Internet Architecture Board (IAB) H. Flanagan, Ed.
Request for Comments: 8700 RFC Editor
Updates: 2555, 5540 December 2019
Category: Informational
ISSN: 2070-1721
Fifty Years of RFCs
Abstract
This RFC marks the fiftieth anniversary for the RFC Series. It
includes both retrospective material from individuals involved at key
inflection points as well as a review of the current state of
affairs. It concludes with thoughts on possibilities for the next
fifty years for the Series. This document updates the perspectives
offered in RFCs 2555 and 5540.
Status of This Memo
This document is not an Internet Standards Track specification; it is
published for informational purposes.
This document is a product of the Internet Architecture Board (IAB)
and represents information that the IAB has deemed valuable to
provide for permanent record. It represents the consensus of the
Internet Architecture Board (IAB). Documents approved for
publication by the IAB are not candidates for any level of Internet
Standard; see Section 2 of RFC 7841.
Information about the current status of this document, any errata,
and how to provide feedback on it may be obtained at
https://www.rfc-editor.org/info/rfc8700.
Copyright Notice
Copyright (c) 2019 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents
(https://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with respect
to this document.
Table of Contents
1. Introduction
2. Key Moments in RFC History
3. Perspectives
3.1. The Origins of RFCs - by Stephen D. Crocker
3.2. The RFC Management and Editing Team - by Vint Cerf
3.3. Formalizing the RFC Editor Model - by Leslie Daigle
3.4. The Continuation, or Creation, of a Stream - by Nevil
Brownlee
3.5. A View from inside the RFC Editor - by Sandy Ginoza
4. The Next Fifty Years of RFCs
4.1. Preservation
4.2. Evolution of the RFC Format
4.3. Stream Structure
5. Conclusion
6. IANA Considerations
7. Security Considerations
8. Informative References
IAB Members at the Time of Approval
Acknowledgements
Contributors
Author's Address
1. Introduction
The RFC Series began in April 1969 with the publication of "Host
Software" by Steve Crocker. The early RFCs were, in fact, requests
for comments on ideas and proposals; the goal was to start
conversations rather than to create an archival record of a standard
or best practice. This goal changed over time, as the formality of
the publication process evolved and the community consuming the
material grew. Today, over 8500 RFCs have been published, ranging
across best practice guidance, experimental protocols, informational
material, and, of course, Internet standards. Material is accepted
for publication through the IETF, the IAB, the IRTF, and the
Independent Submissions streams, each of which have clear processes
on how drafts are submitted and potentially approved for publication
as an RFC. Ultimately, the goal of the RFC Series is to provide a
canonical source for the material published by the RFC Editor and to
support the preservation of that material in perpetuity.
The RFC Editor as a role came a few years after the first RFC was
published. The actual date the term "RFC Editor" was first used is
unknown, but it was formalized by [RFC0902] in July 1984; Jon Postel,
the first RFC Editor, defined the role by his actions and later by
defining the initial processes surrounding the publication of RFCs.
What is certain is that the goal of the RFC Editor is to produce
documents that are readable, clear, consistent, and reasonably
uniform, and that the archival record of what has been published is
maintained.
Change does come to the Series, albeit slowly. First, we saw the
distribution method change from postal mail to FTP and then to email.
RFCs could not be distributed electronically in the beginning, as the
means to do that distribution would not be defined until years after
the first RFC was "published". Not all early RFCs were even created
electronically; some were written out by hand or on a typewriter.
Eventually, the process for creating RFCs became more structured;
authors were provided guidance on how to write an RFC. The editorial
effort went from Steve Crocker to a more official model with a
designated editor, Jon Postel, and later to a team of five to seven
individuals. The actual editing and publishing work split from the
service for registration of protocol code points. The whole RFC
Editor structure was reviewed [RFC4844], refined [RFC5620], and
refined again [RFC6635]. And, in the last few years, the process to
change the format of the RFC documents themselves has started
[RFC7990].
This is evolution; and the Series will continue to be adapted in
order to meet the needs and expectations of the implementers,
operators, historians, and community of authors that uses the RFC
Series. These changes will always be balanced against the core
mission of the Series: to maintain a strong, stable, archival record
of technical specifications, protocols, and other information
relevant to the Advanced Research Projects Agency Network (ARPANET)
and Internet networking communities.
There is more to the history of the RFC Series than can be covered in
this document. Readers interested in earlier perspectives may find
the following RFCs of particular interest. These RFCs focus on the
enormous contributions of Jon Postel, Czar of Socket Numbers
[RFC0433] and first RFC Editor:
* [RFC2441]"Working with Jon, Tribute delivered at UCLA, October 30,
1998"
* [RFC2555]"30 Years of RFCs"
* [RFC5540]"40 Years of RFCs"
In this document, the history of the Series is viewed through the
eyes of several individuals who have been a part of shaping it.
Narratives of this nature offer a limited perspective on events;
there are almost certainly other viewpoints, memories, and
perspectives on events that are equally valid and would reflect a
different history. So, while these retrospectives are enormously
valuable and provide an insight to events of the day, they are just
one lens on the history of the RFC Series.
Steve Crocker, author of [RFC0001], offers his thoughts on how and
why the Series began. Leslie Daigle, a major influence in the
development of the RFC Editor model, offers her thoughts on the
change of the RFC Editor to a stronger, contracted function. Nevil
Brownlee, Independent Submissions Editor from 2010 through February
2018, shares his view on the clarification of the Independent Stream
(IS) and its transition upon the retirement of Bob Braden from the
position. As the current RFC Series Editor, I will put my thoughts
in on the most recent changes in formalizing the digital preservation
of the Series, the process to modernize the format while respecting
the need for stability, and my thoughts on the next fifty years of
RFCs.
This document updates the perspectives offered in [RFC2555] and
[RFC5540].
2. Key Moments in RFC History
+--------------------+----------+-----------------------------------+
| Marker | Date | Event |
+====================+==========+===================================+
| [RFC0001] | April | First RFC published |
| | 1969 | |
+--------------------+----------+-----------------------------------+
| [RFC0114] | April | First distribution of RFCs |
| | 1971 | over the network |
+--------------------+----------+-----------------------------------+
| [RFC0433] | December | First mention of the Czar of |
| | 1972 | Socket Numbers and the |
| | | proposal for a formal |
| | | registry |
+--------------------+----------+-----------------------------------+
| [RFC0690] | June | Relationship starts between |
| | 1975 | the Information Sciences |
| | | Institute (ISI) and the RFC |
| | | Editor (judging by Jon |
| | | Postel's affiliation change) |
+--------------------+----------+-----------------------------------+
| [RFC0748] | April | First April 1st RFC |
| | 1977 | published |
+--------------------+----------+-----------------------------------+
| [IETF1] | January | First Internet Engineering |
| | 1986 | Task Force (IETF) meeting |
+--------------------+----------+-----------------------------------+
| [IAB-19880712] | July | IAB approved the creation of |
| | 1988 | an Internet-Draft series |
+--------------------+----------+-----------------------------------+
| [RFC1122][RFC1123] | December | First major effort to review |
| | 1988 | key specifications and write |
| | | applicability statements |
+--------------------+----------+-----------------------------------+
| [RFC1083] | October | Three-stage standards |
| | 1989 | process first defined |
+--------------------+----------+-----------------------------------+
| [RFC1150] | March | FYI sub-series started |
| | 1990 | |
+--------------------+----------+-----------------------------------+
| [RFC1311] | March | STD sub-series started |
| | 1992 | |
+--------------------+----------+-----------------------------------+
| [RFC1818] | August | BCP sub-series started |
| | 1995 | |
+--------------------+----------+-----------------------------------+
| [RFC-ONLINE] | approx. | RFC Online Project to |
| | 1998 | restore early RFCs that were |
| | | "lost" started |
+--------------------+----------+-----------------------------------+
| [RFC2441] | 15 | Jon Postel's death |
| | October | |
| | 1998 | |
+--------------------+----------+-----------------------------------+
| [RFC4844] | July | RFC Series administrative |
| | 2007 | structure documented |
+--------------------+----------+-----------------------------------+
| [RFC4846] | July | Independent Submission |
| | 2007 | document stream is |
| | | formalized |
+--------------------+----------+-----------------------------------+
| [RFC5620] | August | RFC Editor organization |
| | 2009 | officially established as |
| | | RFC Series Editor, |
| | | Independent Submission |
| | | Editor, RFC Production |
| | | Center, and RFC Publisher |
+--------------------+----------+-----------------------------------+
| [ISI-to-AMS] | October | Transition of RFC Production |
| | 2009 | Center and RFC Publisher |
| | | starts from Information |
| | | Sciences Institute (ISI) to |
| | | Association Management |
| | | Solutions (AMS) |
+--------------------+----------+-----------------------------------+
| [RFC5540] | January | Bob Braden retires from RFC |
| | 2010 | Editor |
+--------------------+----------+-----------------------------------+
| [RFC5743] | December | Internet Research Task Force |
| | 2009 | document stream formalized |
+--------------------+----------+-----------------------------------+
| [RFC-ONLINE] | approx. | RFC Online Project to |
| | 2010 | restore early RFCs that were |
| | | "lost" finished |
+--------------------+----------+-----------------------------------+
| [RFC6360] | August | FYI sub-series ended |
| | 2011 | |
+--------------------+----------+-----------------------------------+
| [RFC6410] | October | Two-stage standards process |
| | 2011 | formalized |
+--------------------+----------+-----------------------------------+
| [RFC6635] | June | Updated responsibilities of |
| | 2012 | RFC Series allocated to RFC |
| | | Series Editor, RFC |
| | | Production Center, and RFC |
| | | Publisher |
+--------------------+----------+-----------------------------------+
| [RFC6949] | May 2013 | RFC format change project |
| | | started |
+--------------------+----------+-----------------------------------+
| [RFC8153] | April | RFCs no longer printed to |
| | 2017 | paper upon publication |
+--------------------+----------+-----------------------------------+
Table 1: Key Moments in RFC History
3. Perspectives
3.1. The Origins of RFCs - by Stephen D. Crocker
(This is a revision of material included more than 30 years ago in
[RFC1000].)
The Internet community now includes millions of nodes and billions of
users. It owes its beginning to the ARPANET, which was once but a
gleam in the eyes of J. C. R. Licklider, Bob Taylor, and Larry
Roberts of the Advanced Research Projects Agency (ARPA). While much
of the development proceeded according to plan, the initial design of
the protocols and the creation of the RFCs was largely accidental.
The procurement of the ARPANET was initiated in the summer of 1968;
remember Vietnam, flower children, etc.? There had been prior
experiments at various ARPA sites to link together computer systems,
but this was the first version to explore packet switching as a core
part of the communication strategy. ("ARPA" didn't become "DARPA"
(Defense Advanced Research Projects Agency) until 1972. It briefly
changed back to ARPA in 1993 and then back again to DARPA.) The
government's Request for Quotations (RFQ) called for four packet-
switching devices, called Interface Message Processors ("IMPs"), to
be delivered to four sites in the western part of the United States:
University of California, Los Angeles (UCLA); SRI International
(Stanford Research Institute) in Menlo Park, CA; University of
California, Santa Barbara (UCSB); and the University of Utah in Salt
Lake City. These sites were running a Scientific Data Systems (SDS)
Sigma 7, an SDS 940, an IBM 360/75, and a DEC PDP-10, respectively.
These machines not only had different operating systems, but even
details like character sets and byte sizes varied. Other sites would
have further variations.
The focus was on the basic movement of data. The precise use of the
ARPANET was not spelled out in advance, thus requiring the research
community to take some initiative. To stimulate this process, a
meeting was called in August 1968 with representatives from the
selected sites, chaired by Elmer Shapiro from SRI. Based on
Shapiro's notes from that meeting, the attendees were Dave Hopper and
Jeff Rulifson from SRI; Glen Culler and Gordon Buck from Santa
Barbara; R. Stephenson, C. Stephen Carr, and W. Boam from Utah;
Vint Cerf and me from UCLA; and a few others from potential future
sites.
That first meeting was seminal. We had lots of questions. How would
IMPs and "hosts" (I think that was the first time I was exposed to
that term) be connected? What would hosts say to each other? What
applications would be supported? The only concrete answers were
remote login as a replacement for dial-up, telephone-based
interactive terminal access, and file transfer, but we knew the
vision had to be larger. We found ourselves imagining all kinds of
possibilities: interactive graphics, cooperating processes, automatic
database query, electronic mail, etc., but no one knew where to
begin. We weren't sure whether there was really room to think hard
about these problems; surely someone senior and in charge, likely
from the East, would be along by and by to bring the word. But we
did come to one conclusion: we ought to meet again. Over the next
several months, we met at each of our sites, thereby setting the
precedent for regular face-to-face meetings. We also instantly felt
the irony. This new network was supposed to make it possible to work
together at a distance, and the first thing we did was schedule a
significant amount of travel.
Over the next several months, a small, fairly consistent set of
graduate students and staff members from the first four sites met.
We used the term Network Working Group (NWG) to designate ourselves.
This was the same term Elmer Shapiro had used when he convened our
first meeting, although it had been used until that point to refer to
the principal investigators and ARPA personnel: senior people who had
been planning the network. Our group was junior and disjointed from
the prior group, except, of course, that each of us worked for one of
the principal investigators.
The first few meetings were quite tenuous, primarily because we
weren't sure how narrow or expansive our goals should be. We had no
official charter or leadership, and it remained unclear, at least to
me, whether someone or some group would show up with the official
authority and responsibility to take over the problems we were
dealing with. Without clear definition of what the host-IMP
interface would look like, or even a precise definition of what
functions the IMP would provide, we focused on broader ideas. We
envisioned the possibility of application-specific protocols, with
code downloaded to user sites, and we took a crack at designing a
language to support this. The first version was known as DEL, for
"Decode-Encode Language" and a later version was called NIL, for
"Network Interchange Language".
In late 1968, Bolt Beranek and Newman (BBN) in Cambridge, MA won the
contract for the IMPs and began work in January 1969. A few of us
flew to Boston in the middle of February to meet the BBN crew. The
BBN folks, led by Frank Heart, included Bob Kahn, Severo Ornstein,
Ben Barker, Will Crowther, Bernie Cosell, and Dave Walden. They were
organized, professional, and focused. Their first concern was how to
meet their contract schedule of delivering the first IMP to UCLA at
the beginning of September and how to get bits to flow quickly and
reliably. The details of the host-IMP interface were not yet firm;
the specification came a few months later as BBN Report 1822. In
particular, BBN didn't take over our protocol design process, nor did
any other source of authority appear. Thus, we doggedly continued
debating and designing the protocols.
A month later, our small NWG met in Utah. As the meeting came toward
an end, it became clear to us that we should start writing down our
discussions. We had accumulated a few notes on the design of DEL and
other matters, and we decided to put them together in a set of notes.
We assigned writing chores to each of us, and I took on the
additional task of organizing the notes. Though I initiated the
RFCs, my role was far less than an editor. Each of the RFCs were
numbered in sequence. The only rule I imposed was the note had to be
complete before I assigned a number because I wanted to minimize the
number of holes in the sequence.
I tried a couple of times to write a note on how the notes would be
organized, but I found myself full of trepidation. Would these notes
look as if we were asserting authority we didn't have? Would we
unintentionally offend whomever the official protocol designers were?
Finally, unable to sleep, I wrote a few humble words. The basic
ground rules were that anyone could say anything and that nothing was
official. And to emphasize the point, I used Bill Duvall's
suggestion and labeled the notes "Request for Comments". I never
dreamed these notes would eventually be distributed through the very
medium we were discussing in these notes: talk about Sorcerer's
Apprentice! (See [APPRENTICE].)
After BBN distributed the specification for the IMP hardware and
software interface to the initial ARPANET sites, our attention
shifted to low-level matters. The ambitious ideas for automatic
downloading of code evaporated. It would be several years before
ideas like mobile code, remote procedure calls, ActiveX, JAVA, and
Representational State Transfer (RESTful) interfaces appeared.
Over the spring and summer of that year, we grappled with the
detailed problems of protocol design. Although we had a vision of
the vast potential for intercomputer communication, designing usable
protocols was another matter. We knew a custom hardware interface
and a custom software addition in the operating system was going to
be required for anything we designed, and we anticipated these would
pose some difficulty at each of the sites. We looked for existing
abstractions to use. It would have been convenient if we could have
made the network simply look like a regular device, e.g., a tape
drive, but we knew that wouldn't do. The essence of this network was
peer-to-peer cooperation among the machines and the processes running
inside them, not a central machine controlling dependent devices. We
settled on a virtual bit stream layer as the basic building block for
the protocols; but even back then, we knew that some applications
like voice might need to avoid that layer of software. (Why a
virtual bit stream instead of a virtual byte stream? Because each
computer had its own notion of how many bits were in a byte. 8-bit
bytes didn't become standard until a few years later.)
Over the next two years, we developed, exchanged, and implemented
ideas. I took a leave from UCLA in June 1971 to spend time working
at ARPA. Jon Postel took over the care and feeding of the RFCs,
evolving the process and adding collaborators over the next twenty-
seven years.
The rapid growth of the network and the working group also led to a
large pile of RFCs. When the 100th RFC was in sight, Peggy Karp at
the MIT Research Establishment (MITRE) took on the task of indexing
them. That seemed like a large task then, and we could have hardly
anticipated seeing more than 1000 RFCs several years later and the
evolution toward Internet-Drafts yet later.
When we first started working on the protocols, the network did not
exist. Except for our occasional face-to-face meetings, RFCs were
our only means of communication. In [RFC0003], I set the bar as low
as possible:
| The content of a NWG note may be any thought, suggestion, etc.
| related to the HOST software or other aspect of the network.
| Notes are encouraged to be timely rather than polished.
| Philosophical positions without examples or other specifics,
| specific suggestions or implementation techniques without
| introductory or background explication, and explicit questions
| without any attempted answers are all acceptable. The minimum
| length for a NWG note is one sentence.
|
| These standards (or lack of them) are stated explicitly for two
| reasons. First, there is a tendency to view a written
| statement as ipso facto authoritative, and we hope to promote
| the exchange and discussion of considerably less than
| authoritative ideas. Second, there is a natural hesitancy to
| publish something unpolished, and we hope to ease this
| inhibition.
Making the RFCs informal was not only a way of encouraging
participation; it was also important in making the communication
effective. One of the early participants said he was having trouble
writing and sending an RFC because his institution wanted to subject
them to publication review. These are not "publications", I
declared, and the problem went away. Another small detail, handled
instinctively and without debate, was the distribution model. Each
institution was required to send a copy directly to each of the other
handful of participating institutions. Each institution handled
internal copies and distribution itself. Submission to a central
point for redistribution was not required so as to minimize delays.
SRI's Network Information Center, however, did maintain a central
repository of everything and provided an invaluable record.
We didn't intentionally set out to challenge the existing standards
organizations, but our natural mode of operation yielded some
striking results. The RFCs are open in two important respects:
anyone can write one for free and anyone can get them for free. At
the time, virtually everyone in the ARPANET community was sponsored
by the government, so there was little competition and no need to use
documents as a way of raising money. Of course, as soon as we had
email working on the ARPANET, we distributed RFCs electronically.
When the ARPANET became just a portion of the Internet, this
distribution process became worldwide. The effect of this openness
is often overlooked; even now, students and young professionals all
over the world have been able to download RFCs, learn about the
technology within, and in turn, build the most amazing software.
(They are also a fantastic resource for historians.)
Where will it end? The ARPANET begat the Internet, and the
underlying technology transitioned from the original host-host
protocol to TCP/IP. But, the superstructure of protocol layers,
community-driven protocol design, and RFCs continued. Through the
changes in physical-layer technology, resulting in speed increases
from thousands to billions of bits per second, and similarly from
thousands to billions of users, this superstructure, including the
RFCs, has continued to serve the community. All of the computers
have changed, as have all of the transmission lines, but the RFCs
march on. Maybe I'll write a few words for RFC 10,000.
Quite obviously, the circumstances have changed. Email and other
media are most often used for the immediate exchange of inchoate
thoughts. Internet-Drafts are the means for exchanging substantial,
albeit sometimes speculative, content, while RFCs are reserved for
fully polished, reviewed, edited, and approved specifications.
Comments to RFCs are not requested, although usage-related
discussions and other commentary on mailing lists often take place
nonetheless. Rather than bemoan the change, I take it as a
remarkable example of adaptation. RFCs continue to serve the
protocol development community. Indeed, they are the bedrock of a
very vibrant and productive process that has fueled and guided the
Internet revolution.
3.2. The RFC Management and Editing Team - by Vint Cerf
As Steve Crocker mentions in Section 3.1, Jon Postel assumed the role
of RFC manager in 1971 when Steve left UCLA for ARPA. Jon took on
this role in addition to his subsequent "numbers Czar"
responsibilities. Initially, his focus was largely on assigning RFC
numbers to aspiring writers, but with time, and as the
standardization of the ARPANET and Internet protocols continued
apace, he began to serve in an editorial capacity. Moreover, as an
accomplished software engineer, he had opinions about technical
content in addition to writing style and did not hesitate to exercise
editorial discretion as would-be published authors presented their
offerings for his scrutiny. As the load increased, he recruited
additional "volunteer" talent, most notably Joyce K. Reynolds, a
fellow researcher at USC/ISI. Over the ensuing years, he also
drafted Robert (Bob) Braden onto the team, and when Jon unexpectedly
passed away in October 1998 (see [RFC2468]), Joyce and Bob undertook
carrying on with the RFC work in his stead, adding Sandy Ginoza to
the team. During the period when Jon and Joyce worked closely
together, Joyce would challenge me to tell which edits had been made
by Jon and which by her. I found this impossible, so aligned were
they in their editorial sensibilities. Sadly, three of these
tireless Internauts have passed on, and we have only the product of
their joint work and Sandy Ginoza's and others' corporate memory by
which to recall history.
3.3. Formalizing the RFC Editor Model - by Leslie Daigle
I was the chair of the Internet Architecture Board, the board
responsible for the general oversight of the RFC Series, at a
particular inflection point in the evolution of all Internet
technology institutions. To understand what we did, and why we had
to, let me first paint a broader picture of the arc of these
institutions.
Like many others who were in decision-making roles in the mid '00s, I
wasn't present when the Internet was born. The lore passed down to
me was that, out of the group of talented researchers that developed
the core specifications and established the direction of the
Internet, different individuals stepped up to take on roles necessary
to keep the process of specification development organized and open.
As the work of specification expanded, those individuals were
generally supported by organizations that carried on in the same
spirit. This was mostly Jon Postel, managing the allocation and
assignment of names and numbers, as well as working as the editor of
RFCs, but there were also individuals and institutions supporting the
IETF's Secretariat function. By the late 20th century, even this
model was wearing thin; the support functions were growing, and
organizations didn't have the ability to donate even more resources
to run them. In some cases (IANA), there was significant industry
and international dependence on the function and its neutrality.
The IETF, too, had grown in size, stature, and commercial reliance.
This system of institutional pieces "flying in formation" was not
providing the kind of contractual regularity or integrated
development that the IETF needed. People who hadn't been there as
the institutions developed, including IETF decision makers, didn't
innately understand why things "had to be the way they were" and were
frustrated when trying to get individual systems updated for new
requirements as well as better integrated across the spectrum of
activities.
Internet engineering had expanded beyond the point of being
supportable by a loosely coupled set of organizations of people who
had been there since the beginning and knew each other well. New
forms of governance were needed along with a rationalized funding
model. The IANA function was absorbed into a purpose-built
international not-for-profit organization. The IETF stepped up to
manage its own organizational destiny, creating the IETF
Administrative Support Activity (IASA), and the Secretariat became
one of its contracted functions.
This left the RFC Editor function as an independent effort supported
by the Internet Society.
That independent nature was necessary for the historic role of the
RFC Series in considering all technical contributions. But, at that
inflection point in the Series' history, it needed a new governance
and funding model, just as the other organizations supporting
Internet technical specification had. Also, the IETF leadership had
some concerns it felt needed to be addressed in its own technical
publication stream. While the RFC Series had been established before
there was an IETF, and had historically continued to have documents
in it that didn't originate from the IETF, the IETF was its largest
and most organized contributor. There was no particular organization
of independent contributors. Equally, the funding for the RFC Editor
was at that point coming from the Internet Society in the guise of
"support for the IETF". For people who hadn't been involved with the
institution from the outset, it was pretty easy to perceive the RFC
Series uniquely as the IETF's publication series. So, the challenge
was to identify and address the IETF's issues, along with governance
and funding, without sacrificing the fundamental nature of the RFC
Series as a broader-than-IETF publication series.
To give a sense of the kind of tensions that were prevalent, let me
share that the one phrase that stuck in my mind from those
discussions was "push to publish". There were those in IETF
leadership who felt that it would significantly reduce costs and
improve timeliness if an RFC could be published by, literally,
pushing a button on a web interface the moment it was approved by the
IESG. It would also, they argued, remove the specification issues
being introduced by copy editors that were hired as occasional
workers to help with improving publication rates but who weren't
necessarily up to speed on terms of art in technical specifications.
(There were some pretty egregious examples of copy editors
introducing changes that significantly changed the technical meaning
of the text that I forbear from citing here; let's just say it wasn't
strictly a problem of Internet engineers getting uptight about their
cheese being moved.) While "push to publish" would have addressed
those issues, it would not have addressed the loss of clarity from
the many significant text improvements copy editors successfully
introduced, or the fact that not all RFCs are approved by the IESG.
Institutionally, it was clear that the target was to have the RFC
Editor function governance within the reach of the Internet technical
community (as opposed to any particular private organization) without
tying it specifically to the IETF. That was reasonably achievable by
ensuring that the resultant pieces were established under the
oversight of the IAB (which is, itself, independent of the IETF even
as it is supported by the IASA organization).
The IETF worked on a document outlining functional requirements for
its technical specification publication. This could have been useful
for establishing its own series, but it also was helpful in
establishing awareness of the challenges in document publishing (it
always looks easy when you haven't thought about it) and also in
laying the groundwork for dialogue with the RFC Editor. The
requirements document was published as [RFC4714] as an Informational
RFC that stands today to provide guidance in the editing processes
surrounding IETF publications.
There was still, however, a certain lack of clarity about
responsibilities for making decisions and changes in the RFC Series
itself. To that end, I and the IAB worked with the various involved
parties to produce [RFC4844]. That document captured the RFC Series
mission (for a purpose greater than IETF technical specification
publication) as well as the roles and responsibilities of the parties
involved. The RFC Editor is responsible for ensuring the
implementation of the mission. The IAB continues to have oversight
responsibilities, including policy oversight, which it could act on
by changing the person (organization) in the role of RFC Editor. At
the same time, operational oversight was migrated to the IASA support
function of the IETF (and IAB).
The discussions, and the resulting publication of [RFC4844], allowed
greater visibility into and commitment to the RFC Series as a general
Internet publication. It also meant that subsequent adjustments
could be made as requirements evolved; the responsible parties are
clearly identified.
3.4. The Continuation, or Creation, of a Stream - by Nevil Brownlee
Arguably starting in 2006 with [RFC4714], the IAB and the IETF
community spent some time in the mid-'00s evolving the structure of
the RFC Series. This work included defining how those groups that
published into the RFC Series (initially including the IETF, the IAB
[RFC4845], and the Independent Submissions Stream [RFC4846], and
later growing to include the IRTF [RFC5743]) would handle approving
documents to be published as RFCs. In 2009, the IAB published "RFC
Editor Model (Version 1)" [RFC5620]. In this model, a new role was
created within the RFC Editor: the RFC Series Editor (RSE). This
individual would oversee RFC publishing and development while leaving
the process for approving documents for publication outside his or
her mandate. While, arguably, this was a role long filled by people
like Jon Postel, Bob Braden, and Joyce Reynolds, [RFC5620] saw the
role of RFC Series Editor defined in such a way as to distinctly
separate it from that of the Independent Submissions Editor (ISE).
Before 2009, the RFC Editor could accept "Independent" submissions
from individuals and, if they were judged significant, publish them
as RFCs; the Independent Stream was set up to continue that function.
From February 2010 through February 2018, I was the ISE. After
reading [RFC4846], I went on to develop the Independent Stream (IS).
First, I spent several days at the RFC Production Center at the
Information Sciences Institute (ISI) in Marina Del Ray with the RFC
Editor (Bob Braden), Sandy Ginoza, and Alice Hagens so as to learn
how RFCs were actually edited and published. All RFCs reach the
Production Center as Internet-Drafts; they are copy edited until the
edited version can be approved by its authors (AUTH48). At any
stage, authors can check their draft's status via the RFC Editor
website.
For the Independent Submissions, Bob kept a journal (a simple ASCII
file) of his interactions with authors for every draft, indexed by
the draft name. Bob also entered the Independent drafts into the RFC
Editor database so that authors could track their draft's status.
After my few days with his team at ISI, he handed that journal
(covering about 30 drafts) over to me and said, "Now it's over to
you!"
I began by following in Bob's footsteps, maintaining a journal and
tracking each draft's status in the RFC Editor database. My first
consideration was that every serious Internet-Draft submitted needed
several careful reviews. At that time, if the ISE knew of suitable
reviewers, he or she could simply ask them. Otherwise, if the draft
related to an IETF or IRTF Working Group, the ISE could ask Working
Group Chairs or Area Directors to suggest reviewers. The Independent
Submissions Editorial Board (Ed Board) was another place the ISE
could request reviewers from. My experience with reviewers was that
most of those I approached were happy to help.
Most drafts were straightforward, but there were some that needed
extra attention. Often, a draft requested IANA code points, and for
that, IANA was always quick to offer help and support. Code points
in some IANA registries require Expert Review [RFC8126]; sometimes
the interactions with Expert Reviewers took quite a long time!
Again, sometimes a draft seemed to fit better in the IETF Stream; for
these, I would suggest that the draft authors try to find an Area
Director to sponsor their work as an individual submission to the
IETF Stream.
After my first few years as ISE, the IETF Tools Team developed the
Datatracker [DATATRACKER] to show draft status and perform all the
"housekeeping" tasks for all of the streams. At that stage, I
switched to using the Datatracker rather than the RFC Editor
database.
Once a draft has been reviewed and the authors have revised it in
dialogue with their reviewers, the ISE must submit that draft to the
IESG for an "IESG Review" [RFC5742]. Overall, each IS draft
benefited from discussions (which were usually simple) with my Ed
Board and the IESG. A (very) few were somewhat controversial; for
those, I was able to work with the IESG to negotiate a suitable "IESG
Statement" and/or an "ISE Statement" to make it clear why the ISE
published the draft.
One rather special part of the Independent Stream is the April 1st
RFCs. These are humorous RFCs that have no formal review and
approval process. The authors must send them directly to the ISE or
the RFC Editor. Only a few of them can be published each year, and
each is reviewed by the ISE and the RSE. Bob Braden's criteria for
April 1st drafts were:
* They must relate to the Internet (like all drafts).
* Their readers should reach the end of page two before realizing it
is an April 1st RFC.
* They must actually be funny!
April 1st RFCs have a large following, and feedback from the Internet
community on April 1st of each year has been enthusiastic and quick!
159 RFCs were published in the Independent Stream during my eight
years as ISE. Over those eight years, I worked with most of their
authors and often met with them at IETF meetings. For me, that was a
very rewarding experience, so I thank all those that contributed.
During those eight years, I also worked with most of the IESG
members, who all also gave me a lot of helpful interaction. Lastly,
I've always enjoyed working with the RSE and all the staff of the RFC
Production Center. The IETF (as a whole) is very fortunate to have
such an effective team of talented professional staff.
3.5. A View from inside the RFC Editor - by Sandy Ginoza
When I joined ISI, shortly after Jon Postel passed away, the RFC
Editor model as we know it today (as defined in [RFC5620] and as
obsoleted by [RFC6548] and [RFC6635]) did not exist. The RFC Editor
functioned as one unit: there was no RSE, Production Center,
Publisher, or Independent Submissions Editor. All of these roles
were performed by the "RFC Editor", which was comprised of four
individuals: Bob Braden, Joyce Reynolds, a part-time student
programmer, and me.
Bob provided high-level guidance and reviewed Independent
Submissions. While Bob was a researcher in "Div 7" (Networking) at
ISI, ostensibly, the percentage of time he had for the RFC Editor was
10%, but he invested much more time to keep the Series running. He
pitched in where he could, especially when processing times were
getting longer; at one point, he even NROFFed a couple of RFCs-to-be.
Joyce was a full-time ISI employee. However, while continuing to
ensure RFCs were published, she was also serving as a User Services
Area Director and a keynote speaker about the Internet, and she was
also temporarily on loan to IANA for 50% of her time while IANA was
getting established after separating from ISI. The student
programmer performed programming tasks as requested and was, at the
time, responsible for parsing MIBs.
I was a full-time staffer and had to quickly learn the ropes so RFCs
would continue to be published. My primary tasks were to manage the
publication queue, format and prepare documents for Joyce's review,
carry out AUTH48 once Joyce completed her review, and publish, index,
and archive the RFCs (both soft and hard copies).
The workload increased significantly over the next few years. As the
workload increased, the RFC Editor reacted and slowly grew their
staff over time. To understand the team growth, let's first take a
look at the publication rates throughout history. The table below
shows average annual publication rates during 5-year periods.
+-------------+--------------------+
| Years | Avg. Pubs per Year |
+=============+====================+
| 1969 - 1972 | 80 |
+-------------+--------------------+
| 1973 - 1977 | 55 |
+-------------+--------------------+
| 1978 - 1982 | 20 |
+-------------+--------------------+
| 1983 - 1987 | 39 |
+-------------+--------------------+
| 1988 - 1992 | 69 |
+-------------+--------------------+
| 1993 - 1997 | 171 |
+-------------+--------------------+
| 1998 - 2002 | 237 |
+-------------+--------------------+
| 2003 - 2007 | 325 |
+-------------+--------------------+
| 2008 - 2012 | 333 |
+-------------+--------------------+
| 2013 - 2017 | 295 |
+-------------+--------------------+
Table 2: Annual Publication Rates
There were significant jumps in the publication rates in the '90s and
onward, with the number of publications almost doubling between 1993
and 2007. The annual submission count surpassed the 300 mark for the
first time in 2004 and reached an all-time high of 385 in 2011. The
submission rate did not drop below 300 until 2016 (284).
As the submissions grew, the RFC Editor experienced growing pains.
Processing times began to increase as the existing staff was unable
to keep up with the expanding queue size. In an attempt to reduce
the training hump and to avoid permanently hiring staff in case the
submission burst was a fluke, ISI brought on temporary copy editors;
this way, the staff could easily be resized as needed. However, as
Leslie noted, this didn't work very well. The effects of the
experiment would be lasting, as this led to a form of the process we
have now, where the RFC Editor asks more questions during AUTH/AUTH48
and technical changes require approval from the relevant Area
Directors or stream managers, depending on the document stream.
These changes added to the workload and extended publication times;
many often now jokingly refer to AUTH48 as the authors' "48 days",
"48 weeks", etc.
In addition to the increase in document submissions, we engaged in
tools testing and went through several editorial process changes.
Because of the lesson learned with temporary copy editors, our team
grew to be more permanent. While we added other editors in between,
two additions are of particular interest, as they experienced much of
the RFC Editor's growing pains, helped work us out of a backlogged
state, shaped the RFC Editor function, and are still with the team
today: Alice Russo joined the team in 2005 and Megan Ferguson joined
us in 2007.
With the understanding that the record-breaking number of submissions
was not an anomaly, we made significant upgrades to the
infrastructure of the RFC Editor function to facilitate document
tracking and reporting. For example, the illustrious "black binder"
(an actual 3-ring binder used to track RFC number assignment), a
manually edited HTML file for the queue page, and a Rube Goldberg set
of text files and scripts that created queue statistics, all were
eventually replaced; an errata system was proposed and implemented;
and XML became a newly accepted source file.
In 2009, [RFC5620] was published, introducing the initial version of
the RFC Editor model we have now. While it was published in 2009, it
did not go into effect until 2010, when the RFC Editor project as I
knew it was disbanded and divvied up into four pieces: RFC Series
Editor (RSE), Independent Submissions Editor (ISE), RFC Production
Center (RPC), and the Publisher function. In addition, the RFC
Series Advisory Group (RSAG) was created to "provide expert, informed
guidance (chiefly, to the RSE) in matters affecting the RFC Series
operation and development" [RSAG].
In 2010, the RPC and Publisher contracts were awarded to Association
Management Solutions (AMS). There, we started with three existing
team members (Alice Russo, Megan Ferguson, and me), and we were
pleased to be joined by Lynne Bartholomew and Rebecca VanRheenen, new
colleagues to anchor us in the AMS office.
I was wary of this model and was especially worried about the hole
Bob Braden's departure would create. Luckily for us, Bob Braden
provided wise counsel and insight during the transition (and beyond).
He gave the staff transitioning to AMS particularly helpful parting
words, "keep the RFCs coming", and that is what we did.
AMS embraced the RFC Series and helped us quickly get set up on new
servers. The RFC Production Center and Publisher were now part of
the AMS family and it was all hands on deck to make sure the
transition went smoothly to minimize the impact on document
processing.
Our focus during transition was to 1) keep the trains running; that
is, we wanted to get ourselves up and running with minimal down time,
and 2) work with the Transitional RSE (a role that concluded before
the transition ended), the ISE (Nevil Brownlee), RSAG, and the IETF
Administrative Director (IAD) to better understand and implement the
newly defined RFC Editor model.
Though some portions of the transition were challenging and lasted
longer than expected, the Acting RSE (Olaf Kolkman) officially handed
the reins over to the new RSE (Heather Flanagan) in 2012. She had to
jump in, learn the RFC Editor and IETF culture, and work through a
backlog of issues that had been left unattended.
Two of the backlogged issues were so old that they were ones someone
had asked me about at my first IETF meeting: When was the RFC Editor
going to allow non-ASCII characters in RFCs? When would the RFC
Editor adopt a more modern publication format?
At that time, while we understood the desire to move toward
supporting a broader range of character sets and having more-modern
outputs, we also routinely received emails from individuals
requesting that we send them plaintext files (instead of pointing
them to the website) because their Internet access was limited. We
also regularly received complaints from users of