I recently commented on a solution and pointed out a small error; the author soon commented back, thanked me and corrected the error. He/she was perfectly nice, and I'm not asking the question because of anything they said; but it made me wonder whether at that point, the proper thing to do would be to delete my comment. I realize there's likely no hard rule, but is there any generally accepted practice for this? I was leaning towards deletion, but thought I'd check first as I don't think I would be able to undelete after the fact. Thanks for any feedback.

  • 2
    $\begingroup$ Probably you can find some related older discussions here on meta, for example: Removing comments when they become irrelevant and Is deletion of unnecessary comments a common practice here? $\endgroup$ – Martin Sleziak Oct 25 '19 at 4:45
  • $\begingroup$ Thanks, I did check for similar questions but didn't run across those. $\endgroup$ – A.J. Oct 25 '19 at 5:03
  • 3
    $\begingroup$ For minor typos you can edit the solution yourself to make the fix. The answerer will be informed of the change and can undo it if it's not just a trivial correction. $\endgroup$ – Ethan Bolker Oct 26 '19 at 13:10
  • 2
    $\begingroup$ Yes. After the problem is fixed the comment no longer serves any usefuul purpose, in fact at best it's going to confuse readers... $\endgroup$ – David C. Ullrich Oct 27 '19 at 19:20
  • 1
    $\begingroup$ Thanks for asking this. I've wondered the same thing and will be more aggressive with my self clean-up. $\endgroup$ – JonathanZ supports MonicaC Oct 29 '19 at 23:57
  • $\begingroup$ If for no other reason, I delete my comment under those circumstances just because I dislike clutter. $\endgroup$ – Michael Hardy Nov 2 '19 at 18:13

Delete your comment, flag any associated comments for deletion.

A toy example:

User 1: "I think there is a mistake on the third paragraph. Not every $X$ is a $Y$!"

User 2: "Sorry, I don't understand. Isn't every $X$ a $Y$ when we are in a $Z$?

User 1: "No, that only holds for specific $Z$'s."

User 2: "Ok, thanks!"

All such comments should be deleted after the issues have been properly fixed. Actually, this is true for every comment: comments are requests for clarification and/or pointers for problems which should be deleted after the associated clarifications have been made, or problems have been solved. (There are some exceptions to this rule, but the major exception is meta itself. Comments are actually used for discussion here, perhaps due to the nature of meta itself. )

An analogy which I'm quite fond of is that comments are quite similar in functionality to "post-it notes". The idea is that comments are intended to help the answers and questions to become better. Comments should not be appendages of quality: whatever their content is, if it is useful, it should live in the question/answer. If it is not useful, it should live nowhere. Either way, the comment has no use after it has served its purpose of improving the question/answer, and should disappear to reduce clutter. (Exactly like a post-it note!)

  • $\begingroup$ Thanks for your reply, that confirms what my gut told me was the proper course of action. $\endgroup$ – A.J. Oct 25 '19 at 5:05
  • $\begingroup$ Thanks for clarifying Alozio. I always thought that users should not delete comments in this situation, because every time this happens to me, there's always that one user who refuses to delete their comments ;) $\endgroup$ – Toby Mak Oct 26 '19 at 10:42
  • 22
    $\begingroup$ "Actually, this is true for every comment" This may be standard SO dogma, but I doubt this applies to m.se to any significant extent. It just doesn't work. I tend to flag lots of comments for deletion, but there are so many more that do not deserve deletion. Most comments don't straightforwardly correct a simple typo. A paper isn't entirely made of theorems and proofs; a thread isn't entirely made of a question and answers. $\endgroup$ – darij grinberg Oct 26 '19 at 23:14
  • 1
    $\begingroup$ @darijgrinberg It is hard to reply to every point of your comment in comments satisfactorily, but it is not just "standard SO dogma". We deal with comments like that here, we receive a significant number of flags regarding those, it works reasonably well (the main constraint is sheer volume) and "deserve" is not the right word to have in mind here. $\endgroup$ – Aloizio Macedo Mod Oct 27 '19 at 5:43
  • 1
    $\begingroup$ The analogy at the end is flawed. It is not the purpose of papers to be entirely made of theorems and proofs. Comments are by design temporary, and lack functionality to be substantially effective as permanent objects. $\endgroup$ – Aloizio Macedo Mod Oct 27 '19 at 5:43
  • $\begingroup$ Furthermore, I mentioned that there are exceptions. Instead of theoretic situations, in practice comments fall predominantly (and overwhemingly) in those categories: pointers to issues, asking for clarifications (those two being their main objectives), displays of approval (like "+1! Very good!") and hint-like suggestions (which is where perhaps there is some controversy, but to flag them for deletion after a proper answer has been given is standard practice). All of which are eventually no longer needed. $\endgroup$ – Aloizio Macedo Mod Oct 27 '19 at 5:43
  • $\begingroup$ When the issues or inquiries are not handled simply, those comments begin turning into a discussion, and they are moved to chat. All this is procedure which occurs routinely. $\endgroup$ – Aloizio Macedo Mod Oct 27 '19 at 5:44
  • 4
    $\begingroup$ @AloizioMacedo: My impression is that a significant number of comments are to clarify insufficiently detailed or ambiguous answers. Example. While simple typos and minor omissions are a good case for editing an answer, these kind of clarifications and extensions are not. I don't know of a better medium for them than comments. $\endgroup$ – darij grinberg Oct 27 '19 at 6:34
  • 2
    $\begingroup$ @AloizioMacedo: There are also some kinds of answers that partly serve as a catcher for second opinions -- for example, an answer can recommend a book, and comments will then provide opinions and experiences with said book. At least currently, we do not tend to edit those comments and opinions into the answer itself, but they are often the more useful part of the whole package. $\endgroup$ – darij grinberg Oct 27 '19 at 6:37
  • $\begingroup$ @darijgrinberg Regarding the type of comments to "clarify insufficiently detailed or ambiguous answers", that is precisely "requests for clarifications/pointers for problems". The outcome should be that either the answerer edits their answer to clarify or, if they don't, you post another answer addressing the unattended issue(s). (Since they are not minor omissions or simple typos, a new answer is most likely a good idea.) $\endgroup$ – Aloizio Macedo Mod Oct 27 '19 at 16:10
  • 1
    $\begingroup$ Regarding the example of recommendations of books, I agree that this case is likely to be an exception. $\endgroup$ – Aloizio Macedo Mod Oct 27 '19 at 16:11
  • 5
    $\begingroup$ @AloizioMacedo: The strategy you are suggesting is rather unusual and will likely lead to lots of confused people downvoting the answer... $\endgroup$ – darij grinberg Oct 27 '19 at 17:21
  • 1
    $\begingroup$ @darijgrinberg It is not unusual and I don't see how it is confusing to have an independent answer which clarifies things which other answers don't. (Of course, if one makes an answer which reads as "In the other answer, one should make this instead of that", that is a bad answer. But this is not the point.) What leads to lots of people downvoting the answer is when someone adds an answer which adds absolutely no value to a question which is years old, which is also not what I mentioned. $\endgroup$ – Aloizio Macedo Mod Oct 27 '19 at 19:36
  • 1
    $\begingroup$ Or when an answer is inherently intended to be adversarial. (Like: "The other answer makes no sense! Disregard it! You should do X instead.") $\endgroup$ – Aloizio Macedo Mod Oct 27 '19 at 19:41
  • $\begingroup$ The post-it-note analogy is brilliant, and to the point. $\endgroup$ – Cyriac Antony Nov 3 '19 at 5:09

You must log in to answer this question.

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