A co-author and I completed a paper and posted it on ArXiV. Since then, the paper has been rejected many times. The paper has a few good ideas in it, but they were not put in context nor executed well. We have decided it is not worth submitting in its current form, and my co-author has repeatedly made clear they do not want to work on the paper any longer.

However, I would like to re-do the paper nearly from scratch and take it in a totally different direction that I believe is much more interesting. The parts of the paper I am interesting in bringing to the new paper are parts I did alone in the original paper, and my co-author agrees they did not contribute to these parts. But if the old paper remains on ArXiV as a separate paper, the new paper will have too many similarities.

So my question is: In this situation, could I replace the old arxiv paper with the new one and remove my co-author as an author (they are fine with this)? Is this something that people would find strange? It will be clear that the papers are very different.

  • 1
    I haven't been in your situation, but I think it'd be better to make a new posting for the new paper, and use the comments field to explain the relation between the two postings.
    – Anyon
    Commented Jul 21, 2020 at 2:56
  • You cannot completely remove things from ArXiv. Commented Jul 21, 2020 at 3:05
  • 4
    I did something like this with one of my own preprints, arxiv.org/abs/1607.06430. I decided to split it into multiple papers, the first of which has been posted as arxiv.org/abs/1807.02613. In the comments on the later paper, I wrote "The results in this paper replace Section 2-4 of arXiv:1607.06430" and I had no trouble (I think there must be plenty of text overlap between the two postings).
    – Dan Ramras
    Commented Jul 21, 2020 at 4:32

1 Answer 1


No, you shouldn't replace the old paper with the new one. Replacements are supposed to be revisions, not "very different" papers. The replacement that you propose would not only revise your contributions; it also looks like an attempt to erase your coauthors' contributions. The change in authors would almost certainly raise a red flag with the moderators (through arXiv's auto-moderation software), and I expect most moderators would reject the replacement.

Submitting a new paper whose text significantly overlaps an existing arXiv preprint will also raise a red flag with the moderators, but as Dan Ramras suggests in a comment, explicitly acknowledging the overlap will convince most moderators to accept it anyway.

(I was an arXiv moderator for over a decade.)

  • Thank you for answering, I will follow your advice because I don't want to appear to be erasing their contributions. I actually don't expect the text to have significant overlap. It's just that one algorithm (of several) and its analysis is similar between the two papers, and I think if you viewed the papers as totally distinct you would say that the algorithms are probably too similar. Commented Jul 21, 2020 at 15:31
  • As another question, do you think there is anything wrong with taking something out of a paper this way? I put a lot of time and effort into the one result, and I would hate to not develop it into a decent publication since it has a lot of potential. Commented Jul 21, 2020 at 15:35
  • No, I don't think there's anything wrong with extracting a subset of results from a formally unpublished paper and publishing that subset separately, as long as everyone involved with the original paper agrees. But even if you're not copying test verbatim, I strongly recommend citing the earlier arXiv paper.
    – JeffE
    Commented Jul 24, 2020 at 6:27

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .