11

In mathematics, the timeline is typically as follows.

  1. Upload a completed draft of the paper to arXiv.
  2. Wait a week or so to see if anyone suggests changes, then submit to a journal.
  3. Once the paper is accepted or major revisions are made, update arXiv. The final arXiv version is the journal version but minus any journal-specific formatting.
  4. When the paper is actually published, add the "Journal ref" to the arXiv entry (this updates some data without creating a new version).

How is this done in CS when there is both a conference version and a journal version of the same paper, which nonetheless may be rather different? I know that the journal version is submitted after the conference version, but what is the timeline, and are the papers different versions of the same arXiv entry?

4
  • 12
    Step 0. Make sure that the journal you want to submit to permits author uploads to arXiv.
    – Buffy
    Nov 29 '20 at 21:00
  • 3
    Indeed, feel free to add that to the question if you'd like, though I'd say in math that step shouldn't be necessary.
    – Anonymous
    Nov 29 '20 at 21:03
  • 5
    Step 0.b if it is not allowed: decide to not publish in such a [expletive removed] journal. We are scientists. Science should be open and accessible. Do not contribute to hiding scientific results behind paywalls! Especially in a time where there really is no reason to anymore. There are always open ways of publishing available. And arXiv can even be your main and only publisher. This goes doubly so if you have tenure, but really in most fields great results will stand out no matter where they are published and things are shifting more and more towards "open science" not hurting your career.
    – Kvothe
    Nov 30 '20 at 16:41
  • 1
    I have submitted papers to arXiv, subsequently submitted them to journals which forbid arXiv, added a note to the submission that they had already been published on arXiv and this could not be revoked, and it didn't seem that anybody cared. I suppose they could subsequently "unpublish" it from their journal but legally copyright has been granted to arXiv first, and the journal has no right or ability to override that. (IANAL) Nov 30 '20 at 18:43
17

In theoretical computer science, a fairly common way to do it is roughly this:

  1. Prepare a nice, complete, readable version of the paper and upload it to arXiv. This is usually known as the "full version".

  2. Prepare a version suitable for conference submission (e.g. move some proofs to the appendix, save space here and there to meet the page limits, change layout if needed).

  3. Revise both versions based on the reviews & comments if needed, and repeat.

  4. Eventually the paper is accepted to a conference. Edit the accepted manuscript it so that it has the right length, the right layout, etc. Often there are page limits, and hence some details will be omitted. Refer to the "full version" in arXiv for the missing details. The publisher may or may not do additional edits before the paper appears in the conference. This is now known as the "conference version", and it is typically fairly different from the full version in arXiv.

For many papers the timeline ends here. However, some papers will also eventually appear in a journal, in which case we continue:

  1. Using both the full version and the latest conference version as a starting point, prepare a paper for journal submission. If there are relevant changes, update the arXiv version.

  2. Revise both versions based on the reviews & comments if needed, and repeat.

  3. Eventually the paper is accepted to a journal. The final published version may be very close to the arXiv version, or there may be changes in the journal version e.g. due to the reviewer's comments that are not incorporated in the arXiv version.

In brief, there is just one arXiv submission (with potentially many revisions), and it is usually much closer to the journal version (if one ever appears) than to the conference version (which tends to omit some details due to lack of space).

Publishers may also have restrictions on how much of their "added value" (e.g. changes made based on reviewers' feedback) can be included in the arXiv version, and because of this the arXiv version may also be an early draft.

Conferences play a big role, they have strict deadlines, and people organize their work based on the deadlines, so it is common to see papers first appearing in arXiv roughly at the same time as when they are first submitted to a conference (± a few days).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.