I don't see many questions that were closed that should have been kept open. There have been some that could be turned into questions appropriate for MO and were closed, but several were edited, reposted, and then remained open. Right now there is one closed question on the first page and 2-3 others that should be closed IMO.

I would be grateful for a clarification regarding this (either way).

Until reading this, I was convinced it was considered good form to rather use a comment than an answer in cases were one believes one wants to contribute relevant information but possibly not a full answer.

A specific situation were this problem arises for me is when the question is (in my opinion) not entirely clear;
then it can happen that I have an answer to one interpretation but not the other.
So, I give a comment along the lines "Not sure you mean this. If so, then... Else, please clarify."

For purely practical reason (eg, character limit) I would often prefer to use an answer for the same purpose.
But thought it was considered rather bad form.

Greg: there is only a really fine line between what an_mo_user describes and what you describes. In particular, you use the words "very small or very rigid reasons"; or, in other words, a judgement call.

So I don't think it is realistic to try to set a policy on when one should use the comments and when one shouldn't. Some of the previous discussions have indicated that it is acceptable to post a comment requesting that the poster of the significant comment to convert his comment to an answer. And that it is also acceptable, if the question is essentially answered in the comments, to post a hash of that answer based on the comments in Community Wiki mode.

---

As to your second question: I agree in principle. But there is a rather large variation between what is considered advanced undergraduate level material and beginning graduate level material among various institutions. So I would prefer if the FAQ not phrase the distinction that way.

It is my impression that there are a lot of questions in the "Unanswered" list that are actually answered in the comments. For people who like to answer questions, it is frustrating. Surely the most efficient version of MO would keep clearing questions and have as large a fraction as possible with an accepted answer (and not just upvoted answers). Now I admit that even my proposal is just lip service. Maybe the right first question is just how many questions there are in the unanswered list that are moribund because they are answered in the comments. If we had that data, then we would know how big the problem is and we could consider what if anything should be done about it.

That is, the advice that one can just post a CW answer would only count for so much if there are already hundreds of questions in this condition.

Let me also say, practicalities aside, that both of these issues can contribute to an air of machismo on MathOverflow. Many people already see MathOverflow this way for various reasons. So we should consider, as a matter of etiquette, things that might be making it worse, even if it's not intentional.

While I see where this is going, and don't necessarily disagree with the aim, I'm not sure I actually agree either. In terms of the FAQ: trying to do one thing well (answer research-level questions) is typically a better aim than doing several things. That's how the Internet works: fill a niche, create the better mousetrap, the community comes. Also if some contributors respect the guideline that answers as such should be full and considered, then they will use the comments to, well, "comment". The appropriateness of doing that is to some extent in the beholder's eye: I have just done it in a question where the "original poster" is plain wrong about something, and I feel a better person for not beating them over the head with that, just noting it in passing. There is an option that fits part of the issue that Greg raises, which is to use a fresh question to ask for amplification of the cryptic.

I think Qiaochu has the solution to this problem (see his post preceding this one).

Quite often questions that have answers that are not formally "accepted" are simply that way because the original poster was an anonymous poster that never came back, or if they did they didn't bother to follow through and officially accept.

If I vote to close a question sometimes I give a hint as to an answer in the comments but I wouldn't supply an answer to a question I had voted to close.

Incidentally, I believe that BCnrd has specifically chosen only to ever respond in comments. (Do a search here to find more on that, I think there's a thread somewhere.) Make of that what you will, but it does mean that whatever consensus is reached there will always be some people who contribute positively to the site but in such a way that it goes against the desired practice.

"If the question was completely answered in the comments, one resolution is to copy the answer into a CW answer with proper attribution."

I would like to also endorse this course of action. In my opinion the post doesn't have be community wiki, but I suspect some other people's oddly moral view of reputation would be offended by this.

As for the aim of having as many questions with an accepted answer as possible, there is another obstacle that I find annoying: users who just refuse to accept answers, even though they come here regularly. There is this user for example, who has asked ten questions, most of them received satisfactory answers, as far as I can tell, and none are accepted. I wonder whether a moderator could email the user and ask him to accept the answers that satisfactorily answer his questions. I left a comment to that effect here, but to no avail.

I vote «cryptuosity» to be for 2011 what «discussion-y» was for 2010!

Dear Gerry,

As something of an aside, I don't think that it's fair to characterize Brian Conrad's comments on 50289 as cryptic. That question was a very technical one, about how certain deformation functors behave when one allows Artinian rings with varying residue field. Brian answered this question quite succinctly in his comments, and his explanations/references are easy enough for an expert to follow. Presumably anyone interested in a technical question of this kind on deformation theory (in particular, the OP) also has the technical wherewithal to follow the answer.

As evience, note that after Brian points out to the OP that he has answered the question, the OP doesn't follow up with more comments, so presumably the OP in fact understood that their question was answered. I don't think there is any reason to believe they were ill-served by Brian's comments, despite the suggestion to the contrary in the last paragraph of your post.

In any event, I would be happy to put Brian's comments into an answer, if people think there's a pressing need.

Regards,

Matthew

Matthew, it was Greg who (first) characterized Brian's comments on 50289 as cryptic. If Greg found the comments cryptic, I'm sure more than a few of us would also find them cryptic. But I defer to your judgement that an expert would find them easy to follow.

I hadn't thought much about the comment issue before, but now that Greg's brought it up, I think he's basically right. I think that answering questions in comments often does come across in a supercilious way, which adds a small amount of pollution to the atmosphere of the site, although in retrospect I'm sure I've done it myself. (I know of only one graphomaniac who puts his answers in the comments to control himself.) It seems that a reasonable rule of thumb would be that if it's worth your time to type up an answer to a question, then it's worth posting it as an answer. If you want to make it clear to everyone that you think your answer is not a big deal, then it's easy enough to add a few modest words to that effect. (Though in my opinion, what makes you look best is to just answer the question as clearly and simply as possible.) You can also silently skip a question if you feel it's too far beneath you. That's ok -- no one will ever know and it won't pollute.

I'm less certain about Greg's second point about math.SE, because I do think it's possible that too many low-level questions could dilute the site and drive people away. The participation of many good mathematicians is the most valuable thing on the site! But I agree that sending people to math.SE can come across as small. One possible solution would be, rather than trying to make a general policy that covers all cases of that, to add something to the FAQ about how to send someone to math.SE respectfully. (Hmmm... Would that look like we're treating people like children? Well, it's an idea.)

As far as redirecting people to math.SE, I can think of a few questions that were not necessarily of an inappropriate level for MO, but for which it appeared the *user* was not at an appropriate level. After all, many questions, especially the vaguer ones, can be answered at a variety of levels. What I've seen happen sometimes is that the question will be taken as a springboard to go into much more high-level considerations, which makes MO look like a bunch of elitist jerks (though I'm not judging intent here, I am quite ready to believe that this happens mostly unintentionally).

This means that I've sometimes encouraged newbies to ask on math.SE instead even when the question was level-appropriate for MO, if my gut feeling was that the MO answers might not be particularly helpful for the user in question. I readily admit that this may be jumping the gun when the question has not been answered yet, but I'd rather do that than having new users get a sour first experience of MO.

@Gerry: Good point. I've posted these kinds of comments, too. I guess I'll have to hunt down a specific question to explain exactly what type of question I meant in my previous remark.

@an_mo_user, I would be interested to see a question showing the phenomenon you have described.

@Gerry Myerson: the following is in some sense a non-example (but still illustrates the pattern I mean), as one of the commenters followed up and provided an official answer, but note that the first version of the ansewer was written after the last but two comments, in particular after the asker already said he understood the answer in the comments.
(Then there was also an edit of the question, and an edit of the answer; but this is unrelated.)

http://mathoverflow.net/questions/55152/this-limit-converges-to-the-partial-derivative

ADDED later: While browsing the site, I came accross the following, which is perhaps a better example:

http://mathoverflow.net/questions/33757/quadratic-solutions

(It is a coincidence that you are involved, I did not specifically search for it; also, as said,
I did not mean to criticize the practise.)

@Gerry Myerson, I essentially share your point of view on the situation.

You won't have to worry about BCnrd leaving answers in comments anymore.

:(

@Gerry: He hasn't been active for a few weeks/possibly months.

One of them uses vowels :)