Recently an answer of mine (https://mathoverflow.net/a/20820/402) was edited (or rather, a suggested edit was approved by two other users). The edit consisted exclusively of minor formatting changes, it mostly inserted dollar signs, e.g., replaced 1 with \$1\$ etc. It also introduced several formatting and typographical mistakes (such as using : instead of \colon, typesetting W*, L, and E in italic instead of roman font, using \Bbb R instead of \bf R, and including an additional sentence in a list item, the latter problem had to be fixed in an independent edit).

Since such edits unnecessarily bump the questions to the top of the front page while adding nothing to the answer, the current policy seems to discourage them, or at least this seems to be the case from reading the highest voted answer at What's our consensus on people resurrecting old questions just to edit formatting?.

What is the appropriate course of action in this case? Should I simply silently rollback to the latest correct version (revision 2 in this case), hoping that the users involved will eventually learn of the above policy?

Or should I also attempt to inform them of this policy? If so, what is the best way to do this? I certainly don't want to leave comments about this below the answer itself, because this would clutter the comment thread with completely irrelevant comments.

Also, would it be possible to codify the somewhat informal policy explained in the answer above? For example, add the relevant clarifications (e.g., specifically discouraging or even disallowing edits that only add \$ signs) to this page: https://mathoverflow.net/help/editing.

  • 2
    $\begingroup$ I think the "peer-review" mechanism for suggested edits in its present form is merely a safeguard against visibly malicious edits, and the edit to your post was certainly not "malicious" in this sense. $\endgroup$
    – Stefan Kohl Mod
    Nov 29, 2013 at 11:53
  • $\begingroup$ @StefanKohl: Your claim contradicts what is written at mathoverflow.net/help/editing: “If a user without edit privileges proposes an edit that does not comply with the guidelines above, it is ordinarily rejected in the review process.” Of course, it can still be the case that in practice the system works according to your description. $\endgroup$ Nov 29, 2013 at 12:00
  • 7
    $\begingroup$ In practice, the system works according to rather inconsistent standards and opinions of the various reviewers. I tend to reject edits such as this one, but clearly other reviewers think otherwise. $\endgroup$ Nov 29, 2013 at 12:51
  • 2
    $\begingroup$ @DmitriPavlov: Not really. -- The guidelines are pretty vague, and leave ample room for interpretation. I think what prevents the review from being a stronger quality control is mainly its bias towards approval -- one click is sufficient to vote to approve an edit, but voting to reject an edit requires specifying a particular reason. Approval is in a certain sense the default ... . $\endgroup$
    – Stefan Kohl Mod
    Nov 29, 2013 at 13:16
  • 3
    $\begingroup$ The answer you link to does not sugggest much anything as regards the general opinion, as the edit that actually says something specific arrived after most of the voting. The version that was voted on was sufficiently vague that most anybody could read in it what was their opinion, which is IMO why it is the highest voted one in the first place. You could check for instance my poll-answer there to see how mixed opinions are (as well as various other things that got written on the subject there and in other threads). $\endgroup$
    – user9072
    Nov 29, 2013 at 13:37
  • 1
    $\begingroup$ Other than that roll-back if you think it is worth the additional bump or leave it if not. And leave a comment for the editor if you want (you can @ comment-notify people that edited though they are not 'suggested'). I read you writing you do not want to do this, but if you are worried about the clutter you could later delete your comment. And phrase it in a way that stirs possibly discussion elsewhere, to meta or to chat. $\endgroup$
    – user9072
    Nov 29, 2013 at 13:39
  • 4
    $\begingroup$ Presumably adding dollar signs and like changes is not what made the resulting post "incorrect". Since it is already bumped, why not edit it again to make it "correct" but leave the formatting fixes? $\endgroup$ Nov 29, 2013 at 14:31
  • 1
    $\begingroup$ @GeraldEdgar why not? You can look at the post and all its revisions, by the way, you do not need to presume. Inserting dollars makes a normal W display in italic, which is then incorrect (in OP's opinion at least). Likewise one can consider R^2 as correct notation, while $\Bbb R^2$ is in some people's opinion incorrect. $\endgroup$
    – user9072
    Nov 29, 2013 at 14:44
  • $\begingroup$ @quid: I tend to distrust poll answers because they cancel positive and negative votes. I think many more people will upvote/downvote an answer they like if the current vote count is negative/positive as opposed to upvoting/downvoting an answer with a positive/negative vote count. So controversial poll answers tend to oscillate around zero. $\endgroup$ Nov 29, 2013 at 16:31
  • 5
    $\begingroup$ I do not claim this poll is very good, but still it should based on it and numerous other contribution on meta as well as from what happens in practice be clear to everybody paying some attention to the matter that there is contrary to what you seem to claim no consensus at all against (granted some will share your opinion, in fact me almost) the type of edit that happened to your post (that there was one glitch and some debatable things is a bit orthogonal). If you have an issue with it rollback or put it in the form you consider as optimal. But please stop ranting about it. $\endgroup$
    – user9072
    Nov 29, 2013 at 20:16
  • 1
    $\begingroup$ To put this differently I am virtually certain there are some that upvoted the answer you invoke as evidence but would actively welcome this edit (minus the minor error with the list) and approve it any time. $\endgroup$
    – user9072
    Nov 29, 2013 at 20:23
  • 9
    $\begingroup$ If you think an edit has introduced more problems than it was worth, and you suspect that the editor is doing this frequently, please flag for moderator attention. We can contact the editor directly if we think it's appropriate. $\endgroup$ Nov 29, 2013 at 20:48

2 Answers 2


Generally, editors should make any corrections that they can (or notice) before saving a post. If you're editing the post to make a minor typographical correction, or a minor formatting fix, it's very much appreciated if you also give the rest of the post a scan to see if anything else should be corrected. This has some advantages:

  • If you have the privilege to edit any post, it helps minimize the churn to the front page, as you noted. However, churn is also a very good thing because it gets more eyes on old content that might need some attention; so long as it doesn't knock more urgent questions out of view.

  • If you're suggesting an edit, you're optimizing for other people's time. Edits need to be reviewed, and ideally not repetitively. It's simply a matter of not creating unnecessary work for others.

There are some that maintain that any edit resulting in an improvement to the post is a good one, and they're not exactly wrong. Sometimes, the only thing that's wrong with a post is a minor formatting mistake. Most, however, would be rather irritated if I were to fix a minor markdown problem and leave glaring spelling and grammatical mistakes for someone else.

Personally, I won't resist a trivial edit if it's the only thing needed to curate the post, but the vast majority of my edits are substantial. If you come across someone making many of these extremely minor edits, you should definitely alert a moderator. Likewise, if you detect someone rubber-stamping really bad suggested edits, the moderators would probably also like to hear about it.

No edit should leave the post worse than it was, so feel free to reject or roll back anything that weakens voice or grammar, inserts unnecessary formatting making the post harder to read, alters the intent of the person writing it or otherwise brings down the level of quality.


This is why we need to enable users to mark their edits as "minor" (or some other term that implies it is not an occasion to bump the posts to the top). Any who disagrees would then have the ability to bump it to the top if they think that is appropriate.

Few TeX or LaTeX and MathJax users understand that those software packages were deliberately designed to make small aesthetic changes easy, or that a large number of little stylistic things that people usually don't notice are the cause of their thinking one font looks good and another looks terrible (which they do notice).

  • $\begingroup$ If a "minor" edit doesn't bump a post to the top, how will any who disagree even know that an edit has been made? $\endgroup$ Nov 6 at 20:26
  • 2
    $\begingroup$ @GerryMyerson : So we create a "recent edits" page and then those who care can look at it from time to time. $\endgroup$ Nov 6 at 20:34
  • 2
  • 2
    $\begingroup$ I'm confused as to how bumping something from just about a decade ago by posting a rant that we, the users of MathOverflow, have absolutely no control over, helps anyone. If you want to lobby for a feature, post a feature request. If you want to lobby for an SE-wide feature, do it on meta.SE. $\endgroup$
    – Asaf Karagila Mod
    Nov 7 at 8:52
  • 2
    $\begingroup$ The question actually gives a good reason why we shouldn't have such a feature: many, possibly most, people who make such minor formatting edits are actually quite clueless about basic standard practices of TeX formatting, and introduce more mistakes than they fix. (One reason for this is that new users are encouraged by the system to edit posts in order to gain reputation.) $\endgroup$ Nov 7 at 9:20
  • $\begingroup$ @EmilJeřábek : "many, possibly most, people who make such minor formatting edits are actually quite clueless about basic standard practices of TeX formatting, and introduce more mistakes than they fix." $${}$$ Do you have some empirical evidence for that? Or is it something that you think "would" happen, independently of empirical evidence? This seems like the opposite of what I've seen happening. $\endgroup$ Nov 7 at 15:15
  • $\begingroup$ The only evidence I have is my impression of seing such edits with regular frequency. $\endgroup$ Nov 7 at 17:21
  • $\begingroup$ @EmilJeřábek : I've been here almost daily for over a decade and I don't recall seeing this. But when I find badly done MathJax code and correct it, usually the user thanks me for having done so. $\endgroup$ Nov 7 at 17:25
  • 2
    $\begingroup$ I admit my perception of the relative frequency is likely skewed by the fact that bad edits of this kind annoy me, whereas the good ones are boring and do not elicit any emotional response. Nevertheless, such edits do occur often. As it happens, the one I’ve run across just now (cstheory.stackexchange.com/revisions/53457/2) was not on MO, but I don’t think the two sites are too different in this regard. $\endgroup$ Nov 8 at 11:07

You must log in to answer this question.

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