Browsing TeX.SE, I have noticed that very frequently people provide answers to a question in the question's comments.

See for example

BibTeX error "inputenc Error: Unicode char \u8: not set up for use with LaTeX"

Is there any package for LaTeX dealing with a database?

How to correctly set up Palatino font with math (related to pxfonts)

This is very different from other SE sites where people tend to provide a lot of answers, and the best one(s) appear at the top by voting.

I am asking because for several unanswered questions I would like to provide an answer, but then I see the very same answer has been submitted as a comment, so I think "ah well, this would look like cheap copying, I rather leave it".

Effectively this seems to leave questions technically unanswered where in reality a pretty satisfying answer exists. To me, this habit seems contradictory to a Q&A site.

  • 14
    The three questions you link to are in fact good examples of the commenting system working as it is intended to. In the first, the OP effectively changed his mode of working based on the comments and so the question is really too localized. In the second, there are very clear duplicates linked in the comments but not enough feedback from the OP to close the question, unfortunately. In the third, the comments help to clarify the question (although without resolution.) Adding answers in any of these cases wouldn't be very productive, I think.
    – Alan Munn
    Feb 5, 2013 at 15:34
  • 59
    there are already more than enough really good answers here, so i will just leave a comment. i now have some experience with a couple of other sites, and find this one to be (1) the most productive in terms of good answers (and questions), (2) the least "greedy" in terms of reputation hunting, and (3) the most polite. i was somewhat turned off at first by the concept of reputation, but just about everyone active here is a good citizen, and respects the other participants. i believe that it's a model of the genre, and am proud to be a member of such a great community. Feb 5, 2013 at 16:35
  • 1
    Related: Why do some people answer in comments?
    – Werner Mod
    Mar 9, 2014 at 22:17
  • 2
    Relevant: What if a comment answers a question?
    – Werner Mod
    Mar 9, 2014 at 22:17

8 Answers 8


There are two ways in which you can view a Question/Answer site. One is as a real-time "I've asked this question and I want a quick answer now" type of site, and the other is as a longer term repository of questions and definitive answers. I think that many of us here take this latter view of the site. As a result, we have lots of very high quality answers, and a relatively low number of competing answers for many questions. We also tend to comment on answers to improve them rather than posting competing answers. This is a very good thing, IMO and not only makes the site a useful long term resource, but also fosters a real sense of community where people are willing to help each other rather than fighting over reputation.

I think that having lots of competing answers to any particular question just generates noise and makes the site less useful. Also, as you can see from other sites where this practice is more common, when there are lots of answers, early answers tend to get upvotes and very good but later answers sometimes don't.

The commenting facility however, provides a real time remedy to this situation: a comment that also provides an answer may solve the original questioner's problem, and (as others have noted) at the same time provide time for us to find and link to other related or duplicate questions. This again adds to the value of the site.

Although it might appear that this system would lead to a lot of unanswered questions, if answers in comments don't get turned into answers, in practice this is not much of a problem. We hold a monthly "Answer the unanswered" session, which catches most, if not all of these cases, and I suspect that the number of unanswered questions that have been answered only by a comment is very low. Also, when comments are resolved they are usually turned into answers. If you see such cases, feel free to ping the commenter in another comment and ask them to convert their comment into an answer.

  • 15
    Bravo. I'd like to give +2 to this but can't. Feb 5, 2013 at 15:00
  • 3
    I've added Stephan's missing vote. Really nice answer.
    – percusse
    Feb 5, 2013 at 15:36
  • 2
    One downside of posting elements of an answer as a comment, though, is that it may mislead new users as to how SE sites operate. I think that practice actually explains a large number of TeX.SE questions that remain "unanswered" even though they have received perfectly satisfactory answers.
    – jub0bs
    Apr 14, 2013 at 17:33
  • 5
    @Jubobs We hold a monthly "Answer the unanswered" session, which catches most, if not all of these cases, so I suspect that the number of unanswered questions that have been answered only by a comment is very low. Also, when comments are resolved they are usually turned into answers. If you see such cases, feel free to ping the commenter in another comment and ask them to convert their comment into an answer.
    – Alan Munn
    Apr 14, 2013 at 20:25

I also often find myself dropping a comment instead of answering. In addition to what was already answered here, my reasons mostly are

Might be too localized. If the answer to a huge question is a one-liner, then more often than not the real problem was a little thing the OP overlooked, having nothing to do with the perceived question at all. In that case, it's often better to close or delete the question, and an answer might be contraproductive.

Not tested / Too little time. When I post an answer, I would at least test it really works, and give an appropriate amount of explanation. When I don't have the time or can't be bothered to test, I often resort to shooting a comment pointing in the intuitively right direction (instead of doing nothing at all). If it happens to hit the point, I can later come back and extend the comment into a "real" answer (if there isn't one already).

  • 29
    This one holds for me too. The key is the phrase "instead of doing nothing at all". I'd rather put a short "is it XYZ?" comment than not help at all. Feb 5, 2013 at 15:19

In general, you are correct that something which answers the question should be given as an answer. However, there are some subtleties to this. The first is as marczellm says that a one-liner may not really 'feel' like an answer: it may be the core of an answer but need expansion, for example. There is then the question of whether the comment really answers the question, which you might want to check out. Not all questions are great, and some of them you may want to comment on with the expectation that they'll be closed. That's particularly the case when you're posting something that can be found in a simple guide or with a web search.

During our 'Answer the Unanswered' sessions, one of the things that happens is that people are encouraged to convert comments to answers where this seems the best approach. (That's done by leaving a comment asking the earlier commenter to take action: I use @<person> Looks like an answer to me.) If the original commenter does not do this, or is not about, then a community wiki answer is often left. This is a way to use the info from a comment without gaining reputation, so without feeling you are 'stealing' from anyone.


Joseph's post speaks about it slightly, but I would like to add one comment. This often happens when there is not enough information in the question, usually because of a missing MWE. This makes the question difficult to answer, so we often just try to guess what is the problem and we shoot a possible solution, as a comment. If later it is found out that the solution was right, it is made an answer so that it can be up-voted, accepted, question resolved.


It may be that people don't consider a one-liner solution to be worth an Answer. There's the feeling that an Answer with capital A is something more detailed. This feeling is (at least in my case) strengthened by the fact that somehow, on this site, there are much more detailed answers than one-liners. The Answerers usually go to great detail even about simple topics because there is the chance that the question poster does not know the correct terminology yet and needs more information.


I sometimes posted advice in a comment, because of being pretty sure that the question has been answered before and I just did not find the duplicate yet. As soon as I would find the duplicate, I would add a link.

It would be good to check for a duplicate at first, before quickly firing answer for rep and badges. I guess commenters often avoid adding an answer to a potential duplicate, or to a very localized question with a small mistake to be corrected.

If a question is likely to be closed, adding a comment may be better than adding an answer (which would conflict with the intention of closing).

If it turns out that it should be a real answer, it's a good etiquette that the commentor makes one, or would be asked so. In case there's not response we often noticed community wiki answers made instead.


I hope people will forgive me if I use the freedom of the Meta group format to add a commentary that is too long for a comment.

@AlanMunn's answer is one that I find extremely helpful for clarifying the culture of TeX.SE. The other answers are also very helpful. However, I think those who function as leaders in TeX.SE should consider the possibility that the prevailing practice discourages participation in TeX.SE. I am not in any way an expert on LaTeX, but I have 20-25 years of experience with it, and have occasionally hacked raw TeX, edited style files, Postscript files, etc. It turns out that most questions in TeX.SE are narrowly focused on issues I don't care about and don't know about--OK, no problem there. But I have gotten incredibly helpful answers on TeX.SE, for which I'm grateful, so I want to contribute. In the past, when I saw a new, unanswered question on a topic I knew something about, I would open it. However, the question had usually already been thoroughly answered in comments. This became frustrating. So I usually gravitate toward not participating in TeX.SE, except when I have a question. What's the point? I don't care about most of the questions, and I can't tell which other ones are answered---and they usually are. The fact that answers are eventually moved out of comments into actual answers doesn't help with this problem. (Of course there are a few questions of interest whose answers I don't know, but those are hard to find.)

Voting on and accepting answers isn't a perfect system, but it's a good system with thought put into it, and it can work pretty well. The TeX.SE practice partly discards the benefits of that system, it seems to me. As for people who are just trying to earn points, well, I don't really see the problem. The SE system is designed to encourage that behavior, and for a good reason. If people give bad answers, they won't get voted up. So what? (If they're really bad answers, they'll get downvoted.)

Comments asking for refinement, broadening, etc. of questions are a different matter. These are of course appropriate and are used in other SE groups. No objection here. I often post possible answers, along with requests for clarification, as comments in other SE groups when I'm not sure what the OP is asking. In TeX.SE, in addition to questions that are looking for clarification, refinement, or broadening, there are many comments that contain what I consider full-fledged answers.

Maybe I'm misinterpreting the situation, but it sounds as if TeX.SE functions as a group in which a few experts end up providing most of the answers, except for more obscure issues. That's an OK system. It might place additional burden on the experts, but I have no problem with that if they don't. However, it may be that the practice also ends up restricting the diversity of the pool of people who are around to answer the more obscure questions.

Anyway, those are the thoughts I have. Thanks.

  • 4
    You raise some interesting points (perhaps this could even be a new Meta question.) But I think your analysis is somewhat off in places. We have a pretty active user base, so I don't think that comments are negatively impacting participation. Furthermore, the issue of one not being able to post an answer because someone has already commented isn't really solved by encouraging answers: the same people who comment could just as easily post an answer. But the point of my answer was that we want to encourage detailed answers if possible so that for most questions there will only a few answers.
    – Alan Munn
    Aug 17, 2013 at 18:43
  • 4
    In addition to @AlanMunn's comment, go ahead and post your own answer without reading the comments. There are experts yes but there are also lots of non-wizards who makes the place super nice. No need to shy out, once you start answering you will see that a full answer requires quite some work :) so if you have a detailed answer please post it. It's a trial and error process not a SO-ruled private TeX support group. The main issue is that we avoid competition for the sake of quality.
    – percusse
    Aug 18, 2013 at 21:36
  • Thanks @AlanMunn. This is a different culture than in some SE groups, obviously. I think I understand the rationale. I'm not sure that I prefer it. But I am grateful for good answers. I get a lot of use out of TeX.SE without asking questions--just searching for answers--as you'd expect.
    – Mars
    Aug 19, 2013 at 17:10
  • Same comment to @percusse.
    – Mars
    Aug 19, 2013 at 17:10

I'm trying to reanimate this question. We now have more than 4.000 unanswered questions and the list of them is full of questions, which have been answered in comments.

Now some folks come and delete answers, because they suffer under the delusion, a comment would be much more suitable! See here: https://tex.stackexchange.com/a/196636/4736

I don't know what to do, but a list of unanswered questions, where at least half of them have an answer is of very limited use.

  • 2
    Thanks for calling me delusional! Much appreciated! :) It's always good to know that having a different opinion isn't just that but – in this case – a delusion!
    – cgnieder
    Aug 16, 2014 at 15:13
  • Well, on the other hand I praised you here: tex.stackexchange.com/a/196547/4736 So you see, sometimes you are right, don't take an occasional error too seriously.
    – Keks Dose
    Aug 16, 2014 at 16:29
  • I don't :) I just find it funny (strange even) that you classify my opinion as delusion and error rather than as a disagreement which we could discuss. As it stands we can't so I won't. :)
    – cgnieder
    Aug 16, 2014 at 16:34
  • 2
    Those Germans, they take everthing so seriously. Hey, you deleted my answer, instead of discussing about it. And after all, it contained the nucleus of the solution. Will you come to the Dante-Meeting in Karlsruhe on September 20th? You could invite me to a beer for deleting my answer by mistake, and I'll pay a beer for you in turn. Sounds like a plan?
    – Keks Dose
    Aug 16, 2014 at 16:56
  • 2
    Just so you know: there are no harmed feelings on my side. :) I just clicked through the review queue and your answer popped up there because of »low quality«. Karlsruhe will probably the first meeting I'll visit. I'm looking very much forward to it! I hope we can drink a beer (or two or three) together :)
    – cgnieder
    Aug 16, 2014 at 17:15
  • I just learned that my nephew will be enrolled in school on September 20th and I promised my brother to attend... so I'm not sure any more if I'll make it to Karlsruhe
    – cgnieder
    Aug 17, 2014 at 19:40
  • 2
    Instead of reanimating this question, why not go through the list of unanswered questions and ping using @<user> each commenter who you think has answered the question in a comment asking them to convert their comment into an answer? Good luck with that.
    – Alan Munn
    Aug 17, 2014 at 21:07
  • 1
    and the main issue in my delusional comment is until it is confirmed by the OP. And EOD is an unknown concept to me ;)
    – percusse
    Aug 19, 2014 at 21:31

You must log in to answer this question.

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