i'm skeptical about actually doing this, but @tantek @aaronpk and i had a brief IRC conversation about it, so i figured i'd capture that here.

snarfed
  tantek: not really, no
  publish retry feature request sounds good...but tricky in practice. lots of different kinds of errors, some transient, some not, and i'll never be able to distinguish 100% confidently
  (eg this one was an HTTP 400 "file type not recognized," and yet it's likely transient)
  also with transient errors, occasionally the publish inside the silo actually did still happen, so retrying would create a dupe
  still feel free to file though!
tantek
  unless you could detect the permalink there
  before the retry
  the silo permalink
snarfed
  i'd have to introspect silo posts and similarity compare
  sounds like a lot of work for an imperfect result :/
aaronpk
  when i use bridgy publish i also make use of the URL it returns, so it would be weird if it went and published the post later without my interaction
tantek
  aaronpk - indeed, perhaps the retrying should be programmatic
  rather than automatic