33
$\begingroup$

I have checked a random sample of questions in the main page, and it seems that almost every one of them has some form of answer in the comments, apart from completely unanswered/untouched ones.

This meta post mentions an old tea discussion (EDIT: added archive.org link, thanks to Martin Sleziak for suggesting it), and the author concludes that "[s]ome people sometimes prefer to answer in comments, and there is not much that can or even should be done against it". I'd like to revisit this, however, since the practice has reached endemic levels.

Just to summarize various drawbacks of answers in comments:

  • comments cannot be accepted. Questions answered only in a comment get periodically bumped up to the main page by bots, producing noise.
  • comments cannot be edited and improved/corrected.
  • the previous point applies also to automated edits to fix broken links.
  • comments cannot be searched using the search box.
  • comments cannot be downvoted if they are wrong; the only way to dispute the content of a comment is starting a long discussion in a comment thread.
  • comments-to-comments often produce long and messy threads, which could then be moved to chat, together with the initial answer-as-comment, and become even more difficult to locate.
  • comments are ephemeral and can be deleted by mods without trace under some conditions.
  • it is not clear which version of the CC license applies to comments.

There is apparently an unwritten tradition that answers that are very short, or just hints, should be written as comments (probably for fear of 'unwelcoming' downvotes?). My opinion is that this tradition needs to change, to avoid these drawbacks. Even bad and link-only answers should be answers, not comments.

What can we do, and what should we do, to address this phenomenon that is getting more and more widespread?

EDIT to clarify: I am speaking about questions that are on-topic for MO, not questions that are just about to be closed.

$\endgroup$
18
  • 1
    $\begingroup$ @MartinSleziak Yes, I am speaking about on-topic questions. Added an edit to specify this. $\endgroup$ Jan 20 at 10:41
  • 2
    $\begingroup$ I haven't myself any clue that it's increasingly more common, so I don't know if your sample is representative. As regards very recent question, it is not rare to first answer in comments: for instance if the answer is straightforward or in case of a doubt about the question, it is legitimate to first comment before posting a full answer. Another reason to postpone a full answer is when one doesn't have the time to immediately provide details. $\endgroup$
    – YCor
    Jan 20 at 13:18
  • 3
    $\begingroup$ @YCor Another reason to postpone a full answer is when one doesn't have the time to immediately provide details This seems a perfect example of what I think is a common misconception: an answer that is short, partial, or with little details should not become a comment. It's still an answer, just a short one. The minimum character length for answers is 30 characters, so unless your remark is very short there is no reason not to post it as one. :) $\endgroup$ Jan 20 at 15:33
  • 3
    $\begingroup$ @FedericoPoloni I should add that in my own case, probably in at lease 25% of such cases it appeared that either is misread the question, or my argument was incomplete (e.g., some fact I expected to be trivial or easy for me, wasn't). So I'm happy I didn't post it. (It also occurred for a few answers, but fortunately much less than 25%— I can't list my own deleted answer so I'm not sure how many.) To me, writing carefully an answer is full part of double checking it. $\endgroup$
    – YCor
    Jan 20 at 16:02
  • 6
    $\begingroup$ @FedericoPoloni I think most mathematicians prefer to be very careful when submitting answers. Sometimes I don't want to be pinned down to an answer I'm not absolutely sure of. If I later convince myself of correctness or fill in the details, I will post it as an answer, but only then. $\endgroup$ Jan 20 at 18:22
  • 1
    $\begingroup$ @HarryGindi thanks! actually Martin told me the same in chat. I have 21 answers deleted by myself (total 28 but I don't wanna include some that became deleted as the question was deleted), this is 4%. $\endgroup$
    – YCor
    Jan 20 at 18:23
  • 3
    $\begingroup$ There's something else I didn't mention. Sometimes I'm really unsure whether the question will be consider on-topic, and the answer might look fairly standard. Then I'm encouraged to post the answer if the question and the comment are appreciated. (Before answering it might be kept in mind that an upvoted answer prevents the question to be deleted.) $\endgroup$
    – YCor
    Jan 20 at 18:25
  • 3
    $\begingroup$ It's lower stakes and lower pressure. Comments can be deleted forever. $\endgroup$ Jan 20 at 20:02
  • 12
    $\begingroup$ Sometimes I answer a question in comments when I have something immediate to say which I think might be useful but without having the time or inclination to flesh out fully. Sometimes I might also feel that the question was not sufficiently interesting, or unsuitable in some way, but still just want to say something that the questioner could use. I don't really see what's wrong with this. People should be free to contribute productively in whatever way they are comfortable with. $\endgroup$
    – Lucia
    Jan 20 at 20:51
  • 27
    $\begingroup$ I knew people would answer this question in comments... $\endgroup$ Jan 20 at 21:53
  • 3
    $\begingroup$ @FedericoPoloni I initially started my answer as a comment before I tasted the irony. $\endgroup$
    – Alec Rhea
    Jan 21 at 1:27
  • 3
    $\begingroup$ My impression is that most people answer in comments not for the reasons you gave, but rather because: (1) the question wasn't appropriate for MO in the first place; (2) they feel a trivial observation doesn't deserve the upvotes an answer might attract. $\endgroup$ Jan 21 at 16:39
  • 3
    $\begingroup$ @ChristianRemling At least (2) should be discouraged since community wiki answers are a better solution. $\endgroup$ Jan 21 at 16:43
  • 2
    $\begingroup$ I once posted an answer to a soft question about the origin of some terminology and was told that it should have been broken up into two comments and was sinful (sic). So I broke it up into two comments, but then the original poster said that it was correct and should be converted into an answer. So now I just stick to comments as it seems safer. $\endgroup$
    – Flounderer
    Jan 23 at 11:47
  • 1
    $\begingroup$ This question prompted Yoav Kallus to CW-answer a question that was answered in the comments. $\endgroup$ Jan 28 at 0:48
20
$\begingroup$

I saw this just after posting a comment on a question that could be considered a partial answer, but I left it as a comment for two reasons: I am not sure it constitutes an answer since it does not address one aspect of the question (enriched categories), and Ivan asked for help understanding cutting edge techniques which I'm not sure Jacobs addresses.

I suspect a good amount of pseudoanswers that pop up as comments are for similar reasons, and if the OP finds the comment-answer suitable they can comment back something to the effect of

'that answers the question, could you post it as an answer to close this baby out?'

I have done this several times with good results, on my own questions and other peoples. If the comment isn't really addressing what the OP is asking about, I think it's better that they're left as clarifying comments.

I have seen some full answers as comments, and I generally use the same approach: reply with thanks, and ask the person to please post their comment as an answer to close out the question.


An edit with a concrete suggestion, in light of the discussion below. We could decide (informally) as a community that if the OP of a question has decided a comment is sufficiently meaty to constitute an answer, the first step is to comment back to that effect and request that the comment be converted into an answer.

After this step is taken, we could allow a ~3 month period in which the original comment-answerer is given prerogative to convert the comment into an answer and get some formalized street cred. If this period passes and no conversion takes place, we could decide that it becomes an implied editorial duty of the community to convert the comment-answer into a CW answer so the OP can close out the question.

$\endgroup$
11
  • 3
    $\begingroup$ In my experience, sometimes even prompting the user to leave the comment-answer as an answer has not yielded results, but yeah, this is a good thing to do before posting the answer yourself as CW. $\endgroup$ Jan 20 at 17:14
  • 6
    $\begingroup$ @HarryGindi There are a few users who have totally ignored my requests more than once (que sera sera), but outside these few I would saw half approximately half of people will post a comment as an answer if prompted within a day of prompting, and the ratio of conversion by answer-ers converges to approximately 3/4 people as more time is allowed. $\endgroup$
    – Alec Rhea
    Jan 21 at 1:24
  • $\begingroup$ Here's an example on one of my recent questions: mathoverflow.net/questions/375806/… $\endgroup$ Jan 21 at 11:55
  • 1
    $\begingroup$ @HarryGindi Here's an example from one of mine; there will naturally be some variation, but notice how there is no mention of what the person should do in my example, just thanks and a request that they convert the comment to an answer. For whatever reason, even if it's just a semantic difference the other person didn't notice, as soon as someone starts telling me what I should do I bristle and instinctually want to do the other thing. Not saying it's reasonable, just saying it exists. $\endgroup$
    – Alec Rhea
    Jan 22 at 23:38
  • $\begingroup$ I doubt that phrasing this another way would have induced Remy to leave an answer, but if one of his friends (or he himself) is reading this, please let me know! $\endgroup$ Jan 23 at 7:55
  • $\begingroup$ @HarryGindi In these situations, I think it is reasonable to write up the comment as a CW answer and accept it. $\endgroup$ Jan 27 at 18:08
  • $\begingroup$ @StevenGubkin That was my suggestion down below!! It didn't seem to be a consensus view so I haven't done anything about it yet. $\endgroup$ Jan 28 at 4:41
  • $\begingroup$ @HarryGindi I agree with Steven that if it's been several months with no reply, and the answerer has been active in the interim but not reposted the comment as an answer, posting their answer as community wiki is appropriate. Perhaps even more; we might decide (informally) as a community that a ~3 month comment-to-answer conversion period be left for the original comment-answerer, after which it is an implied editorial duty for the community to convert the comment to a CW answer. $\endgroup$
    – Alec Rhea
    Jan 28 at 11:37
  • $\begingroup$ 3 months seems a rather long time. With everything on a typical MO user's mind, I fear it would be rather easy to forget about this, leading to no practical change. Why not shorten the time period to a few days? If it happens that the original answerer posts something after the CW answer containing the comment has been made, it can always be deleted. $\endgroup$
    – KReiser
    Jan 29 at 0:50
  • $\begingroup$ @KReiser The idea isn't that a user will hold a comment in their head for three months before correcting it, but rather that if one comes across a comment that meets the above criteria but is less than 3 months old they forget about it, or if it's 3+ months old they immediately just copy-paste it as a CW answer. No need for memorization or rushing IMO. $\endgroup$
    – Alec Rhea
    Jan 29 at 4:39
  • $\begingroup$ Perhaps there was some miscommunication, but I still think that even with your re-explanation, we could do with a shorter time scale. Either way, thanks for your efforts to address this issue. $\endgroup$
    – KReiser
    Jan 29 at 4:52
17
$\begingroup$

Idea: If you see an answer offered in the comments and left there, copy it into a community wiki answer.

To answer Federico's comment: If there is an issue with licensing, rephrase the comment-answer in your own words with attribution. Copyright doesn't apply to ideas, only the actual text.

$\endgroup$
10
  • 1
    $\begingroup$ Note that there are edge cases in which this is illegal, I believe, because it changes the CC version that applies to its content. EDIT: ...and there was quite some backlash when SE tried to relicense stuff. $\endgroup$ Jan 20 at 8:48
  • $\begingroup$ @FedericoPoloni Nobody is going to sue over this. There's not really any way that it could be a matter of monetary damage anyway, so the worst thing that could possibly happen is we have to delete the answer or just rephrase it. You can't copyright the idea of an answer, just the literal text. $\endgroup$ Jan 20 at 8:50
  • 9
    $\begingroup$ Saying who wrote it and linking to their comment means you have satisfied the conditions (IANAL, maybe modulo minor formatting issues of referencing) of the Creative Commons Attribution-Share Alike license. Note that CC-By-SA has been the license for a while, just that the version was updated twice. CC themselves say the licenses v2.x and later are forward compatible $\endgroup$ Jan 20 at 9:14
  • 2
    $\begingroup$ @DavidRoberts IANAL either, but this highly upvoted meta.se post states (with explicit reference to the CC update guidelines) that it is not permitted to unilaterally relicense content from CC-BY-SA 3.0 to CC-BY-SA 4.0, though. Anyhow, these are just edge cases (copying a very old comment). $\endgroup$ Jan 20 at 10:44
  • 5
    $\begingroup$ @FedericoPoloni it's not about re-licensing old content here, but the license one can put on the copied work, since it is meant to be "Share-Alike". CC seem to claim that you don't need to use the exact same version of the license, but can use an updated version of the CC-By-SA license, as long as the original was 2.x or above. The controversy was that SE was re-licensing the original works. $\endgroup$ Jan 20 at 10:55
  • $\begingroup$ Thanks @DavidRoberts; this makes sense. Anyhow, what I don't like about this proposed solution is that it requires additional work from the community (copypasting comments with links, formatting and formulas into an answer is not immediate) and normalizes the misuse of the comment system without attempting to change it. It is like saying "if you see someone littering our common house, you can pick up their trash yourself and put it in the trashcan", without addressing the problem of people that litter in the first place. $\endgroup$ Jan 20 at 15:43
  • 2
    $\begingroup$ @FedericoPoloni Short of coercive action by the moderators (which I certainly don't want), I don't see a way out of this. There are some users who make extremely helpful and useful contributions to this site who, frankly, flout the rules about answering questions in comments. If the choice is between cleaning up after these people or threatening them with suspensions, I prefer that we just clean up after them. It's not that much work. The disincentive to this behavior is actually the reputation system, and it works probably 95% of the time. $\endgroup$ Jan 20 at 18:15
  • 1
    $\begingroup$ @HarryGindi Continuing the analogy, when we see someone littering we don't normally call the police for 'coercive action', but we can still do something: point out their wrong behaviour and let peer pressure work against them, for instance. (Incidentally, are there still users that answer only in comments as a personal rule? The thread linked above mentions one, but he has not been active here since ten years, it seems). $\endgroup$ Jan 20 at 19:21
  • 2
    $\begingroup$ he has not been active here since ten years, it seems ..... wellllll. ;-) $\endgroup$ Jan 20 at 22:06
  • 6
    $\begingroup$ @DavidRoberts I'm sure there's no way anyone who seemingly left the site could have returned to MO using a series of pseudonymous accounts, but, if they did, they might choose to sometimes answer in the comments and sometimes post a regular answer, making them less of a relevant example. $\endgroup$
    – Will Sawin
    Jan 21 at 13:28
13
$\begingroup$

I think one thing that sometimes causes me to answer in a comment is if I suspect that the question, as stated, is not the one the asker meant to ask, and that seeing the answer to the question-as-stated may cause them to edit the question so that it asks something different. I don't like it when I post an answer to a question but then the question gets edited into a different question; it leaves my answer hanging around embarrassingly answering the "wrong question", and I don't believe I even get a notification that a question I posted an answer to has been edited, to give me a chance to edit my answer with a corresponding disclaimer. I think it would be better if edits to questions didn't ever change them into entirely different questions (better to delete the original question, if it was really that bad, and ask a new one), but there's probably not anything we can do about that, and the possibility is one reason for posting answers as comments at first.

$\endgroup$
1
  • 1
    $\begingroup$ Note: if you click "Follow" under the question, you get notified of any edit to it. $\endgroup$ Feb 27 at 17:23
1
$\begingroup$

Rather than saying users are operating the site software the wrong way, I'd rather say the software doesn't properly reflect the way humans want to interact with the site. The solution is fix the software.

$\endgroup$
6
  • $\begingroup$ Thanks for the input. Any suggestions on how you would like to fix the software? I think also the SE folks would be happy to have a method to make the UI more intuitive, too. $\endgroup$ Apr 21 at 6:04
  • 1
    $\begingroup$ One way to start is fix the issues mentioned in the main post. Include comments in the search engine input, make it possible to accept a comment as an answer, turn off the bots bumping the posts, etc. Once those things are out of the way, we can see where we are and whether more needs to be done. $\endgroup$
    – none
    Apr 23 at 6:10
  • $\begingroup$ Comments being 'second-class citizens' is a deliberate design decision from Stack Exchange; if I understand correctly, you wish to revert it. This would probably face a lot of opposition in meta.se, and if it were applied it would produce a completely different system. If you want a forum rather than a Q&A site, then this probably isn't the right place. $\endgroup$ Apr 23 at 6:21
  • 2
    $\begingroup$ @Federico I could imagine at the least a flag that's functionally the reverse of the "this is not an answer" one, and mod tool that allows a comment to be converted to an answer, the reverse of the one now existing. $\endgroup$ Apr 24 at 13:58
  • 1
    $\begingroup$ @theHigherGeometer I agree with you, but note that this is not going to solve the problem by itself; just mitigate its consequences and create more work for the moderators. $\endgroup$ Apr 24 at 14:08
  • $\begingroup$ Yeah, I was trying to put the most positive spin on the suggestion by none, but it would be a big change to the software $\endgroup$ Apr 24 at 22:20

You must log in to answer this question.

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