I have just been presented with an answer in the VLQ queue that consisted essentially in a link to a preprint on arXiv. The user having posted it is a high reputation user, so he definitely knows the rules about link-only answers. Furthermore, he was also the one who had asked the question, and the author of the preprint. He is a serious mathematician, and the preprint is obviously of high level (I cannot judge its correctness, though). This put me in an uncomfortable position, and I recommended the deletion of his post, but only half-heartedly. Which leads me to ask:

should we relax the requirement about link-only posts, and accept links to arXiv, HAL and possibly other repositories that the MO community considers stable (i.e. they offer stable URLs that are not susceptible of change)?

If so, we shall need some sort of poll in order for us to vote on the list of such repositories.

(I, for one, am in favour of my own proposal :) but I would like to hear the opinion of the community.)

  • $\begingroup$ For context, I think you're talking about this answer? $\endgroup$
    – Tim Campion Mod
    Sep 4 at 18:43
  • 4
    $\begingroup$ I'm actually having difficulty finding any kind of meta post or anything where we explicitly say that link-only answers are not allowed. There is some discussion of this on Meta Stack Exchange (though I'm sure the discussion at that link doesn't exhaust the issue). I think I agree that the stability of the link is relevant to how acceptable a link-only answer might be, but probably it's not the only relevant factor. $\endgroup$
    – Tim Campion Mod
    Sep 4 at 18:46
  • $\begingroup$ @TimCampion: Yes, that is the answer that I had in mind; in the meantime it got heavily edited. Indeed, there is no discussion where link-only answers are explicitly forbidden, but when one votes to delete an answer in the review queue one of the "canned" reasons offered by the software is that it is a link-only answer, and links may become invalid over time. This is not a MO-specific "canned comment", it is SE-wide. And I was thinking about triggering a discussion and maybe explicitly delimit what is an acceptable link-only answer from what is not. $\endgroup$
    – Alex M.
    Sep 4 at 20:57
  • 1
    $\begingroup$ Can you define allow, or replace it with something different? I'm fine with suggesting that link-only answers should not get deleted, but in my view it's entirely fair if they get downvoted, edited, or if they attract comments pointing out that they are link-only and suggesting improvements. $\endgroup$ Sep 8 at 6:45
  • $\begingroup$ @FedericoPoloni: Currently, link-only answers are supposed to get deleted, because URLs often become invalid over time. I was hoping to change this regarding a small set of well curated URLs, i.e. the answers containing them should no longer get deleted, as an official policy, but it is clear that this will not happen. Currently, the question has 79 views but only 3 votes and the answer has gathered only 6 votes. And almost no discussion in the comments. It is clear that it did not raise the interest of the community. $\endgroup$
    – Alex M.
    Sep 8 at 7:16
  • $\begingroup$ @AlexM. Are you sure there is an official policy to this effect? Could you add a link? And please edit your question to define "allow" there, not in a comment. $\endgroup$ Sep 8 at 7:26
  • 2
    $\begingroup$ @FedericoPoloni I have posted in MathOverflow chatroom what I was able to find. Basically some posts on Meta Stack Exchange, a brief mention in the help-center. (And there was a rather brief discussion on tea.) Of course, it's not clear whether or not it should be called "an official policy". (Is it actually something that could be considered, in practice, an official policy here on MO? This site is a bit different from other sites in the network.) $\endgroup$ Sep 11 at 4:15
  • 1
    $\begingroup$ @MartinSleziak Thanks Martin. Just to remove a layer of indirection from your "link-only comment", I am copying the pointers here: mathoverflow.net/help/how-to-answer , meta.stackexchange.com/questions/8231/… , msleziak.com/mathoverflow/tea/discussion/1426/… , meta.stackexchange.com/questions/225370/… . $\endgroup$ Sep 11 at 6:40

2 Answers 2


I propose to adopt this policy: link-only answers should not be deleted, as a general practice. They are at least a partial attempt to answer, and contain useful information that should not go lost. Note that the help center does not suggests deleting them.

Downvoting them, pointing out the problem in comments, and editing them to improve them are allowed and encouraged, instead.

  • $\begingroup$ Indeed. Others can edit the question to add some details from the linked site. $\endgroup$ Sep 11 at 13:15
  • 4
    $\begingroup$ The very least, proper bibliographic data should be added, if it's a link to the arXiv, HAL, etc, or to a published paper: at absolute minimum authors and title, enough so that the paper can be identified without the link. $\endgroup$
    – David Roberts Mod
    Sep 12 at 6:00

Upvote this answer if you agree with the proposal and you prefer to relax the rule.

Downvote if you disagree and you prefer to keep the same strict rule as it is today.

Feel free to edit this answer to add more repositories. Once we have a satisfactory list of candidates, a poll should be organized with all the proposed names, to see which ones gather the votes of the majority.



Project Euclid



  • $\begingroup$ I cannot vote on my own answer, but please add a +1 vote from me. $\endgroup$
    – Alex M.
    Sep 4 at 15:14

You must log in to answer this question.

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