infrastructure, which Berlant calls “the living mediation
of what ­organizes life: the lifeworld of structure.”51 Witnessing absence is thus
bound up with the witnessing of invisible infrastructures, or what “binds us
to the world in movement and keeps the world practically bound to itself.”52
The digital makes this intimate distance pos­si­ble by enabling a collapse of
space that si­mul­ta­neously calls attention to its own occurrence: in the digital,
we can touch the distant but always do so in an intensely mediated way.
If the digital continually extends and reworks subjectivity, as Thrift and
­others argue, then traumatic affect is increasingly folded into our digital be-
comings. Such traumatic affects are the vehicle for sensorial manifestations
of radical absence, for the ways in which it is felt bodily. “The body d ­ oesn’t
just absorb pulses or discrete stimulations”; writes Massumi, “it infolds con-
texts, it infolds volitions and cognitions that are nothing if not situated.”53
Rather than an enfolding that envelopes the body, this infolding alters bodily
states, including the a­ ngle at which the body senses events. Sensation “is the
immanent limit at which perception is eclipsed by a sheerness of experience,
as yet un-­extended into analytically ordered, predictably reproducible, pos­si­
ble action.”54 Within the digital’s ever-­present pulsings and infoldings, radical
absence holds its affectivity in this indeterminate zone between perception
and sensation. This sensorial experience of radical absence gives it a visceral
actuality, renders it more than a discrete media object. As affective structures,
radical absences do something. They enact a relation of nonhuman witness-
ing to the terrain of national security, or to faith in technology, or to how one
grieves, or to Country wounded by resource extractivism. Such nonhuman
witnessing is more than material ­because it is always relational, always in the
­process of forging registrations, connections, and attachments when it seems
that relationality itself is u
­ nder assault, even by the most violent of ecological
traumas.

fourth absence: sacred sites

On May 24, 2020, the mining ­giant Rio Tinto detonated two rock shelters
in the Juukan Gorge in the Pilbara region of Western Australia, destroying
sites sacred to the Puutu Kunti Kurrama and Pinikura (pkkp) ­peoples that

Witnessing Absence 167

provided evidence of more than forty-­six thousand years of continuous oc-
cupation.55 Charges had been laid for the extension of the Brockman 4 iron
ore mine days before, but the Traditional O ­ wners of the land w­ ere not noti-
fied despite formal consultations extending back to 2014. As media reports
and outrage spread among First Nations ­people and settlers on social media,
the irreparable damage became inescapable: the absent sacred sites became
sharply, affectively pre­sent. Material witnesses to the enduring habitation of
the land ­were lost: plaited hair four thousand years old, genet­ically linked to
the pre­sent Traditional ­Owners, and bone and stone tools dating back more
than twenty-­eight thousand years, the oldest-­known bone technologies in
Australia. Worse still was the incalculable loss of sites sacred to the pkkp,
places alive with ancestral spirits and an enduring vitality that has no equiva-
lent in Western epistemologies. Testifying to this profound and visceral loss
of living Country to the Standing Committee on Northern Australia’s inquiry
conducted in the wake of the blast, Traditional ­Owner Burchell Hayes told
the committee:

The Juukan Gorge is known to be a place where the spirits of our relatives
who have passed away, even recently, have come to rest. It is a place that
the very, very old p ­ eople still occupy. Purlykuti has been specifically re-
ferred to by the old ­people as a place of pardu, which refers to the special
language only spoken during ceremonies in the Pilbara. Our elders state
that it is certain that the spirits are very disturbed, and their living relatives
are also upset at this. This is why Juukan Gorge is impor­tant. It is in the
ancient blood of our p ­ eople and contains their dna. It h­ ouses history and
the spirits of ancestors and it anchors the ­people to this country.56

Their absence would remain unbearably pre­sent, even as the cascading after-
math of the blasts brought a rare moment of scrutiny and accountability for
extractive capitalism and its ­legal and ­political foundations.
Enabling the destruction ­were two proximate agents of what Aileen
Moreton-­Robinson calls “the possessive logic of patriarchal white sover-
eignty”: the incompetence and negligence of Rio Tinto and the gross dis-
parities of Western Australia’s Aboriginal Heritage Act 1972, u ­ nder which the
destruction of the sites had been approved in 2013.57 On the Rio Tinto side,
the systematic sidelining of heritage reports and Traditional ­Owner con-
cerns became evident, facilitated—or so it was claimed—by the geographic
distance of the com­pany’s executives in London from its mining activities in
Australia. This absence of communication protocols and heritage manage-
ment practices combined with an institutionalized disdain for traditional

168 Chapter Four

owner­ship, exhibited by the com­pany’s sustained and systematic approach in
seeking approval for the destruction of thousands of sacred sites over ­decades
of mining in the Pilbara. Like its competitors bhp, Fortescue Metals Group,
China Shenhua, Roy Hill, and o ­ thers, Rio Tinto had made ruthless use of the
Heritage Act to push through new mines and expansions with ­little regard
for the Traditional ­Owners. ­Under the Heritage Act, the Aboriginal Cultural
Heritage Committee is not required to consult with Traditional O ­ wners, nor
can its decisions be appealed. Even speaking publicly can threaten compensa-
tion for Aboriginal corporations through draconian gag clauses that enshrine
stark inequities into the administrative ­process itself. Widely recognized as
unfair and outdated, reform efforts for the Heritage Act had stalled. But the
furor over what Yawuru man and federal senator Pat Dodson described as
Rio Tinto’s “incremental genocide” brought to mainstream attention the en-
during coloniality of Australia’s extractive capitalism and its supportive ­legal
bedrock. Pre­sent in the radical absence of the sacred sites was settler colonial-
ism as eliminationist structure.58 The destruction of culture operated ­here
as a way of breaking traditional bonds with Country, an expression of the
shifting logics of racialization in response to p ­ olitical activism that stresses
the significance of land to First Nations. The shared investment of settler
government and corporation in the continuation of extractive industries in
the face of First Nations r­ esistance and global heating both depends upon and
reproduces settler sovereignty. A mining g­ iant founded in settler Australia
but now headquartered in London (and subjects of the British Crown in
­either locale) had destroyed sites of sacred importance in the name of profit
and with the imprimatur of law that explic­itly and deliberately marginalizes
and gags Aboriginal ­people. ­Here was the convergence of settler colonialism,
extractive capitalism, and neoliberal corporate structures, suddenly all too
pre­sent against the radical absence of the Juukan Gorge sites.
Even more rare than the visibility of infrastructures too often hidden
in plain sight was the push to hold Rio Tinto to account from within the
­political establishment. ­Under the weight of public scrutiny and a formal
inquiry by Parliament’s Joint Standing Committee on Northern Australia,
three executives and the chair of the board resigned, and the com­pany com-
mitted to a range of remedial actions, internal reforms, and changes in its
engagement with Traditional O ­ wners—­yet this might be read as the new
normal of d ­ oing business in extractive industries rather than meaningful
punishments. As with the rare prosecutions that follow police vio­lence, the
com­pany’s actions hardly constitute justice. Nor do they do anything to undo
the under­lying structure of resource income ­dependency for Traditional

Witnessing Absence 169

­ wners, profit maximization by multinational miners or even the specifically
O
problematic Heritage Act, let alone the ­legal edifice stacked in ­favor of extrac-
tive fossil capitalism. The system endures, the sacred sites remain intensely,
profoundly absent. A ­ fter a brief hiatus, more continue to be blasted with
shocking regularity. Indeed, the sites had borne witness already, examined for
significance within an archaeological framework of knowledge legible to Rio
Tinto and to the ­legal system of Western Australia. Rendered into evidence
of enduring presence by scientific dating techniques, the sites ­were already
testifying—­material witnesses within an epistemic mode legible and contest-
able to the state. This material testimony was in turn accompanied by the
witnessing of the Puutu Kunti Kurrama and Pinikura p ­ eoples in their years-­
long engagements about the significance of the sites. But all this witnessing
found no purchase within the system u ­ ntil it was too late. The destruction
of such sacred sites both enacts and legitimates the settler state, as resource
extraction depends upon the continued denial and devastation of Country.
Throughout the first year of the coronavirus pandemic, the destruction of
the sites at Juukan Gorge continued to reverberate. bhp paused the planned
destruction of forty sites, but de­cided ­after a review of its permissions to
permanently halt the destruction of just ten. Reports of Rio Tinto’s negli-
gence continued to mount. The Parliamentary Inquiry held sessions, took
submissions, and eventually traveled to meet with pkkp Traditional ­Owners.
In its final report released in October 2021, the Joint Standing Committee
[103.158.136.64] Project MUSE (2024-02-06 19:57 GMT) University of Edinburgh

on Northern Australia delivered a scathing indictment of the ­legal archi-

tectures that facilitated the destruction at Juukan Gorge and many other
places across the nation. It called for significant change, ranging from the
codesign with First Nations of new national heritage legislation to addressing
inequities in the negotiating positions of Traditional O ­ wners with mining
corporations and government. With its passionate evocation of First Nations
culture—­along with harsh words for Rio Tinto and the Western Australian
government and heritage laws—­the report made for startling reading: a rare
recognition of the brutal vio­lence of resource extraction and the facilitat-
ing l­egal regime of property rights and heritage laws that makes ecological
trauma the normal condition in settler Australia.
­Whether meaningful change ­will take place is difficult to say. By their own
admission, members of the committee ­were deeply affected by their time
on Puutu Kunti Kurrama and Pinikura Country and by their experiences
with Elders and other community members. The social media presence of
the inquiry itself was a background hum, occasionally punctuating the sur-
face but never sparking back to the intensity of its first days, even when the

170 Chapter Four

final inquiry report was released to a brief flurry of media attention. For the
­peoples of the Pilbara, the wound remains achingly painful. Radical absence
is often like this: it bursts and fades for many, remains intense for some. It is
tempting to conclude that radical absence can be the necessary precondition
for ­political accountability, but the nonhuman witnessing at work h ­ ere was
more complex than mere digital disappearance. Witnessing at Juukan Gorge
occurred through geological vitality, a vitality that was itself all too absent
from the digital mediation of the destruction. Witnessing absence at Juukan
Gorge—­and in the past, pre­sent, and f­ uture of the deliberate destruction of
First Nations heritage—­means attending not only to the radical force of the
lost sacred sites but also to the presence of an entrenched interconnection
of more-­than-­human institutions and infrastructures of law, capital, and
settler-­colonial control.

witnessing absence

“All attachments are optimistic,” writes Berlant, and radical absence is itself a
form of attachment, for all the grief and death to which it attends: a witness-
ing relation with what has dis­appeared, an attachment to what is no longer
pre­sent that enables positive change.59 A witnessing of absence in the absence
of witnesses: such an attachment can be animated by traumatic affect yet
still spark a reparative movement—­even if small, tentative, and threatened
by the very affectivity of the disappearances from which it might emerge.60
While far from a panacea and by no means a politics in itself, nonhuman
witnessing nevertheless widens the aperture from the ­human subject to
assemblages of h ­ uman and nonhuman entities. Witnessing radical absence
is only pos­si­ble due to the sheer materialities of networked infrastructures,
the algorithms and network protocols that enable the flow of machinic affect.
Witnessing radical absence means attending to t­ hose infrastructures, and to
the ecologies they disrupt, the wars they enliven, the extractive industries
they streamline. Witnessing absence in this way makes pos­si­ble a dif­fer­
ent kind of response to systemic oppression than the voice of the testifying
subject, or even the assembled evidentiary force of Schuppli’s material wit-
nesses. Witnessing absence asks that we hold onto the possibility of witness-
ing in nonnormative ways, working outside the frame of courts and public
contestation. If we accept Berlant’s proposition that all attachments contain
some ele­ment of optimism, then an intensive attachment to absence might
well contain within it new forms and dynamics of relation that contain new

Witnessing Absence 171

possibilities for becoming other­wise, for turning radical absence into repara-
tive care.
At the heart of what distinguishes radical absence from other structures
of relation to the affective pre­sent is its disjunctive mode of attachment. This
disjuncture does not negate ­others or community, or not necessarily so. By
calling disappearance into relation, radical absence affords the space for
disjunction to produce change, to enact something new. For violent media-
tion to shift ­toward repair. “Without disjunction ­there would be no cut, no
cleaving, no inflection, no minor gesture,” writes Manning.61 Without such
cuts, the field of experience remains static. What Manning points to h ­ ere is
“the gestural force that opens experience to its potential variation,” such that
“its rhythms are not controlled by a preexisting structure, but open to flux.”62
Attending with an altered ­angle of approach to the wider field of fear within
which execution videos circulate, or to the faith that we place in technology
to protect our ­human fragility, or to the collectivity of loss enacted in digital
memorials, or to the laws of the settler state, might be small acts but they are
not insignificant. However minor, t­ hese gestures contribute to composing
something ­else from what is ­going on around us. “Affect ­matters in a world
that is always promising and threatening to amount to something,” writes
Kathleen Stewart. “Fractally complex, ­there is no telling what ­will come of it
or where it ­will take persons attuned.”63 Traumatic affects that coalesce within
encounters with radical absence can be an opening onto the reparative, but
not ­because they afford the opportunity for treatment in any clinical sense.
Arising from the digital, they bring to visceral life how mediation entangles
experiences in unpredictable ways that vary in intensity and form as they
flow and ebb in time.
Across ­these and other radical absences, such shifts might be small, even
tentative, but it is the minor gestures of the everyday that constitute the first
glimmerings of a ­political other­wise. It is a politics without the requirement
for institutional engagement or party affiliation, for spatial or temporal con-
tiguity. Rather, it is a politics that finds in radical absence a means to move
beyond atomization and isolation, yet not insist on proximity as the founda-
tion for collective feeling and action. Radical absence need not engender a
collapse into a traumatic void, some ruptured space of digital loss. Rather,
witnessing radical absence in all its more-­than-­human complexity might give
the slow work of reparation and care an urgency they can other­wise lack. It
might alter the trajectory with which one approaches the public feelings that
circulate online and off. It might make corporeally real the imbrication of
digital media and mediations into evermore crevices of h ­ uman life and death.

172 Chapter Four

Small movements, but not trivial ones. Finding in the potential for hurt, for
negative inflection, for encounter with traumatic affect, that reside in such
encounters with the violent mediations of digital disappearance some small
space for renewed life. Working through the entrapments, dispossessions,
and disempowerments of algorithmic enclosure requires grappling with the
macroscale of digital capitalism, data colonialism, and platform politics, but
that strug­gle can begin with a witnessing relation. Nonhuman witnessing
tugs the ­human into altered relations with the infrastructural milieus that
make up the material and affective pre­sent. Witnessing radical absence pulls
to prominence continuums of experience: from presence to loss, life to death,
and hope to fear—­however faint and fleeting they might be. How, then, might
the politics of nonhuman witnessing be theorized? It is that question to
which the coda of this book is addressed.

Witnessing Absence 173

​c oda

­toward a
politics
of nonhuman
witnessing
covid-19 has shattered many of the fictions that sustained the global
order, racial capitalism, and the supremacy of Man. Th ­ ese pandemic years
have been a brutal reminder of the nonhuman agencies that impinge upon
and transform us and our ways of living in profound and immea­sur­able ways.
Writing on the growing intimacies with such nonhuman agencies in sites
such as post-­Fukushima Japan, Kath Weston argues that “ecointimacies are
compositional,” born of the “growing conviction that creatures co-­constitute
other creatures, infiltrating one another’s very substance, materially and
other­wise.”1 covid-19 is that most intimate of infiltrators, absorbed through
air and breath, accelerated and intensified by both the desire to share social
and familial space, but even more so by an economic order that demands the
production and distribution of goods passed through h ­ uman hands in tightly
packed spaces in which ­people have no choice but to breathe the same air. Re-
gimes of testing, the continual monitoring for new strains, the (re)instantiation
of borders of all kinds, the clear correlation between changing climate and
new diseases—­the pandemic has forced us to confront our entanglements,
both with one another and with the nonhuman in all its technical and ecolog-
ical variety. More just and equitable f­ utures for h
­ uman life depend not only
on reckoning with covid-19 but also with the enduring crises from which
it is inseparable. Attending to the intimacies and estrangements through
which life is composed is crucial to that task. If the virus teaches us of our
constitutive entanglement within one another, it also insists on difference,
complexity, and the incommensurate opacities through which life coexists.
Produced by new cominglings of h ­ uman and animal, the novel coronavirus
emerged from the forces of expansion, extraction, and enclosure that actual-
ize the compulsions of capital and its handmaiden the state.2 But even as the
pandemic disrupted the seemingly smooth flow of goods and p ­ eople across
the globe, it also accelerated the datafication and informationalization of
life at all scales. From the profusion of Zoom meetings to the normalization
of population health surveillance to the re­distribution of carbon emissions
away from air travel to data centers and compute resources, the pandemic
has intensified the constitutive contradiction of con­temporary life between
the promise of a smooth and knowable World and the collective experience
of disjunctive, agonistic worlds. Collapsing the geopo­liti­cal into everyday
life, the stark inequities in access to covid-19 vaccines and treatments across
the globe—­not to mention the very dif­fer­ent capacity of wealthy nations to
weather the economic storms of lockdowns, deaths, and soaring health care
costs—­are in turn reflected in the classist and racist application of restrictions
within polities, backed by police and militaries. ­Here in Sydney, for example,
armed police and active-­duty soldiers w ­ ere deployed en masse in the diverse
working-­class suburbs in the southwest of the city, while residents of the
affluent east and north went largely untroubled. The biopolitics of health
management fused with an incipient necropolitics of militarized policing,
facilitated by the ontopo­liti­cal capacities of algorithmic analy­sis of the feral
transmission of the virus itself.
Politics as we know it is not equipped to deal with the intimacies of the
entangled and incommensurate, just as it is not equipped to reckon with
crises at the planetary scale. “Only a politics rebuilt on aesthetic princi­ples,
that is, by remaking communications,” writes Cubitt, “offers the possibility
of changing the conduct of relations between ­human beings and nature, and
between both of them and the technologies that so profoundly and multifari-
ously mediate between them.”3 If the neoliberal moment of racial capitalism
has produced a fragmented and ad hoc politics based around the marketiza-
tion and informationalization of life, then an alternative politics must surely
begin with communication within and across difference. As I have argued
throughout this book, nonhuman witnessing is a distinctive communicative
modality, one in which difference is not a prob­lem to be solved but rather
the grounds for flourishing. Many of the nonhuman entities and ecologies
traced in this book lack speech, or lack an inherent verbal or visual language

Toward a Politics of Nonhuman Witnessing 175

equivalent to the h­ uman, but they are nevertheless continually communicat-
ing. Uranium isotopes continue to communicate with the plants, animals,
sand, and ­peoples at Maralinga, and with the Traditional ­Owners driven from
Country. Communication such as this is aesthetic, in that it is at once sensing
and sense-­making, but it is also relational and epistemological.
Nonhuman witnessing makes a claim on what can be known, w ­ hether in
the neural layers of machine learning systems, or in the breath of the artist
blowing glass in the trail of nuclear fallout, or in the digital afterlife of the
sacred sites blasted to extract iron ore. For ­philosopher Jacques Rancière,
politics is disorder, an insistent challenge to the prevailing order, the demand
that the part without a part—­the dispossessed and denied—­obtain agency
over its own fate.4 The prob­lem at hand is not simply to expand who or what
gets to stand for parliament, but to radically reimagine the conditions within
which knowledge is made and the communicative modalities needed to
reckon with the incommensurabilities and opacities that constitute life within
worlds of infinite relation. Even as the politics of the pandemic have for the
most part been grim in the extreme, I hold on to thin hope that t­ hese condi-
tions of crisis contain within them the potential to witness worlds and what
happens within and between them in ways that enable alternative politics
in the flourishing of communication, of connection, and of relation across
difference.

­after the world, many

Many p ­ eoples and worlds know deeply the destructive force of the World:
damming rivers and flooding homelands in the name of pro­gress; clearing
bush for farmland; blasting mountain, hill, and stone to extract fossil fuels;
dispossessing ­peoples and breaking apart families; and severing ties to land,
country, and kin. Even if the World that, like Man, overrepresents itself as
the totality of existence has come to an end as a plausible or coherent notion,
its death throes continue to wrack the planet and life on it in catastrophic
ways. No reckoning has yet been made, despite the urgency. Indebted to the
Zapatista slogan “a world where many worlds fit,” de la Cadena and Blaser
describe “the practice of a world of many worlds, or what we call a plu-
riverse: heterogeneous worldings coming together as a p ­ olitical ecol­ogy
of practices, negotiating their difficult being together in heterogeneity.”5 A
pluriversal reconception of coexistence—­from World to worlds—is the task
at hand for that g­ reat swathe of humanity that has benefited from and main-

176 Coda
tained the fiction of Man. Pluriversality requires a new “­political ontology,”
a “politics of real­ity” grounded in the presumption of “divergent worldings
constantly coming about through negotiations, enmeshments, crossings, and
interruptions.”6
Pluriversality confronts a dominant politics set sharply against the very
notion of many worlds. This politics “emerged (with science) to make a live-
able universe,” writes de la Cadena, “to control conflict among a single if cul-
turally diversified humanity living in a single scientifically knowable nature.”7
This ­political field depends on divisions between friend and ­enemy, as well as
between nature and culture. As de la Cadena argues, “­These two anti­theses—­
between humanity and nature, and between allegedly superior and inferior
­humans—­declared the gradual extinction of other-­than-­human beings and
the worlds in which they existed.”8 To engage in politics, one had to be rec-
ognized within the hierarchical domain of Humanity—of Man—­and not
assigned to Nature, a form of racialization many First Nations p ­ eople have
been, and continue to be, subjected to. Pluralizing politics, then, is not simply
a question of inclusion within Man, but is to be found in the very dissolution
of such a notion to begin with. As I argued in the introduction, witnessing
has long operated as a coconspirator with Man, a guarantor to science, law,
religion, and culture of the coherence and cogency of the World. As I have
articulated the concept, nonhuman witnessing aims to break that binding of
the Witness to Man and, with it, Man to World.
This refiguring of witness and witnessing does not facilitate the smooth
aggregation of politics as usual with pluriversality but enables an adver-
sarial pluralism, in which noncontiguous and mutually exclusive worlds
can coexist—­even if coexistence requires the end of the World of Man. Co-
existence depends upon contact and relationality, not mutual exclusion.
Incommensurate worlds can only coexist when contact with irreducible
difference is the condition for a relational politics. Attending to the nonhu-
man in witnessing is one way to “slow down reasoning and provoke the kind
of thinking that would enable us to undo, or more accurately, unlearn, the
single ontology of politics,” as de la Cadena puts it.9 Nonhuman witnessing
offers the means to trace how knowledge moves between or is animated
across many worlds in a situation in which media, like all resources, are finite.
Media and mediation hold the potential to generate the connective, com-
municative tissue between worlds. For Cubitt, communication constitutes
the ground of a renewed politics, a politics that reckons with the exclusion
of the nonhuman from the forms foisted on the world through the Enlight-
enment, colonialization, and marketization. To build alternative f­ utures, the

Toward a Politics of Nonhuman Witnessing 177

nonhuman must be understood as “an active agent of historical change.”10
For this to hold, humanity must reckon with the fact “that our environments
are not only capable of communication, but are constantly communicating.”11
Communication—­and mediation more generally—­have long been too radi-
cally delimited by modernity’s insistence on the radical distinction between
­human and environment, between nature and culture. Communication must
embrace entities far beyond the ­human, not as sources of evidence or infor-
mation but as agential and vital in and of themselves. In the aftermath of the
end of the World, this communicative politics must reckon with the jostling
and at times agonistic existence of countless worlds on this one planet. Set-
ting agendas between worlds and resolving conflicts between species and
­things where outcomes might be lethal pre­sents an ­immense challenge to
such a proposition, and one that this book can only gesture ­toward. Pluriv-
ersal justice resides in the capacity for coexistence, and the active refusal of
worlds predicated on martial, algorithmic, and ecological vio­lence. Such a
­future requires a kind of faith in the incommensurate, a willingness to build
a pluriversal politics of h
­ uman and nonhuman that begins with the col-
lective witnessing of what must remain opaque, unknowable, and incom-
municable even as it seeks and nourishes connection and communication.
A nonhuman witnessing of the opacity of existence constituted by h ­ uman
and nonhuman relationality.

witnessing opacity

Tracing nonhuman witnessing across entangled crises of war, data, and

ecologies has meant repeated encounters with tensions and paradoxes. Not
bound to the ­human subject or sensorium, nonhuman witnessing necessar-
ily evokes agencies, entities, and aesthetics that cannot be readily resolved in
the ­human communicative terrain of language, gesture, and image. Against
the demand that the modern witness—­the witness of science and the courts
and the media—­have their testimony be verifiable, nonhuman witnessing re-
quires that the incommensurate and unknowable be taken as generative op-
portunities for crafting new relations and knowledges. H ­ ere we might recall
the machine vision analy­sis of military drone images, but also the resistant
potential of such systems when harnessed by the investigative aesthetics of
Forensic Architecture and applied to state vio­lence. Each of the analytic
concepts developed in this book reckon in dif­fer­ent ways with tension,
incommensurability, and unknowability. Violent mediation seeks to describe

178 Coda
how complexity, uncertainty, and the unknowable are erased and elided
through instrumental pro­cesses of mediation. Machinic affect names ­those
relational intensities that animate technoscientific apparatuses, ambivalent
to the h ­ uman and other­wise relegated to the mere operation of technical
systems. Ecological trauma describes the rippling effects of the rupturing of
relations within more-­than-­human ecologies, many of which elude ­human
understanding and can only ever be partially made sensible to the ecological
system itself. Radical absence brings t­ hese questions of the incommensurate
into the quotidian experience of the digital and its nonhuman infrastruc-
tures, accounting for encounters with what has been rendered absent yet
remains forcefully pre­sent. Th ­ ese analytics thus engage with the necessary
opacity of existence, with the fundamental incapacity for entities to disclose
themselves to one another even when bound in relation.
­Here, then, I arrive at a final doubled meaning: witnessing opacity, or the
nonhuman witnessing of opacity, and the opacity of nonhuman witnessing.
Nonhuman witnessing seeks to bring opacity into the space of witnessing, not
as a prob­lem to be resolved but as a site of potential communicative relation.
At the same time, nonhuman witnessing is constituted by its own opacity,
its presence in zones of sensing and sense-­making that cannot be decoded
or even identified at all. The dissolution of the ­human as privileged witness
depends on this potential for withdrawal from anthropocentric epistemology.
Modernity—­with its Enlightenment and colonial under­pinnings—­demands
transparency, as Glissant argues: “This same transparency, in Western His-
tory, predicts that a common truth of Mankind exists and maintains that
what approaches it most closely is action that proj­ects, whereby the world
is realized at the same time that it is caught in the act of its foundation.”12
Opacity works against this “reductive transparency.”13 It is not obscurity but
rather “that which cannot be reduced, which is the most perennial guarantee
of participation and confluence.”14 Opacity emerges with and is the condition
of new and old worlds alike. Opacity does not produce irreconcilable differ-
ence between cultures, languages, or ways of living but rather makes pos­si­ble
the coexistence of multiplicities within a totality. “Opacities can coexist and
converge, weaving fabrics,” Glissant writes. “To understand ­these truly one
must focus on the texture of the weave and not on the nature of its compo-
nents.”15 This weave is Relation, or “what the world makes and expresses of
itself.”16 Glissant’s opening to Relation invokes “a poetics that is latent, open,
multilingual in intention, directly in contact with every­thing pos­si­ble,” but in
his account is very much tied to ­human subjectivities and the traumas they
experience, particularly t­ hose of slavery’s M
­ iddle Passage. Glissant’s right to

Toward a Politics of Nonhuman Witnessing 179

opacity is itself a response to dehumanizing vio­lence and is, in a sense, al-
ready a politics of witnessing traumas that leave no trace, lost to the oceans.
For Glissant, this unrecoverable trauma is both an end and beginning; a
nonpassage for the drowned and an abyssal beginning for ­those who survive.
Opacity thus arises from this unknowable trauma and from the contact
between worlds that it sets in motion. Racialization, in other words, is not a
by-­product of the World of Man, but rather its constitutive force that renders
certain p ­ eoples nonhuman, producing both World and Man through that
dehumanization. Responding to the historical enactment of violent trauma,
Glissant’s opacity offers a way of understanding the Other that does not re-
quire the relinquishing of Otherness. As such, it provides a generative way of
thinking relations with more-­than-­human ecologies and technics that do not
require their submission to ­human forms of knowing or being, but depend
instead on their openness to communication, justice, and the flourishing
of other worlds. In ­doing so, it ­will be necessary to reckon with the incom-
mensurabilities of weaponized drones and military ai, for example, and ask
­whether in the coexistence of worlds such technics should continue. Witness-
ing opacity is itself a ­political proj­ect, with ­political strug­gles inherent to its
articulation beyond ­these pages.17
Against the notion that transparency is the necessary antidote to difference
between beings and worlds, witnessing opacity helps enable a pluriversal poli-
tics. Nonhuman witnessing—as an ethicopo­liti­cal, aesthetic, and epistemic
mode of relation—­provides the potential for a transversal communicative
politics, one that works within and between a pluriverse of worlds. Nonhu-
man witnessing offers an aesthetics of rupture and repair, of connection
and disconnection. The politics at hand h ­ ere are not concerned with policy
per se, or with the demo­cratic ­organization of ­human socie­ties. I am not
proposing that nonhuman witnessing enable a parliament of ­things. Rather,
my proposition is that nonhuman witnessing be mobilized in the name of a
politics of the dispossessed, of the ­human and nonhuman, of ­those denied
humanity and denied agency. Such a politics must contend with the incom-
mensurate at e­ very level, precisely b ­ ecause t­ hose denied p ­ olitical standing
within the World of Man are also deemed to lack transparency or its potential
when in fact what they possess is an unassailable opacity. Justice ­isn’t made
by enforced transparencies and disclosures, but through contact between
opacities that are nonetheless generative of shared knowledges. Nonhuman
witnessing’s p­ olitical potential resides in how a field of relations—­human
and non; technical, cultural, and ecological—­composes itself in the face of
the injunction that witnessing makes, an injunction to become communica-

180 Coda
tive, to become both response-­able and address-­able even while holding the
refusal that resides in the right to opacity.

the politics of nonhuman witnessing

Nonhuman witnessing seeks to bring into being the conditions for an other­
wise by producing communicative relations across and within difference
that refuse to override the opaque and the incommensurate. Nonhuman
witnessing is an ecological mode of communication that arises from the fields
of relations that come together in the encounter between ­human and nonhu-
man, and most intensely so in contexts of vio­lence, domination, and control.
By refusing the supremacy of Man the Witness as the figure through which
events obtain meaning or knowledge is produced, nonhuman witnessing
gives standing to diverse actors and entities, w
­ hether p­ eople denied humanity
or machinic intelligences or wounded ecologies in the aftermath of war. What
the nonhuman bears witness to might well be ruin, death, and trauma—­and
the witness itself might be a perpetrator—­but the fundamental implication
of nonhuman witnessing is to remake the h ­ uman and the witnessing that we
do. Nonhuman witnessing can be mobilized to heal and empower, to bring
to light change in its emergence, and to insist on attending to voices, bodies,
patterns, and materialities denied standing in the pre­sent order. Nonhuman
witnessing is not a panacea, but rather a practice of forging relations with the
incommensurate. Its lure is becoming more h ­ uman through the witnessing
of our constitutive nonhumanity.
The politics of nonhuman witnessing, then, is not one of rights, ­human
or other­wise. Expanding the domain of rights—­granting rights to rivers and
other earth beings, for example—is a worthy enough endeavor but not one
that changes the conditions ­under which politics takes place. If a machine
­were to bear witness as a rights-­bearing subject, what rights would obtain to
it and what would their articulation mean for the rights that already accrue
to the “­human”? Rights, for all the protections they provide, are part and
parcel of the existing order of racial capitalism and neoliberal governance,
guarantors of h­ uman privilege and individual autonomy within the epistemic
domain of the Enlightenment. Rather than extending rights that h ­ umans
have to the nonhuman, the task at hand is to invite nonhumans subjectivities
and agencies into the space of politics and, in d ­ oing so, seek to recompose
what politics is for the ­human. Cubitt again: “It is we ourselves who must
become other in order to produce an other world. The correlative is that we

Toward a Politics of Nonhuman Witnessing 181

must cease to be h ­ uman, and most of all cease to exist as exclusively h ­ uman
polity, which is the medium of communication par excellence. The road to
that goal, however, must lead through the polis, the humanity of h ­ umans,
and most of all through our communications in order to imagine a way out
of stasis.”18 Such a politics nurtures a radical solidarity between ­human and
nonhuman, nourished by a shared capacity to witness vio­lence and won­
der, trauma and healing, and to do so in and across incommensurate time,
space, scales, subjectivities, and materialities. Th
­ ere is no blueprint for such
a politics, no white paper or policy guidance. It is a politics that can only
become thinkable in its particularities through the poesis of its emergence.
That emergence ­will produce its own challenges, not least ­those of setting
agendas, establishing the grounds for lethality, and resolving conflicts. But
to impose its forms in advance would be to foreclose ­futures that cannot be
­imagined from within the epistemic dominance of the Anthropos. At this
critical conjuncture of history, the transformative potential of the politics of
nonhuman witnessing remains unrealized.
­There are already and have always been many worlds, both prior to and
existing alongside or in the shadow of the World of the Anthropos, the World
of the Anthropocene. First Nations worlds, but also worlds on the periphery
of empire, or in the underbelly of cities, or in speculative f­ utures, or in fugi-
tive subjects escaping constraints of all kinds. Worlds of nonhuman beings,
of animals and bacteria and plants and rocks. Worlds of earth beings, as de
la Cadena describes the animacies of mountains, lakes, and forests.19 And
now worlds of technical agencies, and even—­against the odds and despite
the hidden h ­ uman l­ abor that often powers them—­machinic intelligences and
perceptual machines. Not all of ­these worlds would be readily recognized
as such, and it might well be that the existence of some constitutes a risk to
the existence of ­others. Yet such worlds jostle, cohabiting terrains and atmo-
spheres, competing for energy, voice, space, and even time.
To begin with a ­political ontology that allows for such pluriversality is
itself a radical move, since it means the End of the World, and of life and
politics as we know it. “The idea of a pluriverse is utopian indeed,” writes de la
Cadena, “not ­because other socionatural formations and their earth-­practices
do not take place, but b ­ ecause we have learned to ignore their occurrence,
considering it a ­thing of the past or, what is the same, a m
­ atter of ignorance
and superstition.” Like all utopian proj­ects, a world of many worlds ­can’t
20

wind the clock back, undo the damage, or raise the dead. But a world of
many worlds does require a communicative modality that reaches ­toward
the incommensurability of crowding worlds, even as it re­spects the necessity

182 Coda
of ineradicable difference. Rather than rights or demo­cratic participation,
the politics of nonhuman witnessing concerns the emergent composition of
fields of relations out of which incommensurate collectivities and paradoxi-
cal knowledges might form. The politics of nonhuman witnessing is, in this
sense, an ecological poesis, an attunement to and calibration of the ­human
and the nonhuman that dwells in and with opacity. It is a politics of and for
the ­future, even as it provides the means to reimagine the past.
The politics of nonhuman witnessing is a politics of the commons, but
not the commons in a universal, global, or homogenous sense—­rather it is
a profusion of commons, bound by their common commitment to neither
begin with nor seek to resolve homogeneity.21 Such a commons can only ever
be emergent and unfixed, since it must compose itself a new in the ongo-
ing antagonisms, negotiations, sympathies, and alliances between worlds.
Commons are necessarily communicative. Nonhuman witnessing offers the
potential for a distinct communicative mode, one that insists not simply
upon communication but on the demand for response and address. Such
terms carry with them a certain anthropocentrism, but in adopting them I
am not returning to narrow notions of speech or recognition. Address and
response form instead a communicative relation and generative aesthetic.
Fuller and Weizman describe the emergence of an investigative commons in
the new collectivities of forensic architecture, open-­source investigation, and
distributed ­human rights research, which in turn draws on the existence of an
aesthetic commons, in which pro­cesses of sensing and sensing-­making fold
into further such pro­cesses.22 If nonhuman witnessing animates or emerges
within par­tic­u­lar commons, it also does so at the level of aesthetics and in
league with such instrumental investigative modes. But it also exceeds t­ hose
deliberate, ­human interventions, describing too the poesis that can arise in
the strange agonisms and fleeting alliances of machines, ecologies, animals,
and ­people.
To return to the Pacific Forum that opened chapter 3 of this book, nonhu-
man witnessing might galvanize a commons of islands and oceans, ­people
and winds, garbage and atmospheric sensing. Nonhuman witnessing would
not paper over the incapacities of speech or the ephemerality of certain agen-
cies but would be alive to what emerges in the intensive connections that can
arise when worlds are anchored, nurtured, and fought for. It is for this rea-
son that I have attended in this book not only to material events and actu-
ally existing technologies, but also to speculative imaginaries and creative
works. Such phenomena, objects, practices, and pro­cesses are often not at
all contiguous or willing to reveal their workings. Nor should they be. What

Toward a Politics of Nonhuman Witnessing 183

they reveal is the contingent and always incomplete nature of nonhuman
witnessing, how its politics depend on the work of refiguring the ­human in
the face of catastrophic crises.
This book is, too, a necessarily incomplete gesture. A pursuit of some-
thing happening all around us yet refusing to be fixed in place. War, in its
all its turbulent and violent becoming, now escapes the ­human more than
ever before, yet it cannot and w­ ill not leave us b
­ ehind. Algorithmic and data
technologies enclose life and seek to make it operative. Ecological catastrophe
pushes the planet itself to the brink of becoming unlivable. In the shadow of
what refuses to be grasped, that ­won’t submit to contained and discrete ways
of knowing, ­can’t sit still long enough to become knowable to the ­human
alone, nonhuman witnessing widens the ambit through which meaning
comes to m ­ atter, responsibility is forged, and more-­than-­human epistemic
communities become pos­si­ble. Yet while its politics are never far from the
surface, their form and force remain to be realized in the work that lies before
us, ­humans and other­wise.

184 Coda
notes

Introduction. Nonhuman Witnessing

1 Like many, I first encountered the horrifying strike of February 21, 2010,
through the excerpts from the transcript that open Grégoire Chamayou’s
Drone Theory, but it also plays a similar role in journalist Andrew Cockburn’s
Kill Chain and in international relations scholar Lauren Wilcox’s brilliant
analy­sis of racializing and gendering in the drone apparatus. In writing this
account I am deeply indebted to the remarkable work of Derek Gregory,
particularly a series of posts on his Geo­graph­i­cal Imaginations site titled
“­Under Afghan Skies.” In my rendition, I have sought to attend to the points
of contact between technoscientific systems, ­human actions, and the environ-
ment. Chamayou, Drone Theory; Cockburn, Kill Chain; Wilcox, “Embodying
Algorithmic War.”
2 Whyte, “Indigenous Science (Fiction) for the Anthropocene.”
3 Wynter, “Unsettling the Coloniality of Being/Power/Truth/Freedom.”
4 Wynter differentiates between an original Man1 arising from the ­Renaissance
and a l­ater Man2 reshaped by the colonial encounter with the Amer­i­cas
and Darwinian biology, but as the distinctions between ­these two positions
are not central to my argument, I am using “Man” to encapsulate this over-­
representation as the ­human in more general terms.
5 I understand affect in the Spinozan tradition: as the bodily capacity to af-
fect and be affected. ­Here, as elsewhere in my work, affect gels with a
c­ apacious conception of what counts as a body: texts, plants, drones, swans,
databases, publics and ­waters can all be bodies as readily as individuated
­humans. Bodies—­human and non, multiple and individuated—­are enabled
and constrained by the relations in which they are webbed, the resources
and capacities for change, connection, signification and more. ­Those rela-
tions are the stuff of affect. Affect is thus not an exclusively ­human mode of
experience akin to emotion, but the relational dynamics of bodies situated in
contexts, entrained within environments, assembled in machinic pro­cesses,
and so on. As Massumi writes, “The body is as immediately abstract as it is
concrete; its activity and expressivity extend, as on their underside, into an
incorporeal, yet perfectly real, dimension of pressing potential.” Parables, 31.
Affect can be modulated, amplified, intensified and other­wise transformed by
media: this is precisely what makes ubiquitous media so power­ful. But affect
can break, sheer, stretch, distend and rupture bonds between bodies as much
as strengthen or intensify them. See Massumi, “Autonomy of Affect”; Gibbs,
“Panic!”; Gibbs, “Contagious Feelings”; Angerer, Ecol­ogy of Affect; Schae-
fer, Evolution of Affect Theory; Gregg and Seigworth, Affect Theory Reader;
Clough, Affective Turn; Deleuze, “Ethology.”
6 De la Cadena and Blaser, A World of Many Worlds.
7 Amazon, “All In”; Greenpeace, “Clicking Clean ­Virginia.”
8 Sherwood, “Inside Lithium ­Giant sqm’s Strug­gle to Win Over Indigenous
Communities in Chile’s Atacama.”
9 Gilmore, Golden Gulag; Robinson, Black Marxism.
10 Kelley, “Racial Capitalism.”
11 Packer and Reeves, Killer Apps, 5. On the history of computation, screens, and
­enemy production, see also Geoghegan, “An Ecol­ogy of Operations.”
12 Puar, Right to Maim; Mbembe, Necropolitics.
13 Browne, Dark ­Matters.
14 Halpern, Beautiful Data, 8.
15 Furuhata, “Multimedia Environments and Security Operations,” 72.
16 Edwards, A Vast Machine; Edwards, The Closed World.
17 Hogan and Vonderau, “The Nature of Data Centers.”
18 On infrastructural vio­lence, see Khalili, Sinews of War and Trade; Cowen,
The Deadly Life of Logistics; and Easterling, Extrastatecraft. On slow and
structural vio­lence, see Nixon, Slow Vio­lence; Farmer, “Structural Vio­lence.”
On algorithmic vio­lence, see Bellanova et al., “Critique of Algorithmic
Vio­lence.”
19 Whyte, “Against Crisis Epistemology.”
20 Vigh, “Crisis and Chronicity,” 10.
21 Berlant, Cruel Optimism, 10.
22 Murphie, “On Being Affected,” 24.
23 Murphie, “On Being Affected,” 20.
24 Massumi, Ontopower, vii–viii.

186 Notes to Introduction

25 As such, “politics, colonialism, settlement, capitalism, ecological destruction,
racism, and misogynies are not wars by other means—­they are war.” Grove,
Savage Ecol­ogy, 61.
26 Grove, Savage Ecol­ogy, 10.
27 Cubitt, Finite Media, 3; and Cubitt, The Practice of Light.
28 Kember and Zylinska, Life ­after New Media, xv. On mediation as performa-
tive and lively enactment, see also Parks, Rethinking Media Coverage, 2.
29 Drawing on German and Canadian media theory, John Durham Peters
defines media as “ensembles of natu­ral and ­human ele­ments” and “our infra-
structures of being, the habitats and materials through which we act and are.”
Andrew Murphie argues that “we move into relations with media that quite
literally move us/the world and with which we can move the world.” Even
the planet can be understood as a medium, argues Chris Russill, most clearly
in the monitoring apparatuses of climate change and atmospheric military
sensor networks. Peters, The Marvellous Clouds, 3, 15; Murphie, “World as
Medium,” 17; Russill, “Is the Earth a Medium?”; Russill, “Earth Imaging.”
30 Cubitt, Finite Media, 7.
31 Russill, “The Road Not Taken.”
32 Patrick Wolfe has written persuasively about the relationship between settler
colonialism and genocide, arguing that “settler colonialism is inherently
eliminatory but not invariably genocidal” in “Settler Colonialism and the
Elimination of the Native,” 388. ­Today, Aboriginal ­children are currently
being removed from their families at even higher rates than they w ­ ere in
the 1950s, an era of systemic destruction of culture and ­peoples that made
what became known as the Stolen Generation and for which the Australian
government issued a formal apology in 2008. Cashless debit cards to control
welfare benefits are being rolled out to Indigenous communities, many still
living ­under permutations of the military occupation launched in 2007 by an
ailing government looking to stoke anti-­Indigenous resentment. Black deaths
in custody have returned to their heights from the 1980s, while Aboriginal
­people are massively overrepresented in prisons and Indigenous ­women are
far more likely to be subject to sexual vio­lence.
33 Whyte, “Indigenous Science (Fiction) for the Anthropocene,” 226.
34 Estes, Our History Is the F­ uture, 135.
35 De la Cadena and Blaser, A World of Many Worlds, 12.
36 Although, as of writing, it has not yet been recognized by ­either the Interna-
tional Commission on Stratigraphy nor the International ­Union of Geological
Sciences as a division of geologic time.
37 Demos, Against the Anthropocene, 85. For a detailed discussion of Crutzen and
Stoermer’s conception of the Anthropocene, see Grove, Savage Ecol­ogy, 36–40.
38 Moore, Anthropocene or Capitalocene?; Grove, Savage Ecol­ogy; Haraway,
“Anthropocene, Capitalocene, Plantationocene, Chthulucene”; de la Cadena,
“Uncommoning Nature.”

Notes to Introduction 187

39 Writing on what she calls “minimal ethics,” Joanna Zylinska argues
that the response demanded by the Anthropocene is “strongly post-­
anthropocentric . . . in the sense that it does not consider the ­human to be
the dominant or the most impor­tant species, nor does it see the world as
arranged solely for ­human use and benefit.” Zylinska, Minimal Ethics for the
Anthropocene, 20.
40 Davis and Todd, “On the Importance of a Date,” 763. By contrast, see, for
example, McNeill and Engelke, The ­Great Acceleration.
41 Stewart, “Atmospheric Attunements,” 445.
42 Kaplan, Aerial Aftermaths, 17–18.
43 Or, as as Kathryn Yusoff points out, if the end of the world of apocalyptic
imagining is the end of this colonialist and cap­i­tal­ist one for some, then it is
also “the prerequisite for the possibility of imagining ‘living and breathing
again’ for ­others.” Yusoff, A Billion Black Anthropocenes or None, 13.
44 Rather than look outside the narrow frame of knowledges that helped get us
into this mess—­loosely grouped ­under the rubrics of the humanist tradi-
tion and Western scientific rationality—we are called to attend ever more
earnestly to ­those very knowledges in the search for solutions to the damage
they have wrought. See Snaza, “The Earth Is Not ‘Ours’ to Save,” 339.
45 Weizman, Forensic Architecture.
46 Fuller and Weizman, Investigative Aesthetics.
47 Schuppli, Material Witness, 19.
48 Schuppli, Material Witness, 34.
49 Pugliese, State Vio­lence and the Execution of Law, 14.
50 Peters, “Witnessing.”
51 Frosh and Pinchevski, “Why Media Witnessing?,” 1.
52 Chouliaraki, The Spectatorship of Suffering.
53 See, for example, Reading, “Mobile Witnessing”; Andén-­Papadopoulos, “Citi-
zen Camera-­Witnessing”; Andén-­Papadopoulos, “Crowd-­Sourced Video”;
Papailias, “Witnessing in the Age of the Database”; Chouliaraki, “Digital
Witnessing in Conflict Zones”; Gray, “Data Witnessing.”
54 Wu and Montgomery, “Witnessing in Crisis Contexts”; Rae, Holman, and
Nethery, “Self-­Represented Witnessing”; Ristovska, Seeing H ­ uman Rights.
55 For more on torture in ancient Greece and Rome, see Ballengee, Wound and
Witness; DuBois, Torture and Truth; Peters, Torture.
56 On the religious roots of witnessing, see Peters, “Witnessing,” 708. On the
emergence of new ­legal norms, see Langbein, Law of Proof.
57 Ballard, “And They Are like Wild Beasts,” 19. See also Bennett, Vibrant ­Matter,
8–10.
58 Langbein, Law of Proof.
59 Spillers, “Mama’s Baby, Papa’s Maybe,” 69.
60 Dispossession, stigmatisation and deprivation of First Nations served “both
to ‘verify’ the overrepre­sen­ta­tion of Man as if it ­were the ­human, and to

188 Notes to Introduction

legitimate the subordination of the world and well-­being of the latter to ­those
of the former.” Wynter, “Unsettling the Coloniality of Being/Power/Truth/
Freedom,” 268.
61 Law, as Zakkiyah Jackson explains, “denies ­those it deems ‘inhuman’ access
to speech and law, thereby producing the inhumanity it excludes,” showing
how the standard of the “­human” witness is actually “fundamental to law’s
injustice for both ­people of color and animals.” Jackson, “Animal,” 675, 676.
62 Fassin, “The Humanitarian Politics of Testimony,” 534, 541.
63 Daston and Galison, Objectivity.
64 On the relationship of science, testimony, and the production of knowledge
in a classic vein, see Coady, Testimony. On the role of trust and social con-
formity, see Adler, “Testimony, Trust, Knowing”; and Hardwig, “The Role of
Trust in Knowledge.”
65 Daston and Galison, Objectivity.
66 Haraway, Modest−Witness, 24. In her book, Haraway proposes her own femi-
nist modest witness who is suspicious, situated, knowing, ignorant, partial,
and more.
67 Haraway, Modest−Witness, 24.
68 Haraway, “Situated Knowledges.”
69 “Witnessing became a domestic act. . . . ­Television sealed the twentieth
­century’s fate as the ­century of witness.” Ellis in Peters, “Witnessing,” 708.
70 Frosh and Pinchevski, “Why Media Witnessing?,” 9.
71 Peters, “Witnessing.”
72 Boltanski, Distant Suffering; Chouliaraki, The Spectatorship of Suffering;
Kozol, Distant Wars Vis­i­ble.
73 On the significance of witnessing and testimony to the a host of scholarly
disciplines and institutional practices, see Wieviorka, The Era of the Witness;
Felman and Laub, Testimony; Frosh and Pinchevski, Media Witnessing; Givoni,
“Witnessing/Testimony”; Guerin and Hallas, The Image and the Witness.
74 The German media theorist Friedrich Kittler was the leading proponent of
this argument, which I am drawing on ­here by way of Sean Cubitt. See Cubitt,
The Practice of Light, 7–12.
75 For the distilled argument, see the introduction to Cubitt, The Practice of Light.
76 As Halpern puts it, ­there emerged “a new set of investments in ­process, com-
munication, and circulation, now encoded into built environments, ma-
chines, and attention spans.” See Beautiful Data, 84.
77 Gray, “Data Witnessing.”
78 McCosker, “Drone Media”; McCosker, “Drone Vision”; Andén-­
Papadopoulos, “Citizen Camera-­Witnessing”; Chouliaraki and al-­Ghazzy,
“Flesh Witnessing”; Gil-­Fournier and Parikka, “Ground Truth to Fake
Geographies.”
79 The limits of repre­sen­ta­tion in witnessing has been a significant theme in
my own research, see Richardson, Gestures of Testimony; Richardson and

Notes to Introduction 189

Schankweiler, “Affective Witnessing”; Richardson, “Drone’s-­Eye View.” See
also, Pinchevski, Transmitted Wounds; Ettinger, The Matrixial Borderspace.
80 Oliver, Witnessing, 195, 20.
81 Givoni, “Witnessing/Testimony,” 165.
82 ­There are now many valuable interventions in how to rethink the ­human.
Dana Luciano and Mel Y. Chen draw on queer theory to argue that the
­“inhuman”—or inhumanisms—­“points to the vio­lence that the category of the
­human contains within itself ” by “resonating against the ‘inhumane’ ” in “Has
the Queer Ever Been ­Human?,” 197. In alliance with this queer inhumanism,
Julietta Singh calls for dehumanism, a “practice of recuperation, of stripping
away the violent foundations (always structural and ideological) of colonial
and n ­ eocolonial mastery that continue to render some being more ­human
than o ­ thers.” Singh, Unthinking Mastery, 4. Like the ­earlier “antihumanism” of
Louis Althusser and the critical Marxist tradition, the “de-” and “in-” signal
a critical desire to undo the structures of knowledge-­making and world-­
making of the humanities in their traditional form, but are not so geared
­toward the expansive account of agencies needed to think about witnessing
outside the frame of the ­human altogether, which is why some prefer the
“more-­than-­human” and “other-­than-­human.” See Springgay and Truman,
Walking Methodologies, 8–11; Pugliese, Biopolitics of the More-­Than-­Human.
83 Grusin, “Introduction,” ix–­x.
84 “Humanity and nonhumanity have always performed an intricate dance with
each other. ­There was never a time when ­human agency was anything other
than an interfolding network of humanity and nonhumanity; ­today this min-
gling has become harder to ignore.” Bennett, Vibrant ­Matter, 31.
85 Race in general, and blackness in par­tic­u­lar, “cannot be escaped but only
disavowed or dissimulated in prevailing articulations of movement ‘beyond
the ­human.’ ” Jackson, “Outer Worlds,” 216. One consequence of this think-
ing, as Tavia Nyong’o points out, is that “posthumanist theory has tended to
pre­sent the decentering of the ­human as both salutary and largely innocent of
history.” Nyong’o, “­Little Monsters,” 266.
86 Barad, “Posthumanist Performativity,” 809. Stacy Alaimo describes this as
“trans-­corporeality,” or how “the ­human is always intermeshed with the
more-­than-­human world.” Alaimo, Bodily Natures, 17.
87 Luciano and Chen, “Has the Queer Ever Been ­Human?,” 192.
88 Sheikh, “The ­Future of the Witness,” 148.
89 Zylinska, Nonhuman Photography, 15.
90 Cubitt, Finite Media, 151.
91 Vivian, Commonplace Witnessing.
92 See, for example, Caruth, Unclaimed Experience; Felman and Laub, Testi-
mony; Kaplan, Trauma Culture; Abraham and Torok, Shell.
93 Felman writes that “testimony seems to be composed of bits and pieces of a
memory that has been overwhelmed by occurrences that have not settled

190 Notes to Introduction

into understanding or remembrance, acts that cannot be constructed as
knowledge nor assimilated into full cognition, events in excess of our frames
of reference.” Felman and Laub, Testimony, 5.
94 LaCapra, Writing History; LaCapra, Representing.
95 Felman and Laub, Testimony, 24.
96 Agamben, Remnants, 141.
97 Arendt, On Vio­lence, 50, 51.
98 Butler, Precarious Life.
99 Farmer, “Structural Vio­lence,” 308. The term “structural vio­lence” originates
with peace and conflict studies ­founder Johan Galtung but has been elabo-
rated and extended by Farmer and ­others, particularly in anthropology.
100 Das et al., Vio­lence and Subjectivity.
101 Wolfe, “Settler Colonialism and the Elimination of the Native.”
102 Michael Taussig’s account of paramilitary vio­lence in Colombia is a particu-
larly vibrant and nuanced account of the latter. Taussig, Law in a Lawless
Land.
103 Reflecting on her own refusal to revivify spectacles of tortured slaves,
Saidiya Hartman pursues instead the “diffusion of terror and vio­lence per-
petrated ­under the rubric of ­pleasure, paternalism and property.” Hartman,
Scenes of Subjection, 4.
104 Da Silva, “No-­Bodies,” 214.
105 Weizman, Forensic Architecture, 117.
106 Oliver, Witnessing, 7.
107 Oliver, Witnessing, 7.
108 Oliver, Witnessing, 7.
109 Oliver, Witnessing, 68.
110 Oliver, Witnessing, 223.
111 See Chow, Age of the World Target; Bousquet, The Eye of War.
112 Oliver, Witnessing, 224.

Chapter One. Witnessing Vio­l ence

1 As Pailthorpe’s work attests, aesthetic interventions into drone warfare can
possess an ambivalent relation to their subject ­matter. Stubblefield argues
that once drone art is restated in relation to the operative logics and pro­
cesses of networked war, “a more nuanced reading emerges, one in which
the apparent passivity of this genre is not only a conscious response to the
specific conditions of drone power, but in fact the means for reimagining its
relations of vio­lence,” Stubblefield, Drone Art, 2. Further valuable commen-
taries on drone art can be found in Rhee, The Robotic Imaginary; Danchev,
“Bug Splat”; Bräunert and Malone, To See Without Being Seen.
2 Cubitt, Finite Media, 4.
3 Cubitt, The Practice of Light, 8.

Notes to Chapter One 191

4 Packer and Reeves, Killer Apps, 9.
5 For an extensive examination of the relationship between scientific epistemes
and warfare, see Bousquet, The Scientific Way of Warfare.
6 Packer and Reeves, Killer Apps, 7.
7 Bousquet, The Eye of War, 8, 11.
8 In a similar vein, Amoore observes that “algorithmic war appears to make
it pos­si­ble for the imagination of an open global economy of mobile ­people,
objects and monies, to be reconciled with the post-9/11 rendering of a securi-
tized nation-­state.” Amoore, “Algorithmic War,” 51.
9 All too ­human testimony takes place outside the strictures of rights-­based
discourses, of course. In ­Afghanistan, for example, the weaving of war rugs to
represent the weaponry and martial dynamics of successive superpower inva-
sions can be understood as a form of testimony.
10 Tahir, “The Ground Was Always in Play,” 12.
11 Dawes, That the World May Know; Slaughter, ­Human Rights, Inc.
12 Givoni, The Care of the Witness.
13 Jess Whyte argues that neoliberals, threatened by socialist pushes for rights
to social welfare and self-­determination, set about developing an alternate
framework of ­human rights that had the explicit intention of depoliticizing the
social realm in order to protect private investment, strengthen liberal individu-
alism, encourage market-­based solutions, extend imperialism, and entrench a
marketcentric international order. Whyte, The Morals of the Market.
14 Kaplan calls them “hybrid machines” that “make their worlds, materially
and elementally” as “complex assemblages that gather and produce ­people,
objects, discourses, information, terrains, atmospheres, and, in ways we must
make more complicated, images.” Kate Chandler argues that the unmanning
at the center of drone warfare elides its coconstitution by ­human, media and
machine ele­ments that “not only come together but also fail to cohere and the
attendant politics that emerge through ­these relations,” Chandler, Unman-
ning, 7; Kaplan, “Drones and the Image Complex,” 30.
15 McCosker and Wilken, Automating Vision, 89.
16 Parks and Kaplan, Life in the Age of Drone Warfare, 9.
17 See Kaplan, Aerial Aftermaths; Chandler, Unmanning; Wilcox, “Embodying
Algorithmic War.”
18 Kaplan, “Drone-­o-­Rama.”
19 Grégoire Chamayou argues that this “temporary autonomous zone of
slaughter,” enacts “the idea of an invasive power based not so much on the
rights of conquest as on the rights of pursuit: a right of universal intrusion
or encroachment.” Chamayou, Drone Theory, 55, 53. When first deployed in
the 1991 Gulf War, the kill box was an area the size of New York City, but as
both the concept and necessary geolocation and information communication
technologies developed, kill boxes became three dimensional and increas-
ingly granular in time and geo­graph­i­cal specificity.

192 Notes to Chapter One

20 Bousquet, Grove, and Shah, “Becoming War,” 1.
21 Bousquet, Grove, and Shah, “Becoming War,” 5.
22 Gregory, “From a View to a Kill,” 193. See also Grayson and Mawdsley,
“Scopic Regimes in International Relations.”
23 See Gregory, “From a View to a Kill”; Michel, Eyes in the Sky.
24 “Drone vision intervenes at the point of generating a new actionable kind
of visibility or visual knowledge.” McCosker and Wilken, Automating
­Vision, 99.
25 Andrejevic, Automated Media, 85.
26 Parks, Rethinking Media Coverage, 146.
27 Parks, Rethinking Media Coverage, 147.
28 For an outstanding history, see Michel, Eyes in the Sky.
29 On the history and con­temporary practices of photogrammetry and vertical
mediation, see Wilken and Thomas, “Vertical Geomediation.”
30 Pong, For the Duration, 1.
31 Although Behram’s photo­graphs are often disbelieved and dismissed by sup-
porters of the war. See Tahir, “The Ground Was Always in Play,” 14.
32 Dorrian, “Drone Semiosis,” 55.
33 For Stubblefield, Dorrian’s observations are the hinge into reading the photo­
graphs as an inversion of drone vision. Stubblefield, Drone Art, 31.
34 Kapadia, Insurgent Aesthetics, 54–60.
35 Kaplan, Aerial Aftermaths, 33.
36 Kaplan, Aerial Aftermaths, 36.
37 In the Stanford/nyu “Living ­Under Drones” report, Hisham Abrar states:
“When the weather is clear, three or four [drones] can be seen. . . . They are in
[103.158.136.64] Project MUSE (2024-02-06 19:57 GMT) University of Edinburgh

the air 24 [hours a day], seven [days a week], but not when it’s raining. E ­ very
time they are in the air, they can be heard. And ­because of the noise, ­we’re
psychologically disturbed—­women, men, and ­children. . . . When ­there ­were
no drones, every­thing was all right. [­There was] business, ­there was no psy-
chological stress and the ­people did what they could do for a living.” Stanford
Law School and nyu School of Law, “Living ­under Drones,” 164.
38 Stanford Law School and nyu School of Law, “Living ­under Drones,” 81.
39 Edney-­Browne, “The Psychosocial Effects of Drone Vio­lence,” 1347.
40 Schuppli, Material Witness, 124.
41 Kapadia, Insurgent Aesthetics, 69–72.
42 Kapadia, Insurgent Aesthetics, 5.
43 Chishty’s shadowy drones share some loose affinity with James Bridle’s
identically named and more well-­known Drone Shadows (2012–18), in which
1:1 outlines of Reaper and Predator drones are painted onto public spaces in
cities such as London and New York, insisting on the return of the presence
of the drone to the places that authorize their deployment. Jennifer Rhee
provides an excellent critique of the limitations of the politics of identification
enacted by Bridle’s works, demonstrating how such works rest on an obscur-

Notes to Chapter One 193

ing or forgetting of the racialized subjects of vio­lence within ­those polities.
Rhee, The Robotic Imaginary, 144–54.
44 Pakistan’s Federally Administered Tribal Areas (fata) operated u ­ nder a sepa-
rate ­legal and ­political regime to the rest of Pakistan, which is itself governed
by structures inherited from the British colonial era. The securitization of
fata operates in two directions: by the United States and by the Pakistani
state. As Madiha Tahir has powerfully argued, this vio­lence (of torture,
disappearance, and secrecy) constitutes part of the “distributed American
empire,” in which the state and military enact their own desires for power
and control. Tahir, “The Distributed Empire of the War on Terror.”
45 Pugliese, “Death by Metadata.”
46 Andrejevic, Automated Media, 86.
47 Stanford Law School and nyu School of Law, “Living ­under Drones,” 113.
48 Pugliese, “Drone Casino Mimesis.”
49 Stanford Law School and nyu School of Law, “Living ­under Drones,” 74.
50 Pugliese, Biopolitics of the More-­Than-­Human, 191. Pugliese quotes a Yemeni
reporter: “Animals died, and the bodies of all ­those who died ­were disin-
tegrated and scattered over a large area. . . . They ­were all exploded, and we
could not identify them, their limbs ripped apart.”
51 Da Silva, “No-­Bodies”; Pugliese, “Death by Metadata,” 11, 15. Of course, life in
the Afghanistan-­Pakistan borderlands remains much more complex and vi-
brant than the view through the drone suggests. See Bashir and Crews, ­Under
the Drones.
52 Stanford Law School and nyu School of Law, “Living ­under Drones,” 108.
53 Pugliese, Biopolitics of the More-­Than-­Human, 22.
54 See Pugliese, Biopolitics of the More-­Than-­Human, 197.
55 Parks, Rethinking Media Coverage, 9.
56 Mirzoeff, How to See the World; Greene, “Drone Vision.”
57 Stahl, Through the Crosshairs, 68.
58 Not all drone witnessing is of aerial assault. In so-­called overwatch missions,
drone sensors might instead capture the vio­lence of ground combat.
59 Starosielski, Media Hot & Cold, 168.
60 Fish, “Blue Governmentality.”
61 Herscher, “Surveillant Witnessing.”
62 un News, “ ‘Crimes of Historic Proportions’ Being Committed in Aleppo,
un Rights Chief Warns.”
63 Kurgan, “Conflict Urbanism, Aleppo,” 74.
64 For a detailed reflection on this practice, see Kurgan, Close up at a Distance.
65 Weizman, Forensic Architecture, 28. The current 0.25m resolutions are for
black-­and-­white imagery, with 1.6m limits on color.
66 Saldarriaga, Kurgan, and Brawley, “Visualizing Conflict,” 104.
67 Gil-­Fournier and Parikka, “Ground Truth to Fake Geographies.”
68 Kurgan, “Conflict Urbanism, Aleppo,” 76.

194 Notes to Chapter One

69 Chouliaraki and al-­Ghazzy, “Flesh Witnessing.”
70 Kaplan, Aerial Aftermaths, 8.
71 On drone-­human sensoria, see Agostinho, Maurer, and Veel, “Sensorial
Experience of the Drone”; Graae and Maurer, Drone Imaginaries.
72 Azoulay, “The Natu­ral History of Rape,” 167.
73 Azoulay, “The Natu­ral History of Rape,” 169.
74 Barnell et al., “Agile Condor.”
75 Virilio, War and Cinema.
76 Virilio, War and Cinema.
77 Virilio, War and Cinema.
78 On power and the edge of perception, see Massumi, Ontopower.
79 Barnell et al. “Agile Condor,” 2015; Barnell et al, “High-­Performance Com-
puting,” 2018; Isereau et al “Utilizing High-­Performance Embedded Com-
puting,” 2017.
80 See Halpern, Beautiful Data.
81 Andrejevic, Automated Media, 38.
82 Bousquet, The Eye of War, 12.
83 The Institute of Electrical and Electronics Engineers (ieee) is an electrical
engineering professional ­organization that runs some of the most impor­tant
conferences, journals, and magazines for the development of computational
and related technologies.
84 src Defense, “Agile Condor.”
85 Andrejevic, Automated Media, 18.
86 Andrejevic, Automated Media, 27. Andrejevic describes four distinct modali-
ties of automated media: preemption, operationalism, environmentality, and
framelessness.
87 Bousquet, The Eye of War, 12.
88 Andrejevic, Automated Media, 47.
89 DeLanda, War in the Age of Intelligent Machines, 2.
90 DeLanda, War in the Age of Intelligent Machines, 3.
91 DeLanda, War in the Age of Intelligent Machines, 7.
92 DeLanda, War in the Age of Intelligent Machines, 7.
93 DeLanda, War in the Age of Intelligent Machines, 46.
94 Packer and Reeves, Killer Apps, 14.
95 Whittaker, “The Steep Cost of Capture.”
96 Holmqvist, “Undoing War,” 551.
97 See Sharkey in Bhuta et al., Autonomous Weapons Systems.
98 Packer and Reeves, “The Coming Humanectomy,” 266.
99 Amoore, Cloud Ethics, 150.
100 Amoore, Cloud Ethics, 81.
101 On the relations between autonomous killing and liberal subjectivity, see
Atanasoski and Vora, Surrogate Humanity.
102 Massumi, “Affective Fact,” 53.

Notes to Chapter One 195

103 Andrejevic, Automated Media, 86.
104 Suchman, “Algorithmic Warfare and the Reinvention of Accuracy,” 10.
105 Asaro, “Algorithms of Vio­lence,” 36.
106 Grove, Savage Ecol­ogy, 221. Elsewhere, Grove calls for a new “grammar” of
geopolitics that reckons with the transformative ramifications of technologi-
cal development, see Grove, “From Geopolitics to Geotechnics.”

Chapter Two. Witnessing Algorithms

1 Manjoo, “I Tried Microsoft’s Flight Simulator. The Earth Never Seemed So
Real.”
2 Chayka, “The Weird Failures of Algorithm-­Generated Images.” In this vein,
Clement Valla’s Postcards from Google Earth proj­ect seeks to expose ­these
strange and often beautiful moments of breakdown in the automated assem-
bly of images. Valla, Postcards from Google Earth.
3 See, for example, Amoore and Raley, “Securing with Algorithms”; Bellanova
et al., “Critique of Algorithmic Vio­lence”; Noble, Algorithms of Oppression;
Bucher, If . . . Then; Eubanks, Automating ­Inequality; Sadowski, Too Smart;
Finn, What Algorithms Want; Benjamin, Race A ­ fter Technology; Pasquale,
The Black Box Society; Rouvroy and Berns, “Algorithmic Governmentality”;
McKittrick, Dear Science.
4 Finn, What Algorithms Want, 5.
5 Machine learning algorithms are trained on data, but also optimized using a
range of statistical functions. For example, Google’s AlphaGo was optimized
using vari­ous techniques of Bayesian optimization. Chen et al., “Bayesian
Optimization in AlphaGo.”
6 Mackenzie, Machine Learners.
7 Sadowski, “Potemkin ai.”
8 Taffel, “Data and Oil.”
9 Pasquale, The Black Box Society.
10 On algorithmic vio­lence, see, for example, Bellanova et al., “Critique of
Algorithmic Vio­lence”; Safransky, “Geographies of Algorithmic Vio­lence”;
O’Malley and Smith, “ ‘Smart’ Crime Prevention?”; Asaro, “Algorithms of
Vio­lence.”
11 Finn, What Algorithms Want, 5.
12 Halpern, Beautiful Data, 184.
13 Bucher, If . . . Then, 20.
14 Bucher, If . . . Then, 8.
15 Noble, Algorithms of Oppression; Phan and Wark, “What Personalisation
Can Do for You!”
16 Offert and Phan, “A Sign That Spells.”
17 Seaver, “Algorithms as Culture,” 5.

196 Notes to Chapter One

18 “In spite of the rational straightforwardness granted to them by critics and
advocates, ‘algorithms’ are tricky objects to know,” such that “a determined
focus on revealing the operations of algorithms risks taking for granted that
they operate clearly in the first place.” Seaver, “Algorithms as Culture,” 2, 8.
19 Scannell, “What Can an Algorithm Do?”
20 Stark, “Facial Recognition Is the Plutonium of ai.”
21 MacKenzie and Munster, “Platform Seeing,” 2.
22 MacKenzie and Munster, “Platform Seeing,” 8.
23 MacKenzie and Munster, “Platform Seeing,” 6.
24 MacKenzie and Munster, “Platform Seeing,” 4.
25 MacKenzie and Munster, “Platform Seeing,” 3.
26 Amoore, Cloud Ethics, 16.
27 Amoore, Cloud Ethics, 9.
28 Amoore, Cloud Ethics, 135.
29 Amoore, Cloud Ethics, 66. Emphasis in the original.
30 Guattari’s theorizing of the machine is complex and more nuanced than
is necessary to elaborate ­here. See Guattari, Chaosmosis; Guattari, “On
Machines.”
31 My sincere thanks to one of the anonymous reviewers of the manuscript for
the formulation of machinic affect as promiscuous in its adhesiveness.
32 Massumi, Parables, 35.
33 On potential, probability, possibility, and algorithmic systems, see Amoore,
Politics of Possibility.
34 Amoore, Cloud Ethics, 109–29.
35 Massumi, Parables, 26.
36 Cole, “ai-­Assisted Fake Porn.”
37 Cole, “ai-­Assisted Fake Porn.”
38 Goodfellow et al., “Generative Adversarial Networks.”
39 In 2017, University of Washington researchers trained a machine learning
model to transform video footage of Barack Obama to photorealistically
lip-­sync to an audio input track. It was a key early indicator of the potential
for ai-­generated media to mimic real persons, including world leaders. See
SyncedReview 2018 for an overview. Since then, concerns over generative ai
have exploded. The questions sparked by the Obama lip-­sync video take on
a dif­fer­ent hue in the context of generative models, but their implications are
only just beginning to be understood as this book goes into production.
40 Not least, how to think about the knowledge produced in the scientific ap-
plication of gans. See Offert, “Latent Deep Space.”
41 Maddocks, “ ‘A Deepfake Porn Plot’ ”; Kikerpill, “Choose Your Stars and
Studs”; Nagel, “Verifying Images”; Popova, “Reading out of Context.” Saman-
tha Cole has also made the point repeatedly in Motherboard that while the
­political implications of deepfakes have attracted the mainstream headlines,

Notes to Chapter Two 197

the videos remain almost exclusively pornographic and targeted at ­women in
practice.
42 Gerstner, “Face/Off ”; Kirchengast, “Deepfakes and Image Manipulation”;
Maras and Alexandrou, “Determining Authenticity.”
43 Chesney and Citron, “Deepfakes and the New Disinformation War,” 149. In a
similar vein, deepfakes have been articulated as both new forms of “­political
warfare” and as a critical “multi-­level policy challenge” for states. See also,
Paterson and Hanley, “­Political Warfare in the Digital Age.”
44 McCosker, “Making Sense of Deepfakes.”
45 Uliasz, “On the Truth Claims of Deepfakes,” 25, 27.
46 See, for example, Whyte, “Deepfake News”; Fallis, “Epistemic Threat of Deep-
fakes”; Rini, “Epistemic Backstop.”
47 witness Media Lab, “Synthetic Media and Deepfakes,” 15.
48 witness Media Lab, “Synthetic Media and Deepfakes,” 18.
49 Chouliaraki and al-­Ghazzi, “Flesh Witnessing.”
50 Uliasz, “On the Truth Claims of Deepfakes,” 37.
51 Sear, “Xenowar.”
52 Deleuze, Cinema 1; Tomkins, Shame and Its S­ isters.
53 Anderson, Technologies of Vision, 21.
54 Anderson, Technologies of Vision, 23.
55 Experts remain divided on which side ­will win, most major platforms and nu-
merous start-­ups are developing detection—­although questions remain about
who ­will have access to them and ­whether they ­will be usable by ­people in the
Global South. See https://­lab​.­witness​.­org​/­backgrounder​-­deepfakes​-­in​-­2020​/­.
56 Miglio, “ai in Unreal Engine.”
57 Pugliese, Biopolitics of the More-­Than-­Human, 15.
58 Weizman, Forensic Architecture, 65.
59 Weizman, Forensic Architecture, 58.
60 Weizman, Forensic Architecture, 58, 82.
61 ­There is, it should be noted, an increasingly significant domain within
machine learning research that aims for “explainable” ai; however, this is far
from the norm and contains prob­lems of its own, such as the ­limited frames
within which such systems might be “explained.”
62 Amoore, Cloud Ethics, 119.
63 Massumi, Ontopower, 53.
64 Sontag, Regarding Pain; Kozol, Distant Wars Vis­i­ble; Mirzoeff, The Right to
Look.
65 Google’s former ceo Eric Schmidt is a notable proponent of strengthening
ties between big tech and the US military. He is the chair of the Defense In-
novation Board and reportedly an influential voice in the Biden administra-
tion on the advantages of advance technology, often framing its importance
with reference to China’s capabilities.
66 Department of Defense, “awcft Memorandum.”

198 Notes to Chapter Two

67 Department of Defense, “awcft Memorandum.”
68 Bulletin of the Atomic Scientists, “Proj­ect Maven.”
69 Johnson, “Influx of Drone Footage.”
70 ­Human Rights Watch, “A Wedding That Became a Funeral.”
71 Amoore, Cloud Ethics, 128.
72 Amoore, Cloud Ethics, 127.
73 GovernmentCIOMedia, “ai to Help Pentagon.”
74 See, for example, Mbembe, Necropolitics; Wilcox, “Embodying Algorithmic
War”; Rhee, Robotic Imaginary; Wall and Monahan, “Surveillance and Vio­
lence from Afar”; Kapadia, Insurgent Aesthetics.
75 Benjamin, Race ­after Technology, 36.
76 Phan and Wark, “Racial Formations,” 2.
77 Wilcox, “Embodying Algorithmic War,” 21.
78 Bender et al., “On the Dangers of Stochastic Parrots.” Coleader of the Ethical
ai team Margaret Mitchell was fired a few weeks ­later, having been vocally
and publicly supportive of Gebru and critical of her dismissal.
79 Dencik, et al., “Data Justice.”
80 Sadowski, “When Data Is Capital.”
81 Holmes, “Google’s Off-­Limits Data Farm.”
82 Within industry and academia, ai ethics have become their own “economy of
virtue” in which engagements with ethics and ­performances around the issue
have become power­ful currency. See Phan et al., Economies of Virtue.
83 OpenAI, “Proximal Policy Optimization.”
84 Amoore, Cloud Ethics, 23.
85 Glissant, Poetics.

Chapter Three. Witnessing Ecologies

1 Clarke, “Australia Shuts Down Climate Deal.”
2 Lyons, “Fiji PM.”
3 Westfall, “Australia Ranks Last on Climate Action.”
4 DeLoughrey, Allegories of the Anthropocene, 25.
5 DeLoughrey, Allegories of the Anthropocene, 7.
6 McDonald, Ecological Security.
7 Jetñil-­Kijiner, “Tell Them.”
8 DeLoughrey, Allegories of the Anthropocene, 1.
9 DeLoughrey, Allegories of the Anthropocene, 5.
10 DeLoughrey, Allegories of the Anthropocene, 195.
11 Jetñil-­Kijiner, “Tell Them.”
12 Daston, “Epistemic Images,” 17.
13 Cubitt, Finite Media, 4.
14 Fuller, Media Ecologies, 17.
15 Grove, Savage Ecol­ogy, 4.

Notes to Chapter Three 199

16 Cubitt, Finite Media, 117.
17 Haraway, Staying with the Trou­ble.
18 Cubitt, Finite Media, 154.
19 Yusoff, A Billion Black Anthropocenes.
20 Turner, From Counterculture to Cyberculture.
21 Perhaps most powerfully manifest in the back to the land movement, which
tended to be principally pursued by white middle-­class youth who ­imagined
a pure nature in which Indigenous ­people and knowledge was ­either absent
or reduced to readily appropriated totems.
22 Edwards, “Entangled Histories,” 34.
23 On the technical pro­cesses involved in earth imaging, including the role of
satellites, tracking stations, pixels, and analysts, see Russill, “Earth Imaging,”
245. In his study of the formation of the climate-­monitoring system, Paul N.
Edwards shows how the twinned dynamics of making data global (data from
one place that could be analyzed in relation to data from elsewhere) and
making global data (data about climatic change on a global scale) entailed
significant ­labor, which he calls “data friction.” Encompassing every­thing from
adjusting recordings based on known instrument eccentricities to the sheer
effort of transcribing handwritten rec­ords into digital spreadsheet, the chal-
lenge of data friction erodes the promise of bringing ­immense computational
power to bear on climate and environmental analy­sis. Edwards, A Vast
Machine. See also Yang et al., “Satellite Remote Sensing in Climate Change
Studies”; Kidder and Haar, Satellite Meteorology.
24 Parks, Rethinking Media Coverage.
25 Russill, “Earth Imaging,” 232.
26 Parks, Cultures in Orbit; Kurgan, Close up at a Distance; Russill, “Is the Earth
a Medium?”
27 Russill, “Earth Imaging,” 232. The shift ­toward using a range of datalogi-
cal technologies within ­human rights work, particularly via the emergence
of open-­source investigation, has necessitated new expertise in ­these
media forms within ­human rights fields and produced new intermediary
­organizations, see Ristovska, Seeing H ­ uman Rights.
28 Furuhata, Climatic Media.
29 Weizman, Forensic Architecture, 20.
30 Rothe, “Seeing Like a Satellite,” 340, 337.
31 Gabrys, Program Earth, 11.
32 Guattari, The Three Ecologies, 44.
33 Guattari, The Three Ecologies, 43.
34 Fuller, Media Ecologies, 20.
35 Clark, “Scale,” 148.
36 Fuller, Media Ecologies, 147.
37 Fuller, Media Ecologies, 147.
38 Chakrabarty, “Climate of History,” 206–7.

200 Notes to Chapter Three

39 Clark, “Scale,” 152.
40 Woods, “Scale Critique for the Anthropocene,” 138.
41 Alaimo, Exposed, 168.
42 Raymond Williams proposed “structures of feeling” as a way of understand-
ing how par­tic­u­lar meanings and values are ­organized into the lived experi-
ence of the pre­sent. See Williams, Marxism and Lit­er­a­ture.
43 Zylinska, Nonhuman Photography, 37.
44 Woods, “Scale Critique for the Anthropocene.”
45 Clark, “Scale,” 151.
46 Farrier, Anthropocene Poetics, 15.
47 Schankweiler, Straub, and Wendl, Image Testimonies; Guerin and Hallas, The
Image and the Witness; Campt, Listening to Images.
48 Zylinska, Nonhuman Photography, 81.
49 Peters, “Witnessing,” 722.
50 Tsing, “Nonscalability,” 505.
51 James, The Black Jacobins; Taylor, Empire of Neglect; McKittrick, Dear Science;
Rosenthal, Accounting for Slavery.
52 Tsing, “Nonscalability,” 510.
53 Whyte, “Environmental Injustice,” 125.
54 Nixon, Slow Vio­lence.
55 Kaplan, Climate Trauma; Richardson, “Climate Trauma.”
56 Morton, Hyperobjects, 1.
57 While Morton has elsewhere written about the necessity of addressing ecol­ogy
without the overdetermining of Nature, the irony of their hyberobjects is that
they risk replace replacing one problematic enclosure with another. As a theo-
retical endeavor, the object-­oriented ontology movement, of which Morton is a
prominent flag-­waver, results in something of a ­political dead end, not least due
to its insistence on the primacy of objects and the nonrelationality of existence.
58 On bushfires and their elemental properties, see Neale, Zahara, and Smith,
“An Eternal Flame.”
59 Massumi, Ontopower, 22.
60 Mittmann, “Maralinga.”
61 Cited in Barnaby and Holdstock, British Nuclear Weapons, 99.
62 Ladd, “Lesser Known History”; Tynan, Atomic Thunder.
63 Tynan, “Sixty Years On”; Maralinga Rehabilitation Technical Advisory Com-
mittee, “Rehabilitation of Emu and Maralinga.”
64 Barnaby and Holdstock, British Nuclear Weapons, 102. Authorized by Prime
Minister Robert Menzies without cabinet consultation, the tests ­were over-
seen by the Atomic Weapons Test Safety Committee. The awtsc was headed
by scientists pre­sent at the 1952 ­trials (Proj­ect Hurricane), which was the first
atomic bomb tested by the British in Australia (Arnold and Smith, Britain,
Australia, and the Bomb, 27; Smith, Clouds of Deceit, 58). No regard was given
to the effects on Indigenous inhabitants.

Notes to Chapter Three 201

65 Tynan, “Sixty Years On.”
66 Cook et al., “Pu-­Bearing Particles at Maralinga.”
67 ican: International Campaign to Abolish Nuclear Weapons, “Black Mist.”
68 Quoted in Mittmann, “Maralinga,” 10.
69 McLelland, Report of the Royal Commission into British Nuclear Tests in
­Australia, 574.
70 Cited in Alexis-­Martin, “Nuclear Warfare and Weather (Im)Mobilities,” 258.
71 Tynan, “What Is Relationality?,” 1.
72 According to Moreton-­Robinson, “White possessive logics are operational-
ized within discourses to circulate sets of meanings about owner­ship of the
nation, as part of commonsense knowledge, decision making, and socially
produced conventions.” Moreton-­Robinson, The White Possessive, xii.
73 The British government has admitted to using Australian ser­vicemen in
radioactivity experiments, ordering them to walk into blast zones a­ fter tests.
Car­ter, British Nuclear Tests in Australia.
74 Mittmann, “Maralinga,” 9.
75 On terra nullius and the cultural and ­legal logics of colonial possession in
Australia, see Moreton-­Robinson, The White Possessive.
76 Made ­popular by 1990s antinuclear activism, the term nuclear colonialism
“designates con­temporary dynamics of colonial exploitation, in which impe-
rial countries militarily occupy and irreversibly pollute the lands and natu­ral
resources of communities, far away from their own economic and ­political
centers, in order to further the development of their nuclear technologies.”
Maurer, “Snaring the Nuclear Sun,” 372. See also Endres, “Rhe­toric of Nuclear
Colonialism”; Endres, “Most Nuclear-­Bombed Place”; Edwards, “Nuclear
Colonialism.”
77 Maurer, “Snaring the Nuclear Sun,” 373.
78 Hau’ofa, “Our Sea of Islands.”
79 This co-­option of Aboriginal language to name military sites was hardly
new. The nearby town of Woomera was named for an Aboriginal spear-­
throwing implement ­after it was built to facilitate long-­range weapons testing
in 1947.
80 Hydrogen bombs, or thermonuclear weapons, use a combination of nuclear
fission and nuclear fusion, a separate ­process in which the energy is released
through smashing unstable atoms together.
81 Of the first nuclear detonation at White Sands Missile Test Range at Jornada
del Muerto desert of New Mexico, Alexis-­Martin writes: “This first nuclear
weapon detonation marked the dawn of a freshly atomic Anthropocene, cre-
ating localized air temperatures so hot that silica sand seared into glassy trin-
itite, fossilizing the heat of the blast into puzzling.” Alexis-­Martin, “Nuclear
Warfare and Weather (Im)Mobilities,” 251.
82 Ray, “Myth of Empty Country,” 5.
83 Scarce, Delany, and Australian Centre for Con­temporary Art, Missile Park, 92.

202 Notes to Chapter Three

84 The ­measure by which an isotopes radioactivity decreases (and the time it
takes for it to fully transform into its product atom) is called a half-­life.
More specifically, it’s the time that it takes for one half of the original radio-
active isotope to decay. This means that as time passes, each successive half
of the remaining radioactive isotope transforms and its level of radioactivity
also decreases. If the half-­life of an isotope (e.g., Thorium-90) is twenty-­four
days, it takes twenty-­four days for half the isotope to disintegrate; followed
by another twenty-­four days for half of the remaining half, and so on. De-
pending on the isotope, the half-­life could be years or minutes.
85 Vivian, Commonplace Witnessing.
86 Scarce, Delany, and Australian Centre for Con­temporary Art, Missile Park,
126.
87 Ray, “Myth of Empty Country,” 5.
88 Alexis-­Martin, “Nuclear Warfare and Weather (Im)Mobilities,” 251.
89 Harkin, “Anneal This Breath, 86.”
90 Deleuze, Logic of Sense, 73.
91 Deleuze, Pure Immanence. Among other texts, this book contains the essay
“Immanence: A Life. . . .”
92 For more on plural worlds and William James, see Savransky, Around the
Day in Eighty Worlds.
93 De la Cadena and Blaser, A World of Many Worlds.
94 De la Cadena, Earth Beings, 5.
95 Deleuze, Pure Immanence, 31.
96 Deleuze and Guattari, A Thousand Plateaus, 266.
97 Tynan, “What Is Relationality?,” 5.
98 Deleuze, Pure Immanence, 31.
99 Deleuze, Pure Immanence, 31–32.
100 Caruth, Unclaimed Experience, 4.
101 Massumi, Parables, 35.

Chapter Four. Witnessing Absence

1 Kember and Zylinska, Life ­after New Media, 71–72.
2 Farwell, “Media Strategy of isis”; Ingram, “Islamic State’s Information
Warfare.”
3 Hoskins and Illingworth, “Inaccessible War,” 1.
4 Seigworth and Tiessen, “Mobile Affects.”
5 Ahmed, Cultural Politics of Emotion, 71–80.
6 Bourke, “Public Beheading Fears.”
7 Sparrow, “Plastic Sword.”
8 Grusin, Premediation, 72.
9 Clery, “Six Handshakes, Then Silence.”
10 Wise, “How Crazy Am I?”

Notes to Chapter Four 203

11 Christopoulos and Ustinova, “Urgent Hypothesis on Plane mh370
Disappearance.”
12 Day and Lury, “New Technologies of the Observer”; Bremner, “Technologies
of Uncertainty in the Search for mh370”; Bremner, “Fluid Ontologies.”
13 Gabrys, Program Earth.
14 Chris Ashton et al., “mh370.”
15 See Day and Lury, “New Technologies of the Observer”; Taylor, “Fiction
Machines.”
16 Edwards, Closed World.
17 Scannell, ­Television and the Meaning of Live, 66.
18 Jue, Wild Blue Media.
19 Pinchevski, Transmitted Wounds, 14.
20 Pinchevski, Transmitted Wounds, 15.
21 Cubitt, Practice of Light, 2.
22 Berlant, “Feeling Historical,” 6.
23 Gibbs, “Panic!”; Gibbs, “Contagious Feelings”; Richardson, “­There’s Some-
thing G
­ oing On.”
24 Seltzer, “Wound Culture,” 3.
25 Atkinson and Richardson, Traumatic Affect, 12.
26 Pinchevski, Transmitted Wounds, 22.
27 Murphie, “On Being Affected,” 24.
28 Massumi, Parables, 35.
29 Richardson and Schankweiler, “Affective Witnessing as Theory and Practice.”
30 Lapoujade, William James.
31 Kaleem, “Death on Facebook.”
32 Öhman and Watson, “Are the Dead Taking over Facebook?”
33 Leaver, “Social Media Contradiction.”
34 For many, Facebook is a site of testifying to lives lived and to the grief en-
dured by survivors. See Keskinen, Kaunonen, and Aho, “Grief on Facebook.”
35 Frosh, Poetics of Digital Media, 111.
36 Hogan, “Facebook Data Storage Centers,” 11.
37 Brubaker, Hayes, and Dourish, “Beyond the Grave.”
38 Gibbs, “Writing and Danger.”
39 The very distinction between online and offline is, of course, problematic.
“The notion of the offline as real and au­then­tic is a recent invention, corre-
sponding with the rise of the online.” Jurgenson, The Social Photo, 68.
40 Stiegler, “Anamnesis and Hypomnesis.”
41 Chow, Entanglements, 19.
42 Thrift, Non-­Representational Theory, 2.
43 Angerer, Ecol­ogy of Affect, 24.
44 Manning and Massumi, Thought in the Act, 24.
45 Doane, Cinematic Time, 225.
46 Manning, Minor Gesture, 54.

204 Notes to Chapter Four

47 Manning, Minor Gesture, 134.
48 Massumi, Ontopower, 95.
49 Chow, Entanglements, 12.
50 Berlant and Greenwald, “Affect in End Times,” 82.
51 Berlant, “The Commons,” 393.
52 Berlant, “The Commons,” 394.
53 Massumi, Parables, 30.
54 Massumi, Parables, 98.
55 While the Puutu Kunti Kurrama and Pinikura ­peoples use the acronym pkkp,
they are distinct socioterritorial groups. As their submission to the Joint
Standing Committee on Northern Australia inquiry into Juukan Gorge states:
“The Puutu Kunti Kurrama ­people and the Pinikura ­people are separate
­peoples with discrete rights and interests in country, though we have some
shared laws and customs.” Cited in Joint Standing Committee on Northern
Australia, “A Way Forward,” 2.
56 Joint Standing Committee on Northern Australia, “Never Again,” 2.
57 Moreton-­Robinson, The White Possessive, ix.
58 Wolfe, “Settler Colonialism and the Elimination of the Native.”
59 Berlant, Cruel Optimism, 1.
60 This positive potential should not be overstated. As Jasbir Puar argues in
response to optimistic sloganeering in response to depression and suicide
among queer youth, the “tendentious mythologizing that ‘it gets better’ ” ­after
traumatic events obscures and even protects the systemic origins of violent
oppression. Puar, “Cost of Getting Better,” 149.
61 Manning, Minor Gesture, 151.
62 Manning, Minor Gesture, 1.
63 Stewart, “Worlding Refrains,” 340.

Coda. ­Toward a Politics of Nonhuman Witnessing

1 Weston, Animate Planet, 33.
2 See Davis, “How a Pandemic Happens.”
3 Cubitt, Finite Media, 162.
4 Rancière, Disagreement.
5 De la Cadena and Blaser, A World of Many Worlds, 13. ­There is also a rich
body of lit­er­a­ture on pluriversality from both ­process philosophy and non-­
Western perspectives that I cannot do justice to ­here but that interested
readers might wish to pursue. For example, Escobar, Pluriversal Politics;
Escobar, Designs for the Pluriverse; Reiter, Constructing the Pluriverse; Savran-
sky, Around the Day in Eighty Worlds; Yunkaporta, Sand Talk.
6 De la Cadena and Blaser, A World of Many Worlds, 15.
7 De la Cadena, “Indigenous Cosmologies,” 359.
8 De la Cadena, “Indigenous Cosmologies,” 345.

Notes to Coda 205

9 De la Cadena, “Indigenous Cosmologies,” 361.
10 Cubitt, Finite Media, 178.
11 Cubitt, Finite Media, 177.
12 Glissant, Poetics, 62.
13 Glissant, Poetics, 62.
14 Glissant, Poetics, 191.
15 Glissant, Poetics, 190.
16 Glissant, Poetics, 160.
17 I am deeply indebted to Andrew Brooks for his wisdom and generous guid-
ance through Glissant.
18 Cubitt, Finite Media, 6.
19 De la Cadena, Earth Beings.
20 De la Cadena, “Indigenous Cosmologies,” 360.
21 “Our proposal is an invitation to think that instead of the sameness that rec-
ognition supposes, politics might not start from, nor resolve in ontologically
homogeneous grounds. Rather, the grounds of adversarial dispute or of allied
agreement would be what we call uncommons.” De la Cadena and Blaser, A
World of Many Worlds, 27.
22 Fuller and Weizman, Investigative Aesthetics, 165.

206 Notes to Coda

bibliography

Abraham, Nicolas, and Maria Torok. The Shell and the Kernel. Translated by
Nicholas T. Rand. Vol. 1. Chicago: University of Chicago Press, 1994.
Adler, Jonathan E. “Testimony, Trust, Knowing.” Journal of Philosophy 91, no. 5
(1994): 264–75.
Agamben, Giorgio. Remnants of Auschwitz: The Witness and the Archive. Trans-
lated by Daniel Heller-­Roazen. New York: Zone, 2002.
Agostinho, Daniela, Kathrin Maurer, and Kristin Veel. “Introduction to the
Sensorial Experience of the Drone.” Senses and Society 15, no. 3 (Septem-
ber 2020): 251–58.
Ahmed, Sara. The Cultural Politics of Emotion. New York: Routledge, 2004.
Alaimo, Stacy. Bodily Natures: Science, Environment, and the Material Self.
Bloomington: Indiana University Press, 2010.
Alexis-­Martin, Becky. “Nuclear Warfare and Weather (Im)Mobilities: From
Mushroom Clouds to Fallout.” In Weather: Spaces, Mobilities and Affects.
New York: Routledge, 2020.
Amazon. “All In: Staying the Course on Our Commitment to Sustainability,”
2020. https://­sustainability​.­aboutamazon​.­com​/­environment​/­sustainable​
-­operations​/­carbon​-­footprint.
Amoore, Louise. “Algorithmic War: Everyday Geographies of the War on Terror.”
Antipode 41, no. 1 (January 2009): 49–69.
Amoore, Louise. Cloud Ethics: Algorithms and the Attributes of Ourselves and
­Others. Durham, NC: Duke University Press, 2020.
Amoore, Louise. The Politics of Possibility: Risk and Security Beyond Probability.
Durham, NC: Duke University Press, 2013.
Amoore, Louise, and Rita Raley. “Securing with Algorithms: Knowledge, Deci-
sion, Sovereignty.” Security Dialogue 48, no. 1 (February 2017): 3–10.
Andén-­Papadopoulos, Kari. “Citizen Camera-­Witnessing: Embodied P ­ olitical
Dissent in the Age of ‘Mediated Mass Self-­Communication.’ ” New Media
and Society, 2013, 753–69.
Andén-­Papadopoulos, Kari. “Media Witnessing and the ‘Crowd-­Sourced Video
Revolution.’ ” Visual Communication 12, no. 3 (August 2013): 341–57.
Anderson, Steve F. Technologies of Vision: The War between Data and Images.
Cambridge, MA: mit Press, 2017.
Andrejevic, Mark. Automated Media. 1st ed. New York: Routledge, 2019.
Angerer, Marie-­Luise. Ecol­ogy of Affect: Intensive Milieus and Contingent Encoun-
ters. Translated by Gerrit Jackson. Lüneborg: Meson, 2017.
Arendt, Hannah. On Vio­lence. New York: Harcourt, Brace, Jovanovich, 1970.
Arnold, Lorna, and Mark Smith. Britain, Australia and the Bomb: The Nuclear
Tests and Their Aftermath. 2nd ed. London: Palgrave Macmillan, 2006.
Asaro, Peter. “Algorithms of Vio­lence: Critical Social Perspectives on Autono-
mous Weapons.” Social Research: An International Quaterly 86, no. 2
(2019): 20.
Ashton, Chris, Alan Shuster Bruce, Gary Colledge, and Mark Dickinson. “The
Search for mh370.” Journal of Navigation 68 (2015): 1–22.
Atanasoski, Neda, and Kalindi Vora. Surrogate Humanity: Race, Robots, and the
Politics of Technological ­Futures. Perverse Modernities. Durham, NC: Duke
University Press, 2019.
Atkinson, Meera, and Michael Richardson, eds. Traumatic Affect. Newcastle upon
Tyne: Cambridge Scholars, 2013.
Azoulay, Ariella. “The Natu­ral History of Rape.” Journal of Visual Culture 17, no. 2
(August 2018): 166–76.
Ballard, Su. “ ‘And They Are like Wild Beasts’: Violent ­Things in the Anthropo-
cene.” Fibreculture 226 (2019). https://­thirty​.­fibreculturejournal​.­org​/­fcj​-­226​
-­and​-­they​-­are​-­like​-­wild​-­beasts​-­violent​-­things​-­in​-­the​-­anthropocene​/­.
Ballengee, Jennifer R. The Wound and the Witness: The Rhe­toric of Torture.
New York: State University of New York Press, 2009.
Barad, Karen. “Posthumanist Performativity: ­Toward an Understanding of How
­Matter Comes to ­Matter.” Signs: Journal of ­Women in Culture and Society 28,
no. 3 (March 2003): 801–31.
Barnaby, Frank, and Douglas Holdstock. The British Nuclear Weapons Pro-
gramme, 1952–2002. London: Routledge, 2004.
Barnell, Mark, Courtney Raymond, Christopher Capraro, et al. “Agile Condor: A
Scalable High ­Performance Embedded Computing Architecture.” Waltham,
MA: 2015 ieee High ­Performance Extreme Computing Conference
­(September 2015): pp. 1–5. https://­doi​.­org​/­10​.­1109​/­HPEC​.­2015​.­7322447.

208 Bibliography
Barnell, Mark, Courtney Raymond, Chris Capraro, Darrek Isereau, Chris Cicotta,
and Nathan Stokes. “High-­Performance Computing (hpc) and Machine
Learning Demonstrated in Flight Using Agile Condor®.” Waltham, MA: 2018
ieee High ­Performance Extreme Computing Conference (September 2018):
1–4. https://­doi​.­org​/­10​.­1109​/­HPEC​.­2018​.­8547797.
Bashir, Shazad, and Robert D. Crews. ­Under the Drones: Modern Lives in the
Afghanistan-­Pakistan Borderlands. Cambridge, MA: Harvard University
Press, 2012.
Bellanova, Rocco, Kristina Irion, Katja Lindskov Jacobsen, Francesco Ragazzi,
Rune Saugmann, and Lucy Suchman. “­Toward a Critique of Algorithmic
Vio­lence.” International ­Political Sociology 15, no. 1 (2021): 121–50.
Bender, Emily M., Timnit Gebru, Angelina McMillan-­Major, and Shmargaret
Shmitchell. “On the Dangers of Stochastic Parrots: Can Language Models
Be Too Big?” In Proceedings of the 2021 acm Conference on Fairness, Ac-
countability, and Transparency, 610–23. FAccT ’21. New York: Association
for Computing Machinery, 2021.
Benjamin, Ruha. Race ­after Technology: Abolitionist Tools for the New Jim Code.
Medford, MA: Polity, 2019.
Bennett, Jane. Vibrant M ­ atter: A P ­ olitical Ecol­ogy of Th
­ ings. Durham, NC: Duke
University Press, 2010.
Berlant, Lauren. “The Commons: Infrastructures for Troubling Times.” Environ-
ment and Planning D: Society and Space 34, no. 3 (2016): 393–419.
Berlant, Lauren. Cruel Optimism. Durham, NC: Duke University Press, 2011.
Berlant, Lauren. “Thinking about Feeling Historical.” Emotion, Space and Society
1, no. 1 (2008): 4–9.
Berlant, Lauren, and Jordan Greenwald. “Affect in the End Times: A Conversa-
tion with Lauren Berlant.” Qui Parle 20 (2012): 71–89.
Bhuta, Nehal, Susanne Beck, Robin Geiβ, Hin-­Yan Liu, and Claus Kreβ, eds.
Autonomous Weapons Systems: Law, Ethics, Policy. Cambridge: Cambridge
University Press, 2016.
Boltanski, Luc. Distant Suffering: Morality, Media and Politics. Cambridge:
­Cambridge University Press, 1999.
Bourke, Latika. “Public Beheading Fears: Tony Abbott Confirms Police Believed
Terrorists Planned ‘Demonstration Killings.’ ” Sydney Morning Herald,
September 18, 2014. https://­www​.­smh​.­com​.­au​/­politics​/­federal​/­public​
-­beheading​-­fears​-­tony​-­abbott​-­confirms​-­police​-­believed​-­terrorists​-­planned​
-­demonstration​-­killings​-­20140918​-­10ilyq​.­html.
Bousquet, Antoine. The Eye of War: Military Perception from the Telescope to the
Drone. Minneapolis: University of Minnesota Press, 2018.
Bousquet, Antoine. The Scientific Way of Warfare: Order and Chaos on the Battle-
fields of Modernity. New York: Columbia University Press, 2009.
Bousquet, Antoine, Jairus Grove, and Nisha Shah. “Becoming War: ­Towards a
Martial Empiricism.” Security Dialogue 51, nos. 2–3 (2020): 99–128.

Bibliography 209
Bräunert, Svea, and Meredith Malone. To See without Being Seen: Con­temporary
Art and Drone Warfare. Saint Louis, MO: Mildred Lane Kemper Art
­Museum, 2016.
Bremner, Lindsay. “Fluid Ontologies in the Search for mh370.” Journal of the
Indian Ocean Region 11 (2015): 8–29.
Bremner, Lindsay. “Technologies of Uncertainty in the Search for mh370.” In Art
in the Anthropocene: Encounters among Aesthetics, Politics, Environments
and Epistemologies, edited by Heather Davis and Etienne Turpin. London:
Open Humanities, 2015.
Browne, Simone. Dark M ­ atters: On the Surveillance of Blackness. Durham, NC:
Duke University Press, 2015.
Brubaker, Jed R., Gillian R. Hayes, and Paul Dourish. “Beyond the Grave:
Facebook as a Site for the Expansion of Death and Mourning.” Information
Society 29 (2013): 152–63.
Bucher, Taina. If . . . Then: Algorithmic Power and Politics. Oxford Studies in Digi-
tal Politics. New York: Oxford University Press, 2018.
Bulletin of the Atomic Scientists. “Proj­ect Maven Brings ai to the Fight against
isis.” December 21, 2017. https://­thebulletin​.­org​/­2017​/­12​/­project​-­maven​
-­brings​-­ai​-­to​-­the​-­fight​-­against​-­isis​/­.
Butler, Judith. Precarious Life: The Powers of Mourning and Vio­lence. New York:
Verso, 2004.
Campt, Tina. Listening to Images. Durham, NC: Duke University Press, 2017.
Car­ter, Michael. Australian Participants in British Nuclear Tests in Australia. Can-
berra: Department of Veterans’ Affairs, 2006. https://­www​.­dva​.­gov​.­au​/­sites​
/­default​/­files​/­dosimetry​_­complete​_­study​_­1​.­pdf.
Caruth, Cathy. Unclaimed Experience: Trauma, Narrative, and History. Baltimore:
Johns Hopkins University Press, 1996.
Chakrabarty, Dipesh. “The Climate of History: Four ­Theses.” Critical Inquiry 35,
no. 2 (Winter 2009): 197–222.
Chamayou, Grégoire. Drone Theory. Translated by Janet Lloyd. London: Penguin,
2015.
Chandler, Katherine. Unmanning: How H ­ umans, Machines and Media Perform
Drone Warfare. War Culture. New Brunswick, NJ: Rutgers University Press,
2020.
Chayka, Kyle. “The Weird Failures of Algorithm-­Generated Images.” Slate,
August 28, 2020. https://­slate​.­com​/­technology​/­2020​/­08​/­uncanniness​-­of​
-­algorithmic​-­style​.­html.
Chen, Yutian, Aja Huang, Ziyu Wang, Ioannis Antonoglou, Julian Schrittwieser,
David Silver, and Nando de Freitas. “Bayesian Optimization in AlphaGo.”
ArXiv, arXiv:1812.06855 (December 2018). https://­arxiv​.­org​/­abs​/­1812​.­06855v1.
Chesney, Robert, and Danielle Citron. “Deepfakes and the New Disinformation
War: The Coming Age of Post-­truth Geopolitics.” Foreign Affairs 98, no. 1
(2019): 147–55.

210 Bibliography
Chouliaraki, Lilie. “Digital Witnessing in Conflict Zones: The Politics of Reme-
diation.” Information, Communication and Society 18, no. 11 (2015): 1362–77.
Chouliaraki, Lilie. The Spectatorship of Suffering. London: sage, 2006.
Chouliaraki, Lilie, and Omar al-­Ghazzi. “Beyond News Verification: Flesh Wit-
nessing and the Significance of Embodiment in Conflict News.” Journalism
23, no. 3 (2022): 649–67.
Chow, Rey. Entanglements, or Transmedial Thinking about Capture. Durham, NC:
Duke University Press, 2012.
Chow, Rey. The Age of the World Target: Self-­Referentiality in War, Theory, and
Comparative Work. Durham, NC: Duke University Press, 2006.
Christopoulos, Demetris T., and Galina K. Ustinova. “Urgent Hypothesis on
Plane mh370 Disappearance.” ResearchGate, 2014.
Clark, Timothy. “Scale.” In Theory in the Era of Climate Change. Vol. 1, edited by
Tom Cohen, 148–66. Ann Arbor, MI: Open Humanities, 2012.
Clarke, Melissa. “Australia Shuts down Climate Deal ­after Discussions Reduce
Tongan pm to Tears.” abc News, August 15, 2019. https://­www​.­abc​.­net​
.­au​/­news​/­2019​-­08​-­15​/­no​-­endorsements​-­come​-­out​-­of​-­tuvalu​-­declaration​
/­11419342.
Clery, Daniel. “Six Handshakes, Then Silence.” Science 344 (May 2014): 964–65.
https://­www​.­science​.­org​/­doi​/­10​.­1126​/­science​.­344​.­6187​.­964.
Clough, Patricia Ticiento. The Affective Turn: Theorizing the Social. Durham, NC:
Duke University Press, 2007.
Coady, C. A. J. Testimony: A Philosophical Study. Oxford: Oxford University Press,
1994.
Cockburn, Andrew. Kill Chain: Drones and the Rise of High-­Tech Assassins.
New York: Verso, 2015.
Cole, Samantha. “ai-­Assisted Fake Porn Is H ­ ere and ­We’re All Fucked.” Vice,
December 17, 2017. https://­www​.­vice​.­com​/­en​_­us​/­article​/­gydydm​/­gal​-­gadot​
-­fake​-­ai​-­porn.
Cook, Megan, Barbara Etschmann, Rahul Ram, Konstantin Ignatyev, Gediminas
Gervinskas, Steven D. Conradson, Susan Cumberland, Vanessa N. L. Wong,
and Joёl Brugger. “The Nature of Pu-­Bearing Particles from the Maralinga
Nuclear Testing Site, Australia.” Scientific Reports 11, no. 1 (2021): 10698.
Cowen, Deborah. The Deadly Life of Logistics. Minneapolis: University of Min-
nesota Press, 2014.
Cubitt, Sean. Finite Media: Environmental Implications of Digital Technologies.
Durham, NC: Duke University Press, 2017.
Cubitt, Sean. The Practice of Light: A Genealogy of Visual Technologies from Prints
to Pixels. Leonardo Book Series. Cambridge, MA: mit Press, 2014.
Danchev, Alex. “Bug Splat: The Art of the Drone.” International Affairs 92, no. 3
(2016): 703–13.
Das, Veena, Arthur Kleinman, Mamphela Ramphele, and Pamela Reynolds. Vio­
lence and Subjectivity. Berkeley: University of California Press, 2000.

Bibliography 211
da Silva, Denise Ferreira. “No-­Bodies.” Griffith Law Review 18, no. 2 (2009): 212–36.
Daston, Lorraine. “Epistemic Images.” In Vision and Its Instruments: Art, Science,
and Technology in Early Modern ­Europe, edited by A. Payne, 13–35. Univer-
sity Park: Pennsylvania State University Press, 2015.
Daston, Lorraine, and Peter Galison. Objectivity. New York: Zone, 2007.
Davis, Heather, and Zoe Todd. “On the Importance of a Date, or, Decolonizing
the Anthropocene.” acme: An International Journal for Critical Geographies
16, no. 4 (2017): 761–80.
Davis, Mike. “How a Pandemic Happens: We Knew This Was Coming.” Literary
Hub, May 18, 2020. https://­lithub​.­com​/­how​-­a​-­pandemic​-­happens​-­we​-­knew​
-­this​-­was​-­coming​/­.
Dawes, James. That the World May Know: Bearing Witness to Atrocity. Cambridge,
MA: Harvard University Press, 2007.
Day, Sophie, and Celia Lury. “New Technologies of the Observer: #BringBack,
Visualization and Disappearance.” Theory, Culture and Society 34, nos. 7–8
(2017): 51–74.
de la Cadena, Marisol. Earth Beings: Ecologies of Practice across Andean Worlds.
Durham, NC: Duke University Press, 2015.
de la Cadena, Marisol. “Indigenous Cosmopolitics in the Andes: Conceptual Re-
flections beyond ‘Politics.’ ” Cultural Anthropology 25, no. 2 (2010): 334–70.
de la Cadena, Marisol. “Uncommoning Nature: Stories from the Anthropo-­Not-­
Seen.” In Anthropos and the Material, edited by Penny Harvey, Christian Krohn-­
Hansen, and Knut G. Nustad, 35–58. Durham, NC: Duke University Press, 2019.
de la Cadena, Marisol, and Mario Blaser, eds. A World of Many Worlds. Durham,
NC: Duke University Press, 2018.
DeLanda, Manuel. War in the Age of Intelligent Machines. New York: Zone, 1991.
Deleuze, Gilles. Cinema 1: The Movement-­Image. Translated by Hugh Tomlinson
and Barbara Habberjam. London: Continuum, 2005.
Deleuze, Gilles. “Ethology: Spinoza and Us.” In Incorporations, edited by Jonathan
Crary and Sanford Kwinter, translated by Robert Hurley, 625–33. New York:
Zone, 1992.
Deleuze, Gilles. The Logic of Sense. New York: Columbia University Press, 1990.
Deleuze, Gilles. Pure Immanence: Essays on a Life. Translated by Anne Boyman.
Cambridge, MA: Zone, 2001.
Deleuze, Gilles, and Félix Guattari. A Thousand Plateaus: Capitalism and
­Schizophrenia. Translated by Brian Massumi. Minneapolis: University of
Minnesota Press, 1987.
DeLoughrey, Elizabeth M. Allegories of the Anthropocene. Durham, NC: Duke
University Press, 2019.
Demos, T. J. Against the Anthropocene: Visual Culture and Environment ­Today.
Berlin: Sternberg, 2017.
Dencik, Lina, Arne Hintz, and Jonathan Cable. “­Towards Data Justice? The
Ambiguity of Anti-­surveillance ­Resistance in ­Political Activism.” Big

212 Bibliography
Data and Society 3, no. 2 (July–­December 2016). https://­doi​.­org​/­10​.­1177​
/­2053951716679678.
Department of Defense. “Memorandum for the Establishment of an Algorithmic
Warfare Cross-­functional Team (Proj­ect Maven),” April 26, 2017. https://­
www​.­govexec​.­com​/­media​/­gbc​/­docs​/­pdfs​_­edit​/­establishment​_­of​_­the​_­awcft​
_­project​_­maven​.­pdf.
Doane, Mary Ann. The Emergence of Cinematic Time: Modernity, Contingency, the
Archive. Cambridge, MA: Harvard University Press, 2002.
Dorrian, Mark. “Drone Semiosis.” Cabinet, no. 54 (2014): 48–55.
DuBois, Page. Torture and Truth. New York: Routledge, 1991.
Easterling, Keller. Extrastatecraft: The Power of Infrastructure Space. New York:
Verso, 2014.
Edney-­Browne, Alex. “The Psychosocial Effects of Drone Vio­lence: Social Isola-
tion, Self-­Objectification, and Depoliticization.” ­Political Psy­chol­ogy 40,
no. 6 (2019): 1341–56.
Edwards, Nelta. “Nuclear Colonialism and the Social Construction of Landscape
in Alaska.” Environmental Justice 4, no. 2 (2011): 109–14.
Edwards, Paul N. The Closed World: Computers and the Politics of Discourse in
Cold War Amer­i­ca. Cambridge, MA: mit Press, 1996.
Edwards, Paul N. “Entangled Histories: Climate Science and Nuclear Weapons
Research.” Bulletin of the Atomic Scientists 68, no. 4 (2012): 28–40.
Edwards, Paul N. A Vast Machine: Computer Models, Climate Data, and the Poli-
tics of Global Warming. Cambridge, MA: mit Press, 2010.
Endres, Danielle. “The Most Nuclear-­Bombed Place: Ecological Implications of
the US Nuclear Testing Program.” In Tracing Rhe­toric and Material Life, ed-
ited by Bridie McGreavy, Justine Wells, George F. McHendry, and Samantha
Senda-­Cook, 253–87. London: Palgrave Macmillan, 2018.
Endres, Danielle. “The Rhe­toric of Nuclear Colonialism: Rhetorical Exclusion of
American Indian Arguments in the Yucca Mountain Nuclear Waste Siting
Decision.” Communication and Critical/Cultural Studies 6, no. 1 (2009):
39–60.
Escobar, Arturo. Designs for the Pluriverse: Radical Interdependence, Autonomy,
and the Making of Worlds. New Ecologies for the Twenty-­First ­Century.
Durham, NC: Duke University Press, 2018.
Escobar, Arturo. Pluriversal Politics: The Real and the Pos­si­ble. Latin Amer­i­ca in
Translation. Durham, NC: Duke University Press, 2020.
Estes, Nick. Our History Is the F­ uture: Standing Rock versus the Dakota Access
Pipeline, and the Long Tradition of Indigenous R ­ esistance. New York: Verso,
2019.
Ettinger, Bracha. The Matrixial Borderspace. Minneapolis: University of Minne-
sota Press, 2006.
Eubanks, ­Virginia. Automating I­ nequality: How High-­Tech Tools Profile, Police,
and Punish the Poor. New York: St. Martin’s, 2018.

Bibliography 213
Fallis, Don. “The Epistemic Threat of Deepfakes.” Philosophy and Technology 34,
no. 4 (December 2021): 623–43. https://­doi​.­org​/­10​.­1007​/­s13347​-­020​-­00419​-­2.
Farmer, Paul. “An Anthropology of Structural Vio­lence.” Current Anthropology 45,
no. 3 (2004): 305–25.
Farrier, David. Anthropocene Poetics: Deep Time, Sacrifice Zones, and Extinction.
Minneapolis: University of Minnesota Press, 2019.
Farwell, James P. “The Media Strategy of isis.” Survival 56 (2014): 49–55.
Fassin, Didier. “The Humanitarian Politics of Testimony: Subjectification through
Trauma in the Israeli Palestinian Conflict.” Cultural Anthropology 23, no. 3
(2008): 531–58.
Felman, Shoshana, and Dori Laub. Testimony: Crises of Witnessing in Lit­er­a­ture,
Psychoanalysis, and History. New York: Routledge, 1992.
Finn, Ed. What Algorithms Want: Imagination in the Age of Computing. Cambridge,
MA: mit Press, 2017.
Fish, Adam. “Blue Governmentality: Elemental Activism with Conservation
Technologies on Plundered Seas.” ­Political Geography 93 (2022): 102528.
Frosh, Paul. The Poetics of Digital Media. Cambridge: Polity, 2019.
Frosh, Paul, and Amit Pinchevski. “Introduction: Why Media Witnessing? Why
Now?” In Media Witnessing: Testimony in the Age of Mass Communication,
edited by Paul Frosh and Amit Pinchevski, 1–19. Basingstoke, UK: Palgrave
Macmillan, 2009.
Frosh, Paul, and Amit Pinchevski. Media Witnessing: Testimony in the Age of Mass
Communication. Basingstoke, UK: Palgrave Macmillan, 2009.
Fuller, Matthew. Media Ecologies: Materialist Energies in Art and Technoculture.
Leonardo. Cambridge, MA: mit Press, 2005.
Fuller, Matthew, and Eyal Weizman. Investigative Aesthetics: Conflicts and Com-
mons in the Politics of Truth. New York: Verso, 2021.
Furuhata, Yuriko. Climatic Media: Transpacific Experiments in Atmospheric Con-
trol. Ele­ments. Durham, NC: Duke University Press, 2022.
Furuhata, Yuriko. “Multimedia Environments and Security Operations: Expo ’70
as a Laboratory of Governance.” Grey Room 54 (2014): 56–79.
Gabrys, Jennifer. Program Earth: Environmental Sensing Technology and the Making
of a Computational Planet. Minneapolis: University of Minnesota Press, 2016.
Geoghegan, Bernard Dionysius. “An Ecol­ogy of Operations: Vigilance, Radar, and
the Birth of the Computer Screen.” Repre­sen­ta­tions 147, no. 1 (2019): 59–95.
Gerstner, Erik. “Face/Off: DeepFake Face Swaps and Privacy Laws.” Defense
Counsel Journal 87, no. 1 (2020): 1–14.
Gibbs, Anna. “Contagious Feelings: Pauline Hanson and the Epidemiol-
ogy of Affect.” Australian Humanities Review 24 (2001). https://­
australianhumanitiesreview​.­org​/­2001​/­12​/­01​/­contagious​-­feelings​-­pauline​
-­hanson​-­and​-­the​-­epidemiology​-­of​-­affect​/­.
Gibbs, Anna. “Panic! Affect Contagion, Mimesis and Suggestion in the Social
Field.” Cultural Studies Review 14 (2008): 130–45.

214 Bibliography
Gibbs, Anna. “Writing and Danger: The Intercorporeality of Affect.” In Creative
Writing: Theory beyond Practice, edited by Tess Brody and Nigel Krauth,
157–67. Tenerife, qld: Post Pressed, 2006.
Gil-­Fournier, Abelardo, and Jussi Parikka. “Ground Truth to Fake Geographies:
Machine Vision and Learning in Visual Practices.” ai and Society 36 (2020):
1253–62.
Gilmore, Ruth Wilson. Golden Gulag: Prisons, Surplus, Crisis, and Opposition in
Globalizing California. Berkeley: University of California Press, 2007.
Givoni, Michal. The Care of the Witness: A Con­temporary History of Testimony in Cri-
ses. ­Human Rights in History. Cambridge: Cambridge University Press, 2016.
Givoni, Michal. “Witnessing/Testimony.” Mafte’akh 11, no. 2 (2011): 147–69.
Glissant, Édouard. Poetics of Relation. Translated by Betsy Wing. Ann Arbor:
University of Michigan Press, 1997.
Goodfellow, Ian J., Jean Pouget-­Abadie, Mehdi Mirza, Bing Xu, David Warde-­
Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. “Generative
Adversarial Networks.” ArXiv, arXiv:1406.2661 (June 2014). https://­doi​.­org​
/­10​.­48550​/­arXiv​.­1406​.­2661.
GovernmentCIOMedia. “ai to Help Pentagon Prep for Algorithmic Warfare.”
June 11, 2016. https://­governmentciomedia​.­com​/­ai​-­help​-­pentagon​-­prep​
-­algorithmic​-­warfare.
Graae, Andreas Immanuel, and Kathrin Maurer, eds. Drone Imaginaries: The
Power of Remote Vision. Manchester: Manchester University Press, 2021.
Gray, Jonathan. “Data Witnessing: Attending to Injustice with Data in Amnesty
International’s Decoders Proj­ect.” Information, Communication, and Society
22, no. 7 (2019): 971–91.
Grayson, Kyle, and Jocelyn Mawdsley. “Scopic Regimes and the Visual Turn in In-
ternational Relations: Seeing World Politics through the Drone.” ­European
Journal of International Relations 25, no. 2 (2018).
Greene, Daniel. “Drone Vision.” Surveillance and Society 13, no. 2 (2015): 233–49.
Greenpeace. “Clicking Clean ­Virginia: The Dirty Energy Powering Data Center
Alley,” February 13, 2019. https://­www​.­greenpeace​.­org​/­usa​/­reports​/­click​
-­clean​-­virginia​/­.
Gregg, Melissa, and Gregory J. Seigworth. The Affect Theory Reader. Durham,
NC: Duke University Press, 2010.
Gregory, Derek. “From a View to a Kill: Drones and Late Modern War.” Theory,
Culture and Society 28, nos. 7–8 (2011): 188–215.
Gregory, Derek. “­Under Afghan Skies (1).” Geo­graph­i­cal Imaginations.
March 27, 2020. https://­geographicalimaginations​.­com​/­2020​/­03​/­27​/­under​
-­afghan​-­skies​-­1​/­.
Gregory, Derek. “­Under Afghan Skies (2).” Geo­graph­i­cal Imaginations. April 1, 2020.
https://­geographicalimaginations​.­com​/­2020​/­03​/­31​/­under​-­afghan​-­skies​-­2​/­.
Gregory, Derek. “­Under Afghan Skies (3).” Geo­graph­i­cal Imaginations. April 3, 2020.
https://­geographicalimaginations​.­com​/­2020​/­04​/­03​/­under​-­afghan​-­skies​-­3​/­.

Bibliography 215
Grove, Jairus. “From Geopolitics to Geotechnics: Global ­Futures in the Shadow
of Automation, Cunning Machines, and ­Human Speciation.” International
Relations 34, no. 3 (2020): 432–55.
Grove, Jairus Victor. Savage Ecol­ogy: War and Geopolitics at the End of the World.
Durham, NC: Duke University Press, 2019.
Grusin, Richard. “Introduction.” In The Nonhuman Turn, edited by Richard
Grusin, vii–­xxix. Minneapolis: University of Minnesota Press, 2015.
Grusin, Richard. Premediation: Affect and Mediality a­ fter 9/11. New York: Palgrave
Macmillan, 2010.
Guattari, Félix. Chaosmosis: An Ethico-­Aesthetic Paradigm. Translated by Paul
Bains and Julian Pefanis. Bloomington: Indiana University Press, 1995.
Guattari, Félix. “On Machines.” Journal of Philosophy and the Visual Arts 6 (1995):
8–12.
Guattari, Félix. The Three Ecologies. New York: Continuum, 2005.
Guerin, Frances, and Roger Hallas. The Image and the Witness: Trauma, Memory
and Visual Culture. London: Wallflower, 2007.
Halpern, Orit. Beautiful Data: A History of Vision and Reason since 1945. Durham,
NC: Duke University Press, 2015.
Haraway, Donna. “Anthropocene, Capitalocene, Plantationocene, Chthulucene:
Making Kin.” Environmental Humanities 6, no. 1 (2015): 159–65.
Haraway, Donna. Modest−Witness@Second−Millennium.FemaleMan−Meets−Onc-
oMouse™: Feminism and Technoscience. New York: Routledge, 1997.
Haraway, Donna. “Situated Knowledges: The Science Question in Feminism and
the Privilege of Partial Perspective.” Feminist Studies 14, no. 3 (1988): 575–99.
Haraway, Donna. Staying with the Trou­ble: Making Kin in the Chthulucene. Ex-
[103.158.136.64] Project MUSE (2024-02-06 19:57 GMT) University of Edinburgh

perimental ­Futures. Durham, NC: Duke University Press, 2016.

Hardwig, John. “The Role of Trust in Knowledge.” Journal of Philosophy 88, no. 12
(1991): 693–708.
Harkin, Natalie. “Anneal This Breath.” In Missile Park, edited by Australian Cen-
tre for Con­temporary Art, 86–87, 2021.
Hartman, Saidiya V. Scenes of Subjection: Terror, Slavery, and Self-­Making in
Nineteenth-­Century Amer­i­ca. New York: Oxford University Press, 2010.
Hau’ofa, Epeli. “Our Sea of Islands.” Con­temporary Pacific 6, no. 1 (1994): 147–61.
Herscher, Andrew. “Surveillant Witnessing: Satellite Imagery and the Visual Poli-
tics of ­Human Rights.” Public Culture 26, no. 3 (74) (2014): 469–500.
Hogan, Mél. “Facebook Data Storage Centers as the Archive’s Underbelly.”
­Television and New Media 16, no. 1 (2015): 3–18.
Hogan, Mél, and Asta Vonderau. “The Nature of Data Centers.” Culture Machine
19 (2019). https://­culturemachine​.­net​/­vol​-­18​-­the​-­nature​-­of​-­data​-­centers​/­the​
-­nature​-­of​-­data​-­centers​/­.
Holmes, Kevin. “An Artist and a ­Helicopter Capture Google’s Off-­Limits Data
Farm.” Vice, February 14, 2015. https://­www​.­vice​.­com​/­en​/­article​/­aen3xk​
/­artist​-­and​-­helicopter​-­capture​-­off​-­limits​-­data​-­farm.

216 Bibliography
Holmqvist, Caroline. “Undoing War: War Ontologies and the Materiality of
Drone Warfare.” Millennium 41, no. 3 (2013): 535–52.
Hoskins, Andrew, and Shona Illingworth. “Inaccessible War: Media, Memory,
Trauma and the Blueprint.” Digital War 1 (2020): 74–82.
­Human Rights Watch. “A Wedding That Became a Funeral: US Drone Attack on
Marriage Pro­cession in Yemen.” ­Human Rights Watch, February 19, 2014.
https://­www​.­hrw​.­org​/­report​/­2014​/­02​/­19​/­wedding​-­became​-­funeral​/­us​-­drone​
-­attack​-­marriage​-­procession​-­yemen.
ican: International Campaign to Abolish Nuclear Weapons. “Black Mist: The
Impact of Nuclear Weapons on Australia,” January 2014. https://­icanw​.­org​
.­au​/­wp​-­content​/­uploads​/­BlackMist​-­FINAL​-­Web​.­pdf.
Ingram, Haroro J. “Three Traits of the Islamic State’s Information Warfare.” rusi
Journal 159 (2014): 4–11.
Isereau, Darrek, et al. “Utilizing High-­Performance Embedded Computing,
Agile Condor, for Intelligent Pro­cessing: An Artificial Intelligence Plat-
form for Remotely Pi­loted Aircraft.” London: 2017 Intelligent Systems
Conference (IntelliSys) (September 2017): 1155–59. https://­doi​.­org​/­10​.­1109​
/­IntelliSys​.­2017​.­8324277.
Jackson, Zakiyyah Iman. “Animal: New Directions in the Theorization of Race
and Posthumanism.” Feminist Studies 39, no. 3 (2013): 669–85.
Jackson, Zakiyyah Iman. “Outer Worlds: The Per­sis­tence of Race in Movement
‘Beyond the ­Human.’ ” glq: A Journal of Lesbian and Gay Studies 21, no. 2
(2015): 215–18.
James, C. L. R. The Black Jacobins: Toussaint l’Ouverture and the San Domingo
Revolution. 2nd ed., rev. New York: Vintage, 1989.
Jetñil-­Kijiner, Kathy. “Tell Them.” 2011. https://­jkijiner​.­wordpress​.­com​/­2011​/­04​/­13​
/­tell​-­them​/­.
Johnson, Ted. “To ­Handle Its Influx of Drone Footage, Military Should Teach ai
to Watch tv.” Wired, November 26, 2017. https://­www​.­wired​.­com​/­story​/­the​
-­military​-­should​-­teach​-­ai​-­to​-­watch​-­drone​-­footage​/­.
Joint Standing Committee on Northern Australia. “Never Again: Interim Report
into the Destruction of Indigenous Heritage Sites at Juukan Gorge.” Can-
berra: Parliament of the Commonwealth of Australia, December 2020.
Joint Standing Committee on Northern Australia. “A Way Forward: Final Report
into the Destruction of Indigenous Heritage Sites at Juukan Gorge.” Can-
berra: Parliament of the Commonwealth of Australia, October 2021.
Jue, Melody. Wild Blue Media: Thinking through Seawater. Ele­ments. Durham,
NC: Duke University Press, 2020.
Jurgenson, Nathan. The Social Photo: On Photography and Social Media. New York:
Verso, 2019.
Kaleem, Jaweed. “Death on Facebook Now Common as ‘Dead Profiles’ Create Vast
Virtual Cemetery.” Huffington Post, October 1, 2015. https://­www​.­huffpost​
.­com​/­entry​/­death​-­facebook​-­dead​-­profiles​_­n​_­2245397.

Bibliography 217
Kapadia, Ronak K. Insurgent Aesthetics: Security and the Queer Life of the Forever
War. Durham, NC: Duke University Press, 2020.
Kaplan, Caren. Aerial Aftermaths: ­Wartime from Above. Durham, NC: Duke
University Press, 2018.
Kaplan, Caren. “Drone-­o-­Rama: Troubling the Spatial and Temporal Logics of
Distance Warfare.” In Life in the Age of Drone Warfare, edited by Lisa Parks
and Caren Kaplan, 161–77. Durham, NC: Duke University Press, 2017.
Kaplan, Caren. “Drones and the Image Complex: The Limits of Repre­sen­ta­tion in
the Era of Distance Warfare.” In Mediating the Spatiality of Conflicts: Inter-
national Conference Proceedings, edited by Armina Pilav, Marc Schoonder-
beek, Heidi Sohn, and Aleksandar Staničić, 29–43. Delft, Netherlands: bk,
2020.
Kaplan, E. Ann. Climate Trauma: Foreseeing the ­Future in Dystopian Film and Fic-
tion. New Brunswick, NJ: Rutgers University Press, 2016.
Kaplan, E. Ann. Trauma Culture: The Politics of Terror and Loss in Media and Lit­
er­a­ture. New Brunswick, NJ: Rutgers University Press, 2005.
Kelley, Robin D. G. “What Did Cedric Robinson Mean by Racial Capitalism?”
Boston Review, January 12, 2017. https://­bostonreview​.­net​/­race​/­robin​-­d​-­g​
-­kelley​-­what​-­did​-­cedric​-­robinson​-­mean​-­racial​-­capitalism.
Kember, Sarah, and Joanna Zylinska. Life ­after New Media: Mediation as a Vital
­Process. Cambridge, MA: mit Press, 2012.
Keskinen, Niina, Marja Kaunonen, and Anna Liisa Aho. “How Loved Ones Ex-
press Grief ­after the Death of a Child by Sharing Photo­graphs on Facebook.”
Journal of Loss and Trauma 24, no. 7 (2019): 609–24.
Khalili, Laleh. Sinews of War and Trade: Shipping and Capitalism in the Arabian
Peninsula. New York: Verso, 2020.
Kidder, Stanley Q., and Thomas H. Vonder Haar. Satellite Meteorology: An Intro-
duction. Cambridge, MA: Academic Press, 1995.
Kikerpill, Kristjan. “Choose Your Stars and Studs: The Rise of Deepfake Designer
Porn.” Porn Studies 7 no. 4 (2020): 1–5.
Kirchengast, Tyrone. “Deepfakes and Image Manipulation: Criminalisation and
Control.” Information and Communications Technology Law 29, no. 3 (2020):
308–23.
Kozol, Wendy. Distant Wars Vis­i­ble: The Ambivalence of Witnessing. Minneapolis:
University of Minnesota Press, 2014.
Kurgan, Laura. Close up at a Distance: Mapping, Technology, and Politics.
New York: Zone, 2013.
Kurgan, Laura. “Conflict Urbanism, Aleppo: Mapping Urban Damage.” Architec-
tural Design 87, no. 1 (2017): 72–77.
LaCapra, Dominick. Representing the Holocaust: History, Theory, Trauma. Ithaca,
NY: Cornell University Press, 1994.
LaCapra, Dominick. Writing History, Writing Trauma. Baltimore: Johns Hopkins
University Press, 2001.

218 Bibliography
Ladd, Mike. “The Lesser Known History of the Maralinga Nuclear Tests—­and
What It’s like to Stand at Ground Zero.” Australian Broadcasting Corpora-
tion, March 24, 2020. https://­www​.­abc​.­net​.­au​/­news​/­2020​-­03​-­24​/­maralinga​
-­nuclear​-­tests​-­ground​-­zero​-­lesser​-­known​-­history​/­11882608.
Langbein, John. Torture and the Law of Proof. Chicago: University of Chicago
Press, 1977.
Lapoujade, David. William James, Empiricism and Pragmatism. Translated by Thomas
LaMarre. Thought in the Act. Durham, NC: Duke University Press, 2020.
Leaver, Tama. “The Social Media Contradiction: Data Mining and Digital Death.”
m/c Journal 16 (2013). https://­journal​.­media​-­culture​.­org​.­au​/­index​.­php​
/­mcjournal​/­article​/­view​/­625.
Liu, Cixin. Death’s End. Bk. 3 of The Three-Body Problem, Translated by Ken Liu.
London: Head of Zeus, 2016.
Luciano, Dana, and Mel Y. Chen. “Has the Queer Ever Been ­Human?” glq: A
Journal of Lesbian and Gay Studies 21, nos. 2–3 (2015): 183–207.
Lyons, Kate. “Fiji pm Accuses Scott Morrison of ‘Insulting’ and Alienating Pacific
Leaders.” Guardian, August 17, 2019. https://­www​.­theguardian​.­com​/­world​
/­2019​/­aug​/­16​/­fiji​-­pm​-­frank​-­bainimarama​-­insulting​-­scott​-­morrison​-­rift​
-­pacific​-­countries.
Mackenzie, Adrian. Machine Learners: Archaeology of a Data Practice. Cam-
bridge, MA: mit Press, 2017.
Mackenzie, Adrian, and Anna Munster. “Platform Seeing: Image Ensembles and
Their Invisualities.” Theory, Culture and Society 36, no. 5 (2019): 3–22.
Maddocks, Sophie. “‘A Deepfake Porn Plot Intended to Silence Me’: Exploring Con-
tinuities between Pornographic and ‘­Political’ Deep Fakes.” Porn (2020): 1–9.
Manjoo, Farhad. “I Tried Microsoft’s Flight Simulator. The Earth Never Seemed
So Real.” New York Times, August 19, 2020. https://­www​.­nytimes​.­com​/­2020​
/­08​/­19​/­opinion​/­microsoft​-­flight​-­simulator​.­html.
Manning, Erin. The Minor Gesture. Durham, NC: Duke University Press, 2016.
Manning, Erin, and Brian Massumi. Thought in the Act: Passages in the Ecol­ogy of
Experience. Minneapolis: University of Minnesota Press, 2014.
Maralinga Rehabilitation Technical Advisory Committee. “Rehabilitation of For-
mer Nuclear Test Sites at Emu and Maralinga (Australia).” Canberra: De-
partment of Education, Science, and Training, 2003. https://­www​.­industry​
.­gov​.­au​/­sites​/­default​/­files​/­July%202018​/­document​/­pdf​/­rehabilitation​-­of​
-­former​-­nuclear​-­test​-­sites​-­at​-­emu​-­and​-­maralinga​.­pdf ​?­acsf​_­files​_­redirect.
Maras, Marie-­Helen, and Alex Alexandrou. “Determining Authenticity of Video
Evidence in the Age of Artificial Intelligence and in the Wake of Deepfake
Videos.” International Journal of Evidence and Proof 23, no. 3 (2019): 255–62.
Massumi, Brian. “The Autonomy of Affect.” Cultural Critique 31 (1995): 83–109.
Massumi, Brian. “The ­Future Birth of the Affective Fact: The ­Political Ontology of
Threat.” In The Affect Theory Reader, edited by Melissa Gregg and Gregory J.
Seigworth. Durham, NC: Duke University Press, 2010.

Bibliography 219
Massumi, Brian. Ontopower: War, Powers, and the State of Perception. Durham,
NC: Duke University Press, 2015.
Massumi, Brian. Parables for the Virtual: Movement, Affect, Sensation. Post-­
contemporary Interventions. Durham, NC: Duke University Press, 2002.
Maurer, Anaïs. “Snaring the Nuclear Sun: Decolonial Ecologies in Titaua Peu’s
Mutismes: E ‘Ore Te Vāvā.” Con­temporary Pacific 32, no. 2 (2020): 371–97.
Mbembe, Achille. Necropolitics. Durham, NC: Duke University Press, 2019.
McCosker, Anthony. “Drone Media: Unruly Systems, Radical Empiricism and
Camera Consciousness.” Culture Machine 16 (2015). https://­culturemachine​
.­net​/­vol​-­16​-­drone​-­cultures​/­drone​-­media​/­.
McCosker, Anthony. “Drone Vision, Zones of Protest, and the New Camera Con-
sciousness.” Media Fields Journal 9 (2015). http://­www​.­mediafieldsjournal​
.­org​/­drone​-­vision​-­zones​-­of​-­protest​/­2015​/­8​/­21​/­drone​-­vision​-­zones​-­of​-­protest​
-­and​-­the​-­new​-­camera​-­consciousne​.­html.
McCosker, Anthony. “Making Sense of Deepfakes: Socializing ai and Building
Data Literacy on GitHub and YouTube.” New Media and Society (May 2022).
https://­doi​.­org​/­10​.­1177​/­14614448221093943.
McCosker, Anthony, and Rowan Wilken. Automating Vision: The Social Impact of
the New Camera Consciousness. New York: Routledge, 2020.
McDonald, Matt. Ecological Security. Cambridge: Cambridge University Press, 2021.
McKittrick, Katherine. Dear Science and Other Stories. Errantries. Durham, NC:
Duke University Press, 2021.
McLelland, J. R. The Report of the Royal Commission into British Nuclear Tests
in Australia. Vol. 1. Parliamentary Paper. Canberra: The Parliament of the
Commonwealth of Australia, 1985. https://­parlinfo​.­aph​.­gov​.­au​/­parlInfo​
/­download​/­publications​/­tabledpapers​/­HPP032016010928​/­upload​_­pdf​
/­HPP032016010928​.­pdf.
McNeill, J. R., and Peter Engelke. The G ­ reat Acceleration: An Environmental History
of the Anthropocene since 1945. Cambridge, MA: Harvard University Press, 2014.
Michel, Arthur Holland. Eyes in the Sky: The Secret Rise of Gorgon Stare and How
It ­Will Watch Us All. Boston: Houghton Mifflin Harcourt, 2019.
Miglio. “ai in Unreal Engine: Learning through Virtual Simulations.” Unreal
Engine Blog, April 13, 2018. https://­www​.­unrealengine​.­com​/­en​-­US​/­tech​-­blog​
/­ai​-­in​-­unreal​-­engine​-­learning​-­through​-­virtual​-­simulations.
Mirzoeff, Nicholas. How to See the World: An Introduction to Images, from Self-­
Portraits to Selfies, Maps to Movies, and More. New York: Penguin, 2015.
Mirzoeff, Nicholas. The Right to Look: A Counterhistory of Visuality. Durham,
NC: Duke University Press, 2011.
Mittmann, J. D. “Maralinga: Aboriginal Poison Country.” Agora 25 (2017): 7.
Moore, Jason W., ed. Anthropocene or Capitalocene? Nature, History, and the
Crisis of Capitalism. Oakland, CA: pm, 2016.
Moreton-­Robinson, Aileen. The White Possessive. Minneapolis: University of
Minnesota Press, 2015.

220 Bibliography
Morton, Timothy. Hyperobjects: Philosophy and Ecol­ogy ­after the End of the
World. Minneapolis: University of Minnesota Press, 2013.
Murphie, Andrew. “On Being Affected: Feeling in the Folding of Multiple Catas-
trophes.” Cultural Studies 32, no. 1 (2018): 18–42.
Murphie, Andrew. “World as Medium: A Whiteheadian Media Philosophy.” In
Immediation, edited by Erin Manning, Anna Munster, and Bodil Marie
Stavning Thomsen, 16–46. Detroit: Open Humanities, 2019.
Nagel, Emily van der. “Verifying Images: Deepfakes, Control, and Consent.” Porn
Studies 7, no. 4 (2020): 424–29.
Neale, Timothy, Alex Zahara, and ­Will Smith. “An Eternal Flame: The Elemen-
tal Governance of Wildfire’s Pasts, Pre­sents and ­Futures.” Cultural Studies
Review 25, no. 2 (2019): 115–34.
Nixon, Rob. Slow Vio­lence and the Environmentalism of the Poor. Cambridge,
MA: Harvard University Press, 2011.
Noble, Safiya Umoja. Algorithms of Oppression: How Search Engines Reinforce
Racism. New York: New York University Press, 2018.
Nyong’o, Tavia. “­Little Monsters: Race, Sovereignty, and Queer Inhumanism in
Beasts of the Southern Wild.” glq: A Journal of Lesbian and Gay Studies 21,
no. 2–3 (2015): 249–72.
Offert, Fabian. “Latent Deep Space: Generative Adversarial Networks (gans) in
the Sciences.” Media+Environment 3, no. 2 (2021).
Offert, Fabian, and Thao Phan. “A Sign That Spells: DALL-­E 2, Invisual Images
and the Racial Politics of Feature Space.” ArXiv, arXiv:2211.06323 (Octo-
ber 2022). http://­arxiv​.­org​/­abs​/­2211​.­06323.
Öhman, Carl J., and David Watson. “Are the Dead Taking over Facebook? A Big
Data Approach to the ­Future of Death Online.” Big Data and Society 6, no. 1
(2019). https://­doi​.­org​/­10​.­1177​/­2053951719842540.
Oliver, Kelly. Witnessing: Beyond Recognition. Minneapolis: University of Min-
nesota Press, 2001.
O’Malley, Pat, and Gavin J. D. Smith. “ ‘Smart’ Crime Prevention? Digitization
and Racialized Crime Control in a Smart City.” Theoretical Criminology 26,
no. 1 (2020): 40–56.
OpenAI. “Proximal Policy Optimization.” July 20, 2017. https://­openai​.­com​/­blog​
/­openai​-­baselines​-­ppo​/­.
Packer, Jeremy, and Joshua Reeves. “Taking People Out: Drones, Media/Weapons, and
the Coming Humanectomy.” In Life in the Age of Drone Warfare, edited by Lisa
Parks and Caren Kaplan, 261–81. Durham, NC: Duke University Press, 2017.
Packer, Jeremy, and Joshua Reeves. Killer Apps: War, Media, Machine. Durham,
NC: Duke University Press, 2020.
Papailias, Penelope. “Witnessing in the Age of the Database: Viral Memorials, Af-
fective Publics, and the Assemblage of Mourning.” Memory Studies 9, no. 4
(2016): 437–54.

Bibliography 221
Parks, Lisa. Cultures in Orbit: Satellites and the Televisual. Console-­Ing Passions.
Durham, NC: Duke University Press, 2005.
Parks, Lisa. Rethinking Media Coverage: Vertical Mediation and the War on Terror.
New York: Routledge, 2018.
Parks, Lisa, and Caren Kaplan, eds. Life in the Age of Drone Warfare. Durham,
NC: Duke University Press, 2017.
Pasquale, Frank. The Black Box Society: The Secret Algorithms That Control Money
and Information. 1st ed. Cambridge, MA: Harvard University Press, 2015.
Paterson, Thomas, and Lauren Hanley. “­Political Warfare in the Digital Age:
Cyber Subversion, Information Operations and ‘Deep Fakes.’ ” Australian
Journal of International Affairs 74, no. 4 (2020): 439–54.
Peters, Edward. Torture. New York: Basil Blackwell, 1985.
Peters, John Durham. The Marvelous Clouds. Chicago: University of Chicago
Press, 2015.
Peters, John Durham. “Witnessing.” Media, Culture and Society 23, no. 6 (2001):
707–23.
Phan, Thao, and Scott Wark. “Racial Formations as Data Formations.” Big Data
and Society 8, no. 2 (2021).
Phan, Thao, and Scott Wark. “What Personalisation Can Do for You! Or: How to
Do Racial Discrimination without ‘Race.’ ” Culture Machine, 2021. https://­
culturemachine​.­net​/­vol​-­20​-­machine​-­intelligences​/­what​-­personalisation​-­can​
-­do​-­for​-­you​-­or​-­how​-­to​-­do​-­racial​-­discrimination​-­without​-­race​-­thao​-­phan​
-­scott​-­wark​/­.
Phan, Thao, Jake Goldenfein, Monique Mann, and Declan Kuch, editors. Econo-
mies of Virtue—­the Circulation of “Ethics” in ai. Amsterdam: Institute of
Network Cultures, 2022.
Pinchevski, Amit. Transmitted Wounds: Media and the Mediation of Trauma.
New York: Oxford University Press, 2019.
Pong, Beryl. British Lit­er­a­ture and Culture in Second World ­Wartime: For the Dura-
tion. Oxford Mid-­century Studies. New York: Oxford University Press, 2020.
Popova, Milena. “Reading out of Context: Pornographic Deepfakes, Celebrity,
and Intimacy.” Porn Studies 7 no. 4 (2019): 1–15.
Puar, Jasbir. “The Cost of Getting Better: Suicide, Sensation, Switchpoints.” glq:
A Journal of Lesbian and Gay Studies 18 (2012): 149–58.
Puar, Jasbir. Right to Maim: Debility, Capacity, Disability. Durham, NC: Duke
University Press, 2017.
Pugliese, Joseph. Biopolitics of the More-­Than-­Human: Forensic Ecologies of Vio­
lence. Durham, NC: Duke University Press, 2020.
Pugliese, Joseph. “Death by Metadata: The Bioinformationalisation of Life and
the Transliteration of Algorithms to Flesh.” In Security, Race, Biopower:
Essays on Technology and Corporeality, edited by Holly Randell-­Moon and
Ryan Tippet, 3–20. London: Palgrave Macmillan, 2016.

222 Bibliography
Pugliese, Joseph. “Drone Casino Mimesis: Telewarfare and Civil Militarization.”
Journal of Sociology 52, no. 3 (2016): 500–21.
Pugliese, Joseph. State Vio­lence and the Execution of Law: Biopo­liti­cal Caesurae of
Torture, Black Sites, Drones. New York: Routledge, 2013.
Rae, Maria, Rosa Holman, and Amy Nethery. “Self-­Represented Witnessing: The
Use of Social Media by Asylum Seekers in Australia’s Offshore Immigration
Detention Centres.” Media, Culture and Society 40, no. 4 (2018): 479–95.
Rancière, Jacques. Disagreement: Politics and Philosophy. Translated by Julie ­Rose.
Minneapolis: University of Minnesota Press, 1999.
Ray, Una. “The Myth of Empty Country and the Story of ‘Deadly’ Glass.” Di’van,
no. 9 (2021): 42–55.
Reading, Anna. “Mobile Witnessing: Ethics and the Camera Phone in the ‘War on
Terror.’ ” Globalizations 6, no. 1 (2009): 61–76.
Reiter, Bernd, ed. Constructing the Pluriverse: The Geopolitics of Knowledge.
­Durham, NC: Duke University Press, 2018.
Rhee, Jennifer. The Robotic Imaginary: The ­Human and the Price of Dehumanized
­Labor. Minneapolis: University of Minnesota Press, 2018.
Richardson, Michael. “Climate Trauma, or the Affects of the Catastrophe to
Come.” Environmental Humanities 10, no. 1 (2018): 1–19.
Richardson, Michael. “Drone’s-­Eye View: Affective Witnessing and Technicities
of Perception.” In Image Testimonies: Witnessing in Times of Social Media,
edited by Kerstin Schankweiler, Verena Straub, and Tobias Wendl, 64–74.
New York: Routledge, 2018.
Richardson, Michael. Gestures of Testimony: Torture, Trauma, and Affect in Lit­er­a­
ture. New York: Bloomsbury Academic, 2016.
Richardson, Michael. “­There’s Something ­Going On.” Capacious 1, no. 2 (2018):
149–54.
Richardson, Michael, and Kerstin Schankweiler. “Affective Witnessing.” In Affec-
tive Socie­ties: Key Concepts, edited by Jan Slaby and Christian von Scheve,
166–177. New York: Routledge, 2019.
Richardson, Michael, and Kerstin Schankweiler. “Introduction: Affective Wit-
nessing as Theory and Practice.” Parallax 26, no. 3 (2020): 235–53.
Rini, Regina. “Deepfakes and the Epistemic Backstop.” ­Philosophers’ Imprint 20,
no. 24 (2020): 1–16.
Ristovska, Sandra. Seeing H ­ uman Rights: Video Activism as a Proxy Profession.
Information Policy. Cambridge, MA: mit Press, 2021.
Robinson, Cedric J. Black Marxism: The Making of the Black Radical Tradition.
Chapel Hill: University of North Carolina Press, 2000.
Rosenthal, Caitlin. Accounting for Slavery: Masters and Management. Cambridge,
MA: Harvard University Press, 2019.
Rothe, Delf. “Seeing Like a Satellite: Remote Sensing and the Ontological Politics
of Environmental Security.” Security Dialogue 48, no. 4 (2017): 334–53.

Bibliography 223
Rouvroy, Antoinette, and Thomas Berns. “Algorithmic Governmentality and
Prospects of Emancipation.” Reseaux 177, no. 1 (2013): 163–96.
Russill, Chris. “Earth Imaging: Photo­graph, Pixel, Program.” In Ecomedia: Key
Issues, edited by Stephen Rust, Salma Monani, and Sean Cubitt, 228–50.
New York: Routledge, 2015.
Russill, Chris. “Is the Earth a Medium? Situating the Planetary in Media Theory.”
Ctrl​-­z​.­Net​.­Au 7 (2017). http://­www​.­ctrl​-­z​.­net​.­au​/­articles​/­issue​-­7​/­russill​-­is​-­the​
-­earth​-­a​-­medium​/­.
Russill, Chris. “The Road Not Taken: William James’s Radical Empiricism and
Communication Theory.” Communication Review 8, no. 3 (2005): 277–305.
Sadowski, Jathan. “Potemkin ai.” Real Life, August 6, 2018. https://­reallifemag​
.­com​/­potemkin​-­ai​/­.
Sadowski, Jathan. Too Smart: How Digital Capitalism Is Extracting Data, Control-
ling Our Lives, and Taking over the World. Cambridge, MA: mit Press, 2020.
Sadowski, Jathan. “When Data Is Capital: Datafication, Accumulation, and Ex-
traction.” Big Data and Society 6, no. 1 (2019).
Safransky, Sara. “Geographies of Algorithmic Vio­lence: Redlining the Smart City.”
International Journal of Urban and Regional Research 44, no. 2 (2020): 200–18.
Saldarriaga, Juan Francisco, Laura Kurgan, and Dare Brawley. “Visualizing Conflict:
Possibilities for Urban Research.” Urban Planning 2, no. 1 (2017): 100–107.
Savransky, Martin. Around the Day in Eighty Worlds: Politics of the Pluriverse.
Thought in the Act. Durham, NC: Duke University Press, 2021.
Scannell, Josh. “What Can an Algorithm Do?” dis Magazine. Accessed July 14,
2016. http://­dismagazine​.­com​/­discussion​/­72975​/­josh​-­scannell​-­what​-­can​-­an​
-­algorithm​-­do​/­.
Scannell, Paddy. ­Television and the Meaning of Live: An Enquiry into the ­Human
Situation. Cambridge, UK: Polity, 2014.
Scarce, Yhonnie, Max Delany, and Australian Centre for Con­temporary Art. Mis-
sile Park, 2021.
Schaefer, Donovan O. The Evolution of Affect Theory: The Humanities, the
Sciences, and the Study of Power. Cambridge Ele­ments. Cambridge:
Cambridge University Press, 2019.
Schankweiler, Kerstin, Verena Straub, and Tobias Wendl, eds. Image Testimonies:
Witnessing in Times of Social Media. New York: Routledge, 2018.
Schuppli, Susan. Material Witness: Media, Forensics, Evidence. Cambridge, MA:
mit Press, 2020.
Sear, Tom. “Xenowar Dreams of Itself.” Digital War, 1 (2020). https://­doi​.­org​/­10​
.­1057​/­s42984​-­020​-­00019​-­6.
Seaver, Nick. “Algorithms as Culture: Some Tactics for the Ethnography of Algo-
rithmic Systems.” Big Data and Society 4, no. 2 (2017).
Seigworth, Gregory J., and Matthew Tiessen. “Mobile Affects, Open Secrets, and
Global Illiquidity: Pockets, Pools, and Plasma.” Theory, Culture and Society
29 (2012): 47–77.

224 Bibliography
Seltzer, Mark. “Wound Culture: Trauma in the Pathological Public Sphere.”
October 80 (1997): 3–26.
Sheikh, Shela. “The ­Future of the Witness: Nature, Race and More-­than-­Human
Environmental Publics.” Kronos 44, no. 1 (2018): 145–62.
Sherwood, Dave. “Inside Lithium ­Giant sqm’s Strug­gle to Win Over Indigenous
Communities in Chile’s Atacama.” ­Reuters, January 15, 2021. https://­www​
.­reuters​.­com​/­article​/­us​-­chile​-­lithium​-­sqm​-­focus​-­idUSKBN29K1DB.
Singh, Julietta. Unthinking Mastery: Dehumanism and Decolonial Entanglements.
Durham, NC: Duke University Press, 2018.
Slaughter, Joseph R. ­Human Rights, Inc.: The World Novel, Narrative Form, and
International Law. New York: Fordham University Press, 2007.
Smith, Joan. Clouds of Deceit: The Deadly Legacy of Britain’s Bomb Tests. London:
Faber, 1985.
Snaza, Nathan. “The Earth Is Not ‘Ours’ to Save.” In Interrogating the Anthropo-
cene: Ecol­ogy, Aesthetics, Pedagogy, and the ­Future in Question, edited by
Jan Jagodzinski, 339–57. Palgrave Studies in Educational ­Futures. Cham,
Switzerland: Springer International, 2018.
Sontag, Susan. Regarding the Pain of O ­ thers. London: Penguin, 2003.
Sparrow, Jeff. “Plastic Sword the Least of asio’s Bungles in ‘Terror Raid.’ ” Crikey,
October 9, 2014. https://­www​.­crikey​.­com​.­au​/­2014​/­10​/­09​/­plastic​-­sword​-­the​
-­least​-­of​-­asios​-­bungles​-­in​-­terror​-­raid​/­.
Spillers, Hortense J. “Mama’s Baby, Papa’s Maybe: An American Grammar Book.”
Diacritics 17, no. 2 (1987): 65–81.
Springgay, Stephanie, and Sarah E. Truman. Walking Methodologies in a More-­
than-­Human World: WalkingLab. New York: Routledge, 2018.
src Defense. “Agile Condor™ High-­Performance Embedded Computing Archi-
tecture.” YouTube video, 1:57, October 14, 2016. https://­www​.­youtube​.­com​
/­watch​?­v​=­sc1aOFmb3AI.
Stahl, Roger. Through the Crosshairs: War, Visual Culture, and the Weaponized
Gaze. New Brunswick, NJ: Rutgers University Press, 2018.
Stanford Law School and nyu School of Law. “Living ­under Drones: Death,
Injury, and Trauma to Civilians from US Drone Practices in Pakistan.”
International ­Human Rights and Conflict Resolution Clinic at Stanford Law
School and Global Justice Clinic at nyu School of Law, 2012.
Stark, Luke. “Facial Recognition Is the Plutonium of ai.” xrds: Crossroads, the
acm Magazine for Students 25, no. 3 (2019): 50–55.
Starosielski, Nicole. Media Hot and Cold. Ele­ments. Durham, NC: Duke Univer-
sity Press, 2021.
Stewart, Kathleen. “Afterword: Worlding Refrains.” In The Affect Theory Reader,
edited by Melissa Gregg and Gregory J. Seigworth, 339–53. Durham, NC:
Duke University Press, 2010.
Stewart, Kathleen. “Atmospheric Attunements.” Environment and Planning D:
Society and Space 29, no. 3 (2011): 445–53.

Bibliography 225
Stiegler, Bernard. “Anamnesis and Hypomnesis.” Ars Industrialis, March 7, 2016.
https://­arsindustrialis​.­org​/­anamnesis​-­and​-­hypomnesis.
Stubblefield, Thomas. Drone Art: The Everywhere War as Medium. Oakland, CA:
University of California Press, 2020.
Suchman, Lucy. “Algorithmic Warfare and the Reinvention of Accuracy.” Critical
Studies on Security 8, no. 2 (2020): 175–87.
SyncedReview. “Barack Obama Is the Benchmark for Fake Lip-­Sync Videos.” Me-
dium, January 17, 2018. https://­medium​.­com​/­syncedreview​/­barack​-­obama​-­is​
-­the​-­benchmark​-­for​-­fake​-­lip​-­sync​-­videos​-­d85057cb90ac.
Taffel, Sy. “Data and Oil: ­Metaphor, Materiality, and Metabolic Rifts.” New Media
and Society, 25, no 5 (2021): 980–98.
Tahir, Madiha. “The Distributed Empire of the War on Terror.” Boston Review,
September 9, 2021. https://­bostonreview​.­net​/­global​-­justice​/­madiha​-­tahir​
-­war​-­on​-­terror​-­empire​-­pakistan.
Tahir, Madiha. “The Ground Was Always in Play.” Public Culture 29, no. 1 (2017):
5–16.
Taussig, Michael T. Law in a Lawless Land: Diary of a “Limpieza” in Colombia.
Chicago: University of Chicago Press, 2005.
Taylor, Christopher. Empire of Neglect: The West Indies in the Wake of British Lib-
eralism. Radical Américas. Durham, NC: Duke University Press, 2018.
Taylor, Simon. “Fiction Machines: How Drones Read Oceanic Volumes through
‘Technically-­Induced Hallucination.’ ” In Drone Aesthetics: War, Cultures,
Ecologies, edited by Beryl Pong and Michael Richardson. London: Open
Humanities, 2024.
Thrift, Nigel. Non-­representational Theory : Space, Politics, Affect. 1st ed. Hoboken,
NJ: Taylor and Francis, 2008.
Tomkins, Silvan. Shame and Its S­ isters: A Silvan Tomkins Reader. Durham, NC:
Duke University Press, 1995.
Tsing, A. L. “On Nonscalability: The Living World Is Not Amenable to Precision-­
Nested Scales.” Common Knowledge 18, no. 3 (2012): 505–24.
Turner, Fred. From Counterculture to Cyberculture: Stewart Brand, the ­Whole
Earth Network, and the Rise of Digital Utopianism. Chicago: University of
Chicago Press, 2006.
Tynan, Elizabeth. Atomic Thunder: The Maralinga Story. Sydney: NewSouth, 2016.
Tynan, Elizabeth. “Sixty Years on, Maralinga Reminds Us Not to Put Security
over Safety.” Conversation, September 25, 2016. http://­theconversation​.­com​
/­sixty​-­years​-­on​-­maralinga​-­reminds​-­us​-­not​-­to​-­put​-­security​-­over​-­safety​
-­62441.
Tynan, Lauren. “What Is Relationality? Indigenous Knowledges, Practices, and
Responsibilities with Kin.” Cultural Geographies, 28, no. 4 (2021): 597–610.
Uliasz, Rebecca. “On the Truth Claims of Deepfakes: Indexing Images and
Semantic Forensics.” Journal of Media Art Study and Theory 3, no. 1 (2022):
63–84.

226 Bibliography
un News. “ ‘Crimes of Historic Proportions’ Being Committed in Aleppo, un
Rights Chief Warns.” un News, October 21, 2016. https://­news​.­un​.­org​/­en​
/­story​/­2016​/­10​/­543432​-­crimes​-­historic​-­proportions​-­being​-­committed​
-­aleppo​-­un​-­rights​-­chief​-­warns.
Valla, Clement. Postcards from Google Earth, 2012. http://­www​.­postcards​-­from​
-­google​-­earth​.­com​/­.
Vigh, Henrik. “Crisis and Chronicity: Anthropological Perspectives on Continu-
ous Conflict and Decline.” Ethnos 73, no. 1 (2008): 5–24.
Virilio, Paul. War and Cinema: The Logistics of Perception. New York: Verso, 1989.
Vivian, Bradford. Commonplace Witnessing: Rhetorical Invention, Historical Re-
membrance, and Public Culture. New York: Oxford University Press, 2017.
Wall, T., and T. Monahan. “Surveillance and Vio­lence from Afar: The Politics
of Drones and Liminal Security-­Scapes.” Theoretical Criminology 15, no. 3
(2011): 239–54.
Weizman, Eyal. Forensic Architecture: Vio­lence at the Threshold of Detectability.
New York: Zone, 2017.
Westfall, Sammy. “Australia Ranks Last on Climate Action in U.N. Report.”
Washington Post, July 2, 2021. https://­www​.­washingtonpost​.­com​/­world​
/­2021​/­07​/­02​/­australia​-­climate​-­action​-­un​-­sustainable​-­development​/­.
Weston, Kath. Animate Planet: Making Visceral Sense of Living in a High-­Tech, Eco-
logically Damaged World. anima. Durham, NC: Duke University Press, 2017.
Whittaker, Meredith. “The Steep Cost of Capture.” Interactions 28, no. 6 (2021):
50–55.
Whyte, Christopher. “Deepfake News: ai-­Enabled Disinformation as a Multi-­level
Public Policy Challenge.” Journal of Cyber Policy 5, no. 2 (2020): 199–217.
Whyte, Jessica. The Morals of the Market: H ­ uman Rights and the Rise of Neoliber-
alism. New York: Verso, 2019.
Whyte, Kyle. “Against Crisis Epistemology.” In Routledge Handbook of Critical In-
digenous Studies, edited by Brendan Hokowhitu, Aileen Moreton-­Robinson,
Linda Tuhiwai-­Smith, Chris Andersen, Steve Larkin, and Brendan Hokow-
hitu, 52–64. New York: Routledge, 2021.
Whyte, Kyle. “Indigenous Science (Fiction) for the Anthropocene: Ancestral Dys-
topias and Fantasies of Climate Change Crises.” Environment and Planning
E: Nature and Space 1, nos. 1–2 (2018): 224–42.
Whyte, Kyle. “Settler Colonialism, Ecol­ogy, and Environmental Injustice.” Envi-
ronment and Society 9, no. 1 (2018): 125–44.
Wieviorka, Annette. The Era of the Witness. Ithaca, NY: Cornell University Press,
2006.
Wilcox, Lauren. “Embodying Algorithmic War: Gender, Race, and the Posthu-
man in Drone Warfare.” Security Dialogue 48, no. 1 (2017): 11–28.
Wilken, Rowan, and Julian Thomas. “Vertical Geomediation: The Automation
and Platformization of Photogrammetry.” New Media and Society 24, no. 11
(2022): 2531–47.

Bibliography 227
Williams, Raymond. Marxism and Lit­er­a­ture, Oxford: Oxford University Press, 1977.
Wise, Jeff. “How Crazy Am I to Think I Actually Know Where That Malaysia
Airlines Plane Is?” New York Magazine, February 23, 2015. https://­nymag​
.­com​/­intelligencer​/­2015​/­02​/­jeff​-­wise​-­mh370​-­theory​.­html.
witness Media Lab. “Mal-­Uses of ai-­Generated Synthetic Media and Deepfakes:
Pragmatic Solutions Discovery Convening.” witness, June 11, 2018.
Wolfe, Patrick. “Settler Colonialism and the Elimination of the Native.” Journal of
Genocide Research 8, no. 4 (2006): 387–409.
Woods, Derek. “Scale Critique for the Anthropocene.” Minnesota Review 2014,
no. 83 (2014): 133–42.
Wu, Xiaoping, and Martin Montgomery. “Witnessing in Crisis Contexts in the
Social Media Age: The Case of the 2015 Tianjin Blasts on Weibo.” Media,
Culture and Society 42, no. 5 (2020): 675–91.
Wynter, Sylvia. “Unsettling the Coloniality of Being/Power/Truth/Freedom:
­Towards the H ­ uman, ­After Man, Its Overrepresentation—an Argument.”
cr: The New Centennial Review 3, no. 3 (2003): 257–337.
Yang, Jun, Peng Gong, Rong Fu, Minghua Zhang, Jingming Chen, Shunlin Liang,
Bing Xu, Jiancheng Shi, and Robert Dickinson. “The Role of Satellite Re-
mote Sensing in Climate Change Studies.” Nature Climate Change 3, no. 10
(2013): 875–83.
Yunkaporta, Tyson. Sand Talk: How Indigenous Thinking Can Save the World.
Melbourne: Text, 2019.
Yusoff, Kathryn. A Billion Black Anthropocenes or None. Minneapolis: University
of Minnesota Press, 2018.
Zylinska, Joanna. Minimal Ethics for the Anthropocene. Ann Arbor, MI: Open
Humanities, 2014.
Zylinska, Joanna. Nonhuman Photography. Cambridge, MA: mit Press, 2017.

228 Bibliography
Index

Italicized page numbers refer to figures. aesthetic interventions and artwork, 13,
98; about nuclear testing, 21, 115, 138–45;
Aboriginal ­peoples, 18, 146–47, 205n55; drones in, 34–35, 37–39, 53, 107–10, 191n1,
­Aboriginal Heritage Act of 1972 (Aus- 192n9, 193n33, 193n43; ecological trauma
tralia), 168–70; military cooptation of and, 117, 125–29, 138–46, 149, 154; insur-
Aboriginal languages, 202n78; nuclear gent aesthetics, 53; significance to the
weapons testing and, 133–39, 141–42, book, 9, 21, 23; and trauma theory, 29–30.
144–45; Stolen Generation, 187n32. See also individual artworks and artists
See also First Nations ­peoples; Indig- affect theory, 6–7, 185n5. See also machinic
enous ­peoples; individual nations and affect; traumatic affect
­peoples ­Afghanistan, 32, 38, 46, 52–53, 55, 57, 150,
Abrar, Hisham, 193n37 192n9; Pakistan border, 50, 101–2, 194n51;
activism, 104, 119, 154, 192n13, 200n27; of Uruzgan Province, 1–3, 7, 67
Aleppo Media Center, 60, 63–­64; anti- Agile Condor (src Inc.), 43, 49, 59, 66–74,
nuclear, 137, 144, 202n75; call for cloud 76
ethics, 110; of Forensic Architecture, Air Force Research Laboratory (afrl), 66;
20–21, 87, 93–99, 109, 178; humanitarian Moving and Stationary Target Acquisi-
testimony and, 4, 22, 24, 32, 42–43, 95, tion and Recognition (mstar), 69–70
183, 192n9; of Indigenous ­peoples, 6, 18, Alaimo, Stacy, 122, 190n86
20–21, 115, 137, 144, 169; prodemocracy al-­Asad, Bashir, 59–60
protests in Daraa, 59; protests at US–­ Aleppo Media Center, 60, 63–­64
Mexico border, 94; of witness, 90–91. Aleppo, Syria: aftermath of war in, 43,
See also aesthetic interventions and 59–65, 98, 119; ­Battle of Aleppo, 60; Con-
artwork flict Urbanism proj­ect and, 60–63, 65
Alexander, David R., 68 Atacan community of Chile, 12
Alexis-­Martin, Becky, 202n80 Atkinson, Meera, 158
Al-­Ghazzi, Omar, 63 Australia, 18, 37, 104, 107, 115; Aboriginal
algorithmic enclosure, 4–6, 8, 11, 118, 192n8; Heritage Act of 1972, 168–70; Atomic
autonomous warfare and, 2, 12–14, 42, 56, Weapons Test Safety Committee
79, 83; ethics and, 84–88, 91–92 (awtsc), 201n63; Brockman 4 iron ore
Algorithmic Warfare Cross-­Functional mine, 168; climate crisis and, 17, 112–13,
Team (awcft), 99–100, 102, 109. See also 132, 148; “Digital Earth Australia,” 125;
Proj­ect Maven Eora Nation, 7; Juukan Gorge, 167–71,
Althusser, Louis, 190n82 205n55; Maralinga, 133–39, 141–42, 144,
Amazon, 11–12, 14, 75 176; Stolen Generation, 187n32; terrorist
Amoore, Louise, 77, 85–86, 96, 101, 110, raids in, 152; Woomera, 138, 142, 202n78
192n8 author positionality, 7, 17–18
ancestral dystopias, 6 Autonomous Real-­Time Ground Ubiq-
Anderson, Stephen F., 92 uitous Surveillance Imaging System
Andrejevic, Mark, 46, 69, 72, 78, 195n86 (argus-­i s), 47–48, 100
Angerer, Marie-­Luise, 164 Azoulay, Ariella, 64–65
anthropocentrism, 33, 116, 179, 182–83;
atomic bomb and, 202n80; climate Baichal, Jennifer, 123
change and, 6, 13, 121–23, 126, 128, 130–32, Bainimarama, Frank, 112
149; minimal ethics and, 188n39; naming Ballard, Su, 23
of the Anthropocene, 19, 121–23; trauma Barad, Karen, 27
theory and, 30 Behram, Noor, 50–­51, 193n31
Arab Spring, 59 Benjamin, Ruha, 102–3
ArcGIS, 100 Berlant, Lauren, 15, 158, 166–67, 171
Arendt, Hannah, 31 Biden, Joe, 198n65
Armenia–­Azerbaijan war (2020), 75 Bidjigal ­people of Eora Nation, 7
artificial intelligence (ai), 26, 38, 40, 75, biopolitics, 12, 16, 22, 29, 35; covid-19 pan-
110, 180, 199n82; Agile Condor target- demic and, 175–76. See also ontopower
ing system, 43, 49, 59, 66–74, 76; Azure Biopolitics of the More-­Than-­Human
platform, 80; ChatGPT, 82, 105–6; (Pugliese), 22
convolutional neural networks (cnns), black boxing, 68, 77, 82, 86, 104–5, 110, 156
88, 96, 100; Dall-­E 2, 83, 92; deepfakes Blaser, Mario, 7, 18, 146, 176
and, 88, 90–91, 197n39; FAccT confer- “Blue Marble,” 117
ence, 105; Figure Eight, 100; generative bnngina (slang for drones), 52
adversarial networks (gans), 88–90, 92, “Boiling Milk” (Halperin), 127–29, 145
197n40; Google Brain, 88; Google Ethical Borges, Jorge Luis: “On Exactitude in Sci-
ai team, 104, 199n78; ­human ­labor and, ence,” 80
76, 82, 85, 100–101; Mechanical Turk, 82; Boulminwi, Joy, 104
Midjourney, 83; Potemkin ai, 82; Proj­ect Bousquet, Antoine, 41, 45, 69, 73
Maven, 99–100, 103; Stable Diffusion, Brand, Stewart: ­Whole Earth Cata­logue, 118
92; Synthesis.ai, 93. See also machine Bridle, James: Drone Shadows, 193n43
learning Brimblecombe-­Fox, Kathryn: Theatre of
Asaro, Peter, 79 War: Photons Do Not Care, 34–35
Atacama salt flats, Chile, 123. See also Ata- British Royal Commission into Nuclear
can community of Chile Testing (1984), 133, 144

230 Index
Browne, Simone, 13 communicative politics: and nonhuman
Browning, Daniel, 139 witnessing, 3, 8, 20, 116, 175–83
Bucher, Taina, 83 Computer Learns Automation (Tan), 106–10
Burtynsky, Edward: “Clear Cut #3,” 123; computer vision, 39, 58, 77, 93, 99–100,
“Morenci Mine #2,” 124; “Salt Pan #18,” 108–9. See also drone vision; machine
123 vision; surveillance
Butler, Judith, 31 Conflict Urbanism proj­ect, 60–63, 65
convolutional neural networks (cnns), 88,
capitalism, 18, 25, 43, 114, 133, 148, 187n25, 96, 100
188n43; algorithms and, 79, 104, 152, 160; Cooke, Grayson: “Open Air,” 125–26
extractive industries and, 12, 31, 130–31, corporate vio­lence, 5, 20, 38, 87, 93, 119.
168–71; racial, 12–14, 19, 174–75, 181. See also state vio­lence; structural vio­lence
See also settler colonialism; slavery covid-19 pandemic (2020), 17, 26, 80, 158,
Caruth, Cathy, 147 170, 174–76
Center for Spatial Research, Columbia Creech Air Force Base (US), 1
University: Conflict Urbanism: Aleppo crisis epistemologies, 14–15
proj­ect, 60–61 crisis ordinariness, 15, 158–59
Chakrabarty, Dipesh, 121 CrowdFlower platform, 100
Chamayou, Grégoire, 185n1, 192n19 Crutzen, Paul, 19
Chandler, Kate, 192n14 Cubitt, Sean, 16–17, 39–40, 116–17, 157, 175,
ChatGPT (OpenAI), 82, 105–6 177, 181–82; The Practice of Light, 26, 28,
Chayka, Kyle, 81 189n74
Chen, Mel Y., 28, 190n82 cyborg tendency, 26, 63, 164
Chile, 12, 123
Chishty, Mahwish: Drone Art Paintings, 53; da Silva, Denise Ferreira, 31, 56
Drone Shadows, 53–­54, 193n43; Reaper, Daston, Lorraine, 24, 115
53–­54 data friction, 12, 71, 76, 200n23
Chouliaraki, Lilie, 22, 63 data justice, 104–5, 110. See also activism
Chow, Rey, 166 Davis, Heather, 19
Chris­tian­ity, 6, 24–25 Death Zephyr (Scarce), 139, 141
Clark, Timothy, 121, 124 deepfakes, 9, 25, 87–92, 102, 197n41; of
“Clear Cut #3” (Burtynsky), 123 Barack Obama, 197n39; ­political warfare
climate crisis, 11, 133; artwork about, 123–30; and, 198n43. See also synthetic media
climate trauma, 132; climatic media and, Defense Advanced Research Proj­ects
119; Marshall Islands and, 112–15, 148; Agency (darpa), 13, 47; Moving and
scale of, 13, 121–22, 125; settler colonial- Stationary Target Acquisition and Recog-
ism and, 14–15, 18–19, 110, 113, 131. See also nition (mstar), 69–70; offset program,
ecological trauma 75; role in internet creation, 118; Systems
closed world computation, 13, 156 of Neuromorphic Adaptive Plastic Scal-
Cockburn, Andrew, 185n1 able Electronics (SyNAPSE), 69
Cold War era, 113, 117–18; closed world de la Cadena, Marisol, 7, 18, 146, 176–77, 182
computation and, 13, 156 DeLanda, Manuel, 73–75
Cole, Samantha, 88, 197n41 Deleuze, Gilles, 6, 45, 74, 91, 145–49
Columbian Exchange (1610), 19 DeLoughrey, Elizabeth, 113–14
communicative aesthetics, 29 Dencik, Lina, 104
communicative commons, 115 de Pencier, Nicholas, 123

Index 231
disappearances. See radical absence edge computing, 29, 49, 68
disinformation, 21, 90. See also deepfakes Edney-­Browne, Alex, 52
Doane, Mary Ann, 164 Edwards, Paul N., 13, 156, 200n23
Dodson, Pat, 169 Ellis, John, 25, 189n69
Dorrian, Mark, 50, 193n33 empiricism, 7, 22, 76, 87; radical, 6, 17, 45,
Dowdell, Anthony, 160–61 117, 132
Drone Art Paintings (Chishty), 53 Enlightenment, the, 7, 23–24, 31, 177, 179, 181
drones and drone warfare, 8–9, 14–15, 20, Epic Games: Unreal Engine, 93, 95–96
99, 164, 178, 180; AeroVironment rq-11 Estes, Nick, 18
Raven, 44; Bayraktar tb2, 38; bnngina ethics, 28, 42, 59, 77, 102, 120; algorithms
(slang), 52; ­Castle Bravo, 137; Datta Khel and, 84–88, 91–92; of care, 129, 147; cloud
airstrike, 50, 101–2; gendering and, 50, ethics, 110; Google Ethical ai team,
56, 102–3, 185n1; Global Hawks, 38, 46; 104, 199n78; humanism and, 23, 32–33;
global increase in, 12; kill boxes and, memory and, 29; minimal, 188n39; video
44–45; life ­under, 32, 38, 46, 51–52, 55–58, editing and, 151
193n37, 194n51; Operation Kamikazi, 137; Eubanks, ­Virginia, 104
Predator, 1–3, 38, 44, 46, 53, 75, 193n43; Eye in the Sky, 58
racialization and, 50–51, 56, 102–3, 185n1,
193n43; Reaper, 37–­39, 44, 46, 53–­54, 58, FAccT conference, 105
66, 70, 75–76, 100, 193n43; search for Facebook (Meta), 75, 151, 155–56; bias and,
Malaysian Airlines Flight 370, 156–57; 82–83, 159; digital death and, 153, 160–62,
spatial dynamics and, 35, 37–38, 47, 50, 165–66, 204n34
52, 57, 59–65, 68, 192n19; temporality and, facial recognition, 11, 70, 72, 84; bias and,
44, 47–49, 59, 61–62, 64, 68, 78, 192n19; 104; deepfakes and, 88, 90–91
unmanning and, 44, 192n14; Uruzgan Farid, Idris, 56
airstrike, 1–3, 7, 67. See also aesthetic Farmer, Paul, 31, 191n99
interventions and artwork; drone vision; Farrier, David, 128
surveillance; violent mediations Federally Administered Tribal Areas
drone vision, 193n24, 193n33, 194n51; war- (fata), Pakistan, 55, 194n44
fare and, 46, 50, 57–59, 61, 65, 109, 194n58. feminist critique, 16, 24–25, 64–65, 117,
See also computer vision; machine vision; 189n66, 197n41
surveillance Finn, Ed, 83
First Nations ­peoples, 6–8, 20, 24, 158, 177,
Earth imaging, 118–21; “Blue Marble,” 117; 182; crisis epistemologies and, 14–15; min-
“Digital Earth Australia,” 125; “Earthrise,” ing sacred sites of, 167–71, 176; nuclear
117; Google Earth and, 80, 100, 196n2; weapons testing and, 9, 115; settler colo-
Landsat and, 61–62, 115, 125–26 nialism and, 17–18, 131, 136, 138, 188n60.
earwitnessing, 52 See also Aboriginal ­peoples; Indigenous
ecological trauma, 10, 30, 116–17, 132–33, ­peoples; individual nations and ­peoples
149; Aboriginal ­peoples and, 135, 142, Foley, James, 150–52, 166
144–45; aesthetic interventions and, 117, Forensic Architecture (Goldsmiths, Uni-
125–29, 138–45; First Nations ­peoples and, versity of London), 20, 87, 102, 178; ­Triple
115, 131; Pacific Islands and, 113–15, 148; Chaser, 21, 93–99, 109
radical absence and, 154, 158, 167, 170, 179; forensic ecol­ogy, 22, 56
wounding and, 145–48. See also trauma; Freedom of Information Act (US), 3
traumatic affect Frosh, Paul, 22, 25, 161

232 Index
Fuller, Matthew, 116, 120–21; Investigative Hellfire missiles, 2, 30, 50, 57, 66
Aesthetics, 21–23, 183 Hogan, Mél, 161
Furuhata, Yuriko, 13, 119 Holmqvist, Caroline, 75
Hoskins, Andrew, 151
Gabrys, Jennifer, 120, 156
Gadigal p­ eople of Eora Nation, 7 Illingworth, Shona, 151
Galison, Peter, 24 imperialism, 80, 156, 192n13; First Nations
Galtung, Johan, 191n99 and, 18, 113, 136, 144, 182; insurgent aes-
gaming. See video games thetics and, 53; vio­lence and, 12, 18, 24, 29,
Gebru, Timnit, 104, 199n78 44, 113, 115, 194n44
gender, 24, 53; deepfakes and, 90, 197n41; Indigenous ­peoples, 18, 23–24, 146–47,
drone systems and, 50, 56, 102–3, 185n1; 201n63; back-­to-­the-­land movement
sexual vio­lence and, 64–65, 187n32; tar- and, 200n21; genocide and, 11–12, 187n32;
geting of “military-­aged males,” 2, 58, 102 nuclear weapons testing and, 136, 144.
General Atomics, 68; Reaper drones, 37–39, See also Aboriginal ­peoples; First Nations
44, 46, 53, 58, 66, 70, 75–76, 100, 193n43 ­peoples; individual nations and ­peoples
generative adversarial networks (gans), Institute of Electrical and Electronics Engi-
88–90, 92, 197n40 neers (ieee), 69, 195n83
geographic information systems (gis), 100 insurgent aesthetics, 53. See also aesthetic
Gerrard, John: “The Farm,” 105 interventions and artwork
gilgamesh, 2, 55 Investigative Aesthetics (Fuller and
GitHub repository, 88, 90, 94 Weizman), 21–23, 183
Givoni, Michal, 27 invisual perception, 85, 101, 103, 108. See also
Glissant, Édouard, 110, 179–80 computer vision
Google (Alphabet), 2, 11, 13–14, 105, 198n65; Iraq, 42; Islamic State in Iraq and Syria
bias and, 83, 104; Brain, 88; digital death (isis) and, 59–60, 100, 103, 150, 152; US
and, 161; Earth, 80, 100, 105, 196n2; invasion of, 57
Ethical ai team, 104, 199n78; machine Iron Dome technology (Israel), 74
learning and, 75, 88, 196n5; PageRank, Islamic State in Iraq and Syria (isis), 59–60,
166; Proj­ect Maven, 99–100, 103 100, 103, 150, 152
Gregory, Derek, 46, 185n1
ground truth, 26, 56, 70, 91, 164; aftermath Jackson, Zakiyyah Iman, 27, 189n61
in Aleppo and, 61–62 James, C. L. R., 130
Grove, Jairus, 16, 42, 45, 79, 116, 187n25, James, William, 17, 45, 146, 160
196n106 Jetñil-­Kijiner, Kathy, 148; “Tell Them,”
Grusin, Richard, 27, 153 113–14
Guattari, Félix, 86, 120, 147, 197n30 Joint Standing Committee on Northern
Gulf War (1991), 46, 192n19 Australia, 168–70, 205n55
Juukan Gorge sacred sites, 167–71, 205n55
Halperin, Ilana: “Boiling Milk,” 127–29, 145
Halpern, Orit, 13, 26, 83, 189n76 Kainaki II Declaration, 112
Haraway, Donna, 24–25, 117, 189n66 Kanders, Warren B., 94
Harkin, Natalie, 144; “mine and refine this Kapadia, Ronak, 53
float . . . ,” 145 Kaplan, Caren, 20, 44, 50, 63, 192n14
Hartman, Saidiya, 31, 191n103 Kausar, Mohammed, 52
Hayes, Burchell, 168 Kelley, Robin D. G., 12

Index 233
Kember, Sarah, 16, 151 and, 174; opacity and, 179–80; racializa-
kill boxes, 44–45, 192n19 tion and, 31, 177, 180
Kittler, Friedrich, 189n74 Manjoo, Farhad, 80
Kurgan, Laura, 60 Manning, Erin, 164, 172
Mãori ­peoples, 20
Landsat (nasa), 61–62, 115, 126; “Digital Maralinga, Australia, 133–39, 141–42, 144,
Earth Australia,” 125 176
Lapoujade, David, 159–60 Marshall Islands, 113–14, 137, 148
law, Western, 21, 52, 76, 90, 104, 194n44; martial gaze, 41–42, 44, 56, 69, 103
Aboriginal ­peoples and, 135–37, 168–70; Massumi, Brian, 78, 159; on ontopower, 15,
extractive capitalism and, 31, 168–70; First 72, 87; theorizing affect, 7, 96, 132, 148,
Nations ­peoples and, 20, 24, 136, 169–71; 164–65, 167, 185n5
terra nullius doctrine, 136, 202n74; Material Witness (Schuppli), 21
witnessing in, 23, 42–43, 56–57, 98, 144, McCosker, Anthony, 44
178, 189n61 media studies, 16, 25; witnessing subfield,
Liu, Cixin: The Three-Body Prob­lem, 129–30 22–23
liveness, 25, 59, 166 mediation, 26, 98, 120, 175, 177; and absence,
“Living ­Under Drones” report (Stanford 151–67, 171–73; and deepfakes, 91; defini-
University and New York University), 56, tion, 16; and drones, 2–3, 22, 35, 39–79;
193n37 and ecol­ogy, 116–19, 125–28, 132; and ma-
Luciano, Dana, 28, 190n82 chinic affect, 86, 101, 114; and nonhuman
witnessing, 7, 28; and nuclear testing, 21,
machine learning, 68–69, 80–81, 83–84, 119, 114, 144–45; violent, 10, 22, 39–79, 101, 114,
196n5, 198n61; aesthetic interventions 119, 132, 146, 149, 178–79
and, 93–99, 105–9; deepfakes and, 88–89, media witnessing, 25–26, 91, 158, 189n69;
92, 102, 197n39; ­human ­labor and, 1–2, 12, definition, 22. See also social media;
77, 79, 82, 96, 100, 106–8. See also artificial violent mediations
intelligence (ai); machinic affect Menzies, Robert, 133, 201n63
machine vision, 26, 49, 73, 75, 164, 178. Meta. See Facebook (Meta)
See also computer vision; drone vision; Microsoft, 75; AirSim, 93; Bing, 105; Flight
surveillance Simulator, 80–81, 105, 109; Planetary
machinic affect, 10, 84, 101–3, 197n31; Computer, 119
Computer Learns Automation and, 109, military-­aged males, targeting of, 2, 58, 102
110; deepfakes and, 91; definition, 86–87; “mine and refine this float . . .” (Harkin), 145
ecological trauma and, 114–15, 149, 179, misinformation, 21, 88. See also deepfakes
181; invisual perception and, 108; radical Missile Park (Scarce), 139, 142–­43
absence and, 154, 158, 163–64, 166, 171; Mitchell, Margaret, 199n78
­Triple Chaser and, 94, 96, 98. See also Mittman, J. D., 135–36
affect; machine learning mnemotechnologies, 163. See also drones
Mackenzie, Adrian, 82, 84–85, 108 and drone warfare
Malaysian Airlines Flight 370 (mh370), Moore, Jessa, 160–62
154–57, 165 “Morenci Mine #2” (Burtynsky), 124
Man, the Witness, 6, 19, 23–25, 28–29, 32, Moreton-­Robinson, Aileen, 135, 168, 202n71
35–36, 185n4, 188n60; climate crisis and, Morgan, Nyarri, 135
115–17, 126–27, 149; concept of the World Morrison, Scott, 112–13
and, 176–77, 180–81; covid-19 pandemic Morton, Timothy, 132, 201n56

234 Index
mq-9 Reaper I–­III (Pailthorpe), 37–39 Pacific Islands Forum (2019), 112, 183
Multi-­Spectral Targeting System (msts), Packer, Jeremy, 12, 40–41, 74, 76
1–2 Paglen, Trevor, 105
Munster, Anna, 84–85, 108 Pailthorpe, Baden, 191n1; mq-9 Reaper I–­III,
Murphie, Andrew, 15, 159, 187n29 37–39
Pakistan, 42, 53, 194n51; Federally Adminis-
nasa: Earth Resources Technology Satellite, tered Tribal Areas (fata) of, 55, 194n44;
126; Landsat, 61–62, 115, 125–26 Waziristan, 50, 101–2
Nature, 116, 147, 177, 201n56 Parks, Lisa, 44, 47, 57
natureculture, 10, 117 Pasquale, Frank, 82
necropolitics, 12, 16, 29, 35, 68, 78, 175. Pax for Peace, 119
See also biopolitics; ontopower Pearl, Daniel, 152
neuromorphic computing, 69–­7 1 Peters, John Durham, 22, 25, 130, 187n29
9/11 attacks (September 11, 2001), 12, 25–26, Phan, Thao, 103
158; aftermath of, 46, 150, 152, 192n8 Pinchevski, Amit, 22, 25, 157, 159
Nixon, Rob, 131 pluralism, 146
Noble, Safiya, 83, 104 pluriversal politics, 3, 11, 19, 110, 153, 176–78,
nonhuman witnessing, definition, 3–11, 180, 182, 205n5
16–17, 23–29 Pohiva, Akilisi, 112
nonhuman witnessing, politics of, 175–84 Poitras, Laura: ­Triple Chaser, 21, 93–99,
nuclear colonialism, 19, 136–38, 142, 144–45, 109
202n75 police brutality, 30, 152, 158, 169, 175, 187n32
nuclear weapons testing, 67, 118, 202n79; “Polymorphism” (Tan), 105
artworks about, 21, 113–14, 138–45; in Pong, Beryl, 49
Australia, 9, 115, 133–39, 201n63, 202n72, porn studies, 90
202n78; on Marshall Islands, 113–14, 137, posthumanist theory, 27, 190n85
148; “Operation Antler,” 134; “Operation Proj­ect Maven (Google and US Department
Buffalo,” 134, 137–38; White Sands Missile of Defense), 99–100, 103
Testing Range (US), 202n80. See also proximal policy optimization (ppo), 107–8
nuclear colonialism Puar, Jasbir, 12, 205n60
Nvidia, 95 Pugliese, Joseph, 23, 43, 56, 94–95, 194n49;
Nyong’o, Tavia, 190n85 Biopolitics of the More-­Than-­Human, 22
Puutu Kunti Kurrama and Pinikura (pkkp)
Obama, Barack, 90, 150, 197n39 ­peoples, 167–71, 205n55
Oceti Sakowin ­peoples of Dakota, 18
Öhman, Carl J., 160 queer critique, 27–28, 104, 190n82, 205n60
Oliver, Kelly, 26–27, 32–33, 36, 101
ontopower, 29, 35, 68, 78–79, 87, 175; defini- race and racialization, 16, 31, 118, 180,
tion, 15–16, 72 190n85; Aboriginal ­peoples and, 145,
opacities, 86, 100, 175–76, 178–81, 183. 187n32; algorithmic bias and, 82–83,
See also black boxing 103–4, 110–11, 159; drone warfare and,
OpenAI, 107; ChatGPT, 82, 105–6; Two-­ 50–51, 56, 102–13, 185n1, 193n43; First
Minute Papers YouTube channel, 105–6 Nations p­ eoples and, 169, 177; insurgent
“Open Air” (Cooke and Walker), 125–26 aesthetics and, 53; racial capitalism, 12–14,
open-­source tools, 20, 60, 90, 93–94, 183, 19, 174–75, 181; surveillance logics and, 13,
200n27 44. See also settler colonialism; slavery

Index 235
radical absence, 10, 50, 153–54, 163–67, Shah, Nisha, 45
172–73, 179; digital death and, 153, 160–63, Shanahan, John, 100
165–66, 204n34, 204n39; execution of Sharkey, Noel, 76
James Foley and, 150–52; Malaysian Air- Sheikh, Shela, 28
lines Flight 370 and, 155–57; sacred sites Singh, Julietta, 190n82
and, 168–71; traumatic affect and, 157–60 slavery, 12–13, 23–24, 29, 31, 179–80, 191n103;
radioactivity, 137, 141–42, 176, 202n72, plantations and, 18–19, 130–31. See also
202nn79, 203n83 capitalism
Ramstein Air Base, Germany, 1 social media, 13, 17, 20, 98, 109, 111, 130, 154;
Rancière, Jacques, 176 bias and, 82–83, 159; digital death and, 153,
rape, 42, 64–65 160–62, 165–66, 204n34; disappearance
Ray, Una, 139, 142 of Malaysian Airlines Flight 370 and,
Reeves, Joshua, 12, 40–41, 74, 76 155–57; execution of James Foley and,
Rhee, Jennifer, 193n43 150–52; grief and, 203n34; mining sacred
Rio Tinto, 167–70 sites and, 168, 170–71, 176; traumatic
Rise of the Drones (pbs), 48 affect and, 159–60, 162–63, 166. See also
Rothe, Delf, 119–20 individual platforms
Russill, Chris, 17, 118, 187n29 Sotloff, Steven, 151
spatial dynamics, 16, 20, 25, 88, 93, 96, 114,
Sadowski, Jathan, 82 142; climate crisis and, 9, 118, 120–23,
Safariland, 94–96 126–32; drone warfare and, 35, 37–38,
“Salt Pan #18” (Burtynsky), 123 47, 50, 52, 57, 59–65, 68, 192n19; radical
savage ecol­ogy, 116, 120 absence and, 154, 157–59, 162–64, 172, 174
Scannell, Paddy, 156–57 Spiking Neural Networks, 69
Scarce, Yhonnie, 115, 138, 144–45; Death Spillers, Hortense, 24, 31
Zephyr, 139, 141; Missile Park, 139, 142–­43; src Inc., 66, 69–70. See also Agile Condor
Thunder Raining Poison, 139–­40 Stahl, Roger, 57
Schmidt, Eric, 14, 198n65 Starosielski, Nicole, 58
Schuppli, Susan, 22–23, 43, 52, 63, 171; state vio­lence, 5, 12–16, 20, 31–32, 79, 84, 119,
Material Witness, 21 178. See also drones and drone warfare;
Sear, Tom, 91 settler colonialism; structural vio­lence
Seaver, Nick, 83 Stewart, Kathleen, 19–20, 172
Seltzer, Mark, 158 Stiegler, Bernard, 163
September 11, 2001 (9/11), attacks, 12, 25–26, Stoermer, Eugene, 19
158; aftermath of, 46, 150, 152, 192n8 Stolen Generation, 187n32
settler colonialism, 11, 19–20, 29, 111, 117, structural vio­lence, 5, 14, 31, 61, 190n82,
121, 148, 185n4, 187n25, 188n43; author 191n99. See also settler colonialism;
positionality and, 7, 17–18; environmental slavery; state vio­lence
injustice and, 5, 12, 115, 131; genocide and, structures of feeling, 122, 172, 201n41
19, 169, 187n32; possessive logics of, 39, Stubblefield, Thomas, 50, 191n1, 193n33
135–36, 169–70, 202n74; racist other- Suchman, Lucy, 78
ing and, 24–25, 44, 82–83, 103, 190n82; surveillance, 44, 56, 155–56; Agile Con-
­resistance to, 13, 53, 110, 137–39, 142, 145, dor, 43, 49, 59, 66–74, 76; argus-­i s,
168–69, 172, 177, 202n75; structural nature 47–48, 100; climate-­monitoring systems,
of, 5, 14, 23, 31, 35, 113, 144, 169–71, 194n44 9, 13–14, 20, 113, 119–20, 126, 187n29,
sexual vio­lence, 42, 64–65, 187n32 200n23; covid-19 pandemic and, 174–75;

236 Index
drones and, 12, 32, 46, 51–52, 59, 194n58; trauma studies, 29–30, 32, 147–48
Earth imaging, 61–62, 80, 100, 115, 117–21, traumatic affect, 151–54, 156–60, 164, 166–67,
125–26, 196n2; gilgamesh, 2, 55; nuclear 171–73; digital death and, 162–63, 165.
monitoring systems, 67, 118; skynet, See also affect theory; trauma
55; wide area motion imagery (wami) ­Triple Chaser (Poitras and Forensic Archi-
initiatives, 41, 47–49, 71, 100. See also tecture), 21, 93–99, 109
computer vision; drone vision; machine Tsing, Anna, 130
vision Tynan, Lauren, 135, 147
synthetic media, 66, 88, 90–91, 93–99.
See also deepfakes Ukraine, 2022 ­Russian invasion of, 38, 75
Syria, 42; aftermath of war in Aleppo, Uliasz, Rebecca, 90
43, 59–65, 98, 119; Islamic State in Iraq uncommmons, 206n21
and Syria (isis) and, 59–60, 100, 103, United Nations Institute for Training and
150, 152 Research (unitar): unosat (United
Nations Satellite Center), 60
Taffel, Sy, 82 US Defense Innovation Board, 14, 198n65
Tahir, Madiha, 42, 194n44 US Department of Defense (DoD), 198n65;
Tan, Kynan: Computer Learns Automation, Defense Intelligence Enterprise, 99–100;
106–10; “Polymorphism,” 105 Proj­ect Maven, 99–100, 103; report on
Taussig, Michael, 191n102 Uruzgan airstrike, 2–3
­television, 25, 158, 189n69 US Special Forces, 2, 68
“Tell Them” (Jetñil-­Kijiner), 113–14
temporality, 16, 18, 87, 98, 147, 187n36; Valla, Clement: Postcards from Google
climate crisis and, 114, 117–23, 125–30, 132, Earth, 196n2
145, 149; drone warfare and, 44, 47–49, video games, 38, 70, 162–63; Microsoft
59, 61–62, 64, 68, 78, 192n19; machine Flight Simulator, 80–81, 105, 109;
learning and, 100, 105, 107–8; of nuclear Open­AI, 106; Second Life, 93; Unigene,
radiation, 116, 133, 141–42, 203n83; radical 105; Unreal Engine, 93, 95–96
absence and, 154–55, 161–64, 172; signifi- Vigh, Henrik, 15
cance to the book, 8–9 violent mediations, 10, 101–2, 178–79;
terra nullius, 18, 136 aftermath of drone warfare and, 2–3,
Theatre of War: Photons Do Not Care 22, 49–50, 52, 57–65; climate crisis and,
(Brimblecombe-­Fox), 34–35 114–16, 119, 132, 146, 149; drone systems
“The Farm” (Gerrard), 105 and, 39–49, 67, 72–73, 76–79, 84; radical
Thirft, Nigel, 163, 167 absence and, 151–54, 157–60, 172–73; rape
Three-Body Prob­lem, The (Liu), 129–30 and, 64–65
Thunder Raining Poison (Scarce), 139–­40 Virilio, Paul, 67
Todd, Zoe S., 19 von Neumann, John, 118
Tomkins, Silvan, 91
trans-­corporeality, 190n86 Walker, Emma: “Open Air,” 125–26
trauma, 9, 15, 59; mythologizing and, Wark, Scott, 103
205n60; used to radicalize, 151; witnessing war rugs, 53, 192n9
in response to, 4, 21, 29, 32–33, 180–82; Watson, David, 160
wounding and, 50, 145–48. See also eco- Waup, Lisa, 142
logical trauma; traumatic affect; trauma Waziristan, ­Afghanistan–­Pakistan border,
studies 50, 101–2

Index 237
Weizman, Eyal, 22, 32, 93, 95, 102, 119, 183; Woods, Derek, 122
Forensic Architecture, 21; Investigative Woomera, Australia, 138, 142, 202n78
Aesthetics, 21–23, 183. See also Forensic Work, Robert, 99
Architecture worldings, 19–20, 176–77
Weston, Kath, 174 wound culture, 158. See also trauma; violent
Whanganui River, New Zealand, 20 mediations
white settler sovereignty, 135, 137, 168–69, Wynter, Sylvia, 6, 31, 116, 118, 185n4
202n71. See also nuclear colonialism; set-
tler colonialism xenowar, 91. See also deepfakes
­Whole Earth Cata­log (Brand), 118
Whyte, Jess, 192n13 Yami, Lester, 133, 135, 139, 144
Whyte, Kyle Powis, 6, 14, 18, 131 YouTube, 58, 90, 156; aftermath in Aleppo
wide area motion imagery (wami) initia- and, 61, 63; Agile Condor videos, 71;
tives, 41, 47–49, 71, 100 execution of James Foley and, 150, 152;
Wilcox, Lauren, 103, 185n1 Two-­Minute Papers channel (OpenAI),
Wilken, Rowan, 44 105–6
Williams, Raymond, 201n41 Yusoff, Kathryn, 188n43
witness, 90–91
Wolf, Asher, 104 Zylinska, Joanna, 16, 28, 122, 128, 151; mini-
Wolfe, Patrick, 31, 187n32 mal ethics, 188n39

238 Index
This page intentionally left blank