# Are there codified or at least customary acceptance criteria?

In this question, there are two answers, both of which I up-voted, but I didn't "accept" either since there didn't appear to be a reason why one stands out as clearly preferable to the other. Apparently, my lack of "acceptance" offended someone who's taking me to task for it in comments on one of my questions. Some people (or at least one person) seem to get rather passionate about acceptances. Has anyone defined codified some criteria that say when one should "accept" an answer and what that is supposed to mean?

• I believe Willie's answer to the previous question "What factors should be considered to accept an answer" is relevant here. – user856 Mar 6 '12 at 21:20
• It's just reputation points. One criteria might be lack of obnoxiousness. Sometimes people win marathons (a 26+ mile race) by a fraction of a second. – Jay Mar 7 '12 at 0:01
• Well, if I wanted to be a grammar nazi, I might say that you should have written "One criterion might be..." and that "criteria" is plural. – Michael Hardy Mar 7 '12 at 0:44

If your only problem is that there isn't a clear reason to prefer one answer to the other, but you still feel that your question has been answered satisfactorily, then I'd say to go ahead, flip a coin and accept one of them. Or choose the one whose author needs the rep the most, or whatever.

Getting the question marked as "no longer open" is more important than making sure nobody is treated more fairly than someone else, I think.

That being said, I think there is too much mindless pressure for users to get their "accept rate" up, just for the sake of having it up. The person who took you to task complained about a 42% accept rate, which is slightly ridiculous in my eyes. I wish we could reserve that kind of remarks for users who seem to be avoiding to accept anything on general principle (or out of ignorance, but then the remarks really should be phrased more kindly than they usually are).

One reason for this is that the warm fuzzy feeling that comes from having an answer accepted is becomes considerably cheapened for me by the possibility that the answer may not have been very helpful for asker at all, perhaps he's just being bullied into "working on his accept rate" for its own sake.

• I didn't know there was such a thing as getting a question marked "no longer open". Why should one want to do that? – Michael Hardy Mar 7 '12 at 16:44
• Surely you have seen that in lists of questions, questions with an accepted answer have the number-of-answers shown on a green background? Why one should want to do that is to help other users of the site quickly find the questions where their work will do the most good. – hmakholm left over Monica Mar 7 '12 at 16:47
• Indeed, one can argue that pushing readers to accept answers that they don't want to accept is highly detrimental to the site, because it discourages later answers that might be more comprehensible or more enlightening to the OP (and readers). IMO, accepting answers too quickly is one of the major problems of the site. The content of the site would be much richer if the software forced the OP to wait a few days before accepting answers. – Bill Dubuque Mar 7 '12 at 18:33
• For a peon's perspective: indeed, I have been peer-pressured into keeping a 95-100% acceptance rate. When I am about to ask a new question, I go back and make sure to accept an answer to my most recent question. Sometimes it comes down to a coin flip. sometimes there is no best answer, but I would rather (for whatever subjective reason or misconception on my part) accept "unacceptable" (or "underacceptable") answers than risk having future questions ignored because DP doesn't like my acceptance rate. – The Chaz 2.0 Mar 7 '12 at 21:03
• @TheChaz, I'm one of the people who bugs others about accept rates, but it would never occur to me to chide someone for having "only", say, 90%. I don't think I've ever commented on any rate over 40%, and mostly when I comment the rate is in the 0 to 20 range. I don't know who has been pressuring you to keep over 95%, but I'd advise you to resist any such pressure. – Gerry Myerson Mar 8 '12 at 6:29
• @Gerry Myerson: Thanks for your input. – The Chaz 2.0 Mar 8 '12 at 12:02
• Are questions with accepted answers really "no longer open"? If I'm not mistaken, it is possible to post an answer after another answer is accepted. – Michael Hardy Mar 8 '12 at 16:12
• They are "no longer open" in the everyday sense of "open questions" as ones for which a good answer is not known. – hmakholm left over Monica Mar 8 '12 at 18:36

Firstly, I fully endorse Rahul's comment. :-)

Secondly, let me explain, or at least give my point of view, on why some users are getting worked up about the acceptance rate. There are two main ways of user-interaction which can be used to express gratitude for providing an answer beyond a simple "thanks." They are "up votes" and "answer acceptances". (There's also "bounties", which we'll ignore for now.) Many people have their own ideas about what each of the two means, so I won't try to attach meaning to them. But I will point out that there are two main technical differences:

1. Up votes are worth less reputation than acceptances (+10 vs +15)
2. Acceptances can be more easily rescinded than votes (a vote can only be changed within a certain time frame, or after an edit was made to the corresponding answer; an answer acceptance can be removed any time).

Because of the difference, people regard the two with different values. Now, in spite of the actual intentions behind a questioner deciding not to accept an answer, some users may interpret a low acceptance rate as (the list is not intend to be complete, but just a description of some of the comments I've seen)

1. the questioner being "stingy" with an intangible commodity with no actual worth and which violates the "have a cake and eat it too" rule.
2. the questioner is "leeching" from the community; while answers here are given voluntarily and time spent here are of the users' own wills, some people still think that people who ask questions should "give something back" to the community, be it providing answers or giving people reputation points.

I will not debate whether either of the views are reasonable. This is just to give you some ideas on what may be behind comments calling you on your accept rate.

Thirdly, for (not-so-)new users with low reputation and a habit of asking poorly motivated questions, I do sometimes decide not to answer a question because of extremely low acceptance rates. This is due to mainly lack of feedback. If the user consistently refuse to provide the context in which the questions are asked, nor any details about his/her personal mathematical background, and if the user does not indicate whether previous answers to his or her questions are actually useful, how am I to know whether the answer I have in mind will be useful to the person who asked? This is especially frustrating in the case when I know I can explain the same thing in several different levels, but the user gives no indication which one level is most suitable.

My opinion is that it is perfectly fine not to accept any answer if one has good reasons to do so.

For example, during the month I was testing MathOverflow, I posted a question on a tantalizing problem of Halmos, namely: explain why the following "proof" gives the correct answer.

$$\rm\begin{eqnarray} (1 - ba)^{-1} &=&\rm: 1 + ba + baba + bababa + \cdots \ &=& \rm: 1 + b (1 + ab + abab + ababab + \cdots) a \ &=&\rm: 1 + b (1 - ab)^{-1} a \end{eqnarray}$$ One easily checks that the final answer is correct in every ring, even though the derivation using (formal) power series generally cannot be interpreted in every ring. As Halmos wrote

Why does it all this work? What goes on here? Why does it seem that the formula for the sum of an infinite geometric series is true even for an abstract ring in which convergence is meaningless? What general truth does the formula embody? I don't know the answer, but I note that the formula is applicable in other situations where it ought not to be, and I wonder whether it deserves to be called one of the (computational) elements of mathematics.

A long time ago (perhaps sparked by a conversation with G. C. Rota), I found what I thought was a good answer to this question. But none of the current MO answers (or comments) have helped me to remember this answer. So I have not yet accepted any answer, in the hope that someday the question will get bumped into the view of MO reader who knows the method that I had in mind.