Sometimes, I accept one answer, and it's a good answer, and partly solved my question. So I accepted that answer.

After some time, someone gives one more good answer, maybe better, maybe with much greater details filled in.

Is it good to accept the new answer? Would people be hurt by OP's action?

Though I upvoted them both.

  • 25
    $\begingroup$ This is very much ok, and while it might hurt some, they should understand. If you want to avoid hurting anyone's feelings, it might a good idea to wait for at least a day before accepting any answer. When you accept an answer, you are signalling to other users that you consider your question fully solved. There are some users who will not even look at questions with an accepted answer. There are also others who will add an extra answer, but only if they think they can offer an interesting different viewpoint. Anyway, accepting any answer will prevent many new answers from appearing. $\endgroup$ Jun 29 '13 at 5:08
  • 5
    $\begingroup$ Better, definitely yes. With much more detail filled in, that poses problems. If the filled in detail is stuff you likely would not be able to fill in yourself, then yes. But if the less detailed answer that you accepted let you contribute to the solution, then there there is reason to think it may be better than a fully detailed solution. $\endgroup$ Jun 29 '13 at 7:12
  • $\begingroup$ @AndréNicolas hi, could you give me a link where with colored tex redering? In posts or answers, I saw some of them, but I cannot find one now... $\endgroup$ Jun 29 '13 at 7:16
  • $\begingroup$ Here is one. One answer has some colour. I have seen instances with much more. $\endgroup$ Jun 29 '13 at 7:26
  • $\begingroup$ @AndréNicolas ah, thanks. I'm trying to convert mathematica formula to tex with color. This is one question about this at Mathematica.SE. mathematica.stackexchange.com/q/27840/6648 $\endgroup$ Jun 29 '13 at 7:34
  • $\begingroup$ Regarding the use of color: meta.math.stackexchange.com/questions/4195/…, don't overdo it. $\endgroup$
    – Asaf Karagila Mod
    Jun 29 '13 at 9:24
  • $\begingroup$ @AsafKaragila thanks for reminding me of that, I'll try that in such case: I added [step1] such sign posts/tags in the long formula derivation where I have problems, which maybe not so easy to locate. $\endgroup$ Jun 29 '13 at 10:08
  • 3
    $\begingroup$ Don't you ever do that to me. Ever! (Of course it is ok to do that and people should not take offense or get "hurt"). $\endgroup$
    – Julien
    Jun 29 '13 at 22:51
  • 1
    $\begingroup$ To add to what @Asaf said, I looked at the use of color for an e-text I'm writing, The consensus research is that, as Asaf said, to use color sparingly if you want to be considerate to the 5%-10% of males who have a problem with this. If you must use color, use just blue unless there's a very compelling reason otherwise. $\endgroup$ Jul 1 '13 at 1:33
  • $\begingroup$ To add to what @Rick said, here is a really awesome link on the topic: lighthouse.org/accessibility/design/accessible-print-design/… $\endgroup$
    – Asaf Karagila Mod
    Jul 1 '13 at 10:34
  • 4
    $\begingroup$ The few times this has happened to me I have found it a little irksome, mainly because it was not clear to me why the change was made (and sometimes after a long delay). I have the same reaction to downvotes. So my suggestion would be leave a brief comment explaining why, otherwise only change if there is a fairly obvious and significant improvement. $\endgroup$
    – copper.hat
    Jul 1 '13 at 15:53
  • 1
    $\begingroup$ The first (and only) time this happened to me, my answer had 5 upvotes, and the new accepted answer had 0. I was a bit miffed at first, but then I felt silly, like the fable of the greedy shepherd. It's always good to remember that the OP is just one person, where there are likely several more people who found it even more helpful than the OP did. $\endgroup$ Jul 31 '13 at 14:19

Admittedly, there are times when changing the accepted answer may annoy someone. Here are some circumstances:

  1. When the question is a homework question, the accepted answer is a well written hint and the newer answer is a full solution. Some may think that by accepting the full solution, you are encouraging the practice of doing homeworks for the others on MSE.
  2. Sometimes, an OP may ask what's wrong with his/her lengthy, messy proof to a problem statement. Someone goes over the details, pinpoints the error and the answer is accepted. Later, another one posts a very clever proof. Although this new answer mentions absolutely nothing about the error in OP's proof (which is what is being asked), it gets accepted.
  3. When the original accepted answer is correct, but the newer one is not. In this case, the new answer is accepted, probably because the OP cannot judge what is correct and what is not.
  4. When the original accepted answer is complete and correct, but the new is incomplete, and the one who gives this new answer does not seem to have any clue to complete the answer. Yet this new answer may be accepted for various reasons (e.g. the partial proof looks flashier and more advanced, or it looks less advanced and more down-to-earth so that the OP feels more at home).

There are tons of reasons that changing the accepted answer may frustrate someone. In my humble opinion, as we cannot please everybody (and we are not obliged to do so), it is fine to change the accepted answer when one sees fit, as long as the new accepted answer is technically correct.

However, I think it is also courtesy to upvote the original accepted answer when you change your mind (again, assuming that the original answer is technically correct).

  • 1
    $\begingroup$ Another example is when the OP just unaccepts and then deletes their user... :| $\endgroup$
    – Asaf Karagila Mod
    Jun 30 '13 at 6:43
  • $\begingroup$ @AsafKaragila Yes, hit-and-run is the worst, but usually, those people who delete their one-time user accounts don't even care to accept an answer. $\endgroup$
    – user1551
    Jun 30 '13 at 9:21
  • $\begingroup$ Oh, I'm not talking about hit-and-run. I'm talking about some well established user that unaccepted my answer (and said it was great in the comments when I asked why they'd done that), and requested that his account is to be deleted. $\endgroup$
    – Asaf Karagila Mod
    Jun 30 '13 at 9:26
  • 1
    $\begingroup$ Strange story indeed. This seems more of an exception for this user, who left most acceptances in place when departing the site. They were automatically transferred to Community user. As a result of deletions, Community has lots of accepts and more votes (over 15000) than any real user. $\endgroup$ Jul 1 '13 at 12:39
  • 1
    $\begingroup$ Usually if I accept an answer, I also upvote it. That means should I ever decide to change the accepted answer, I cannot upvote on the originally accepted one again. Which means I cannot do what your last paragraph suggests. $\endgroup$
    – celtschk
    Jul 6 '13 at 8:52

Would people be hurt?

No. Poking a person with a sharp stick usually hurts them. Clicking a checkmark on SE site doesn't. As the OP, you are to choose the answer that's most helpful to you. When the set of answers expands, the maximum of helpfulness among them may be attained at a different point. It's mathematics:
$$\operatorname{arg\,max}\limits_{x\in A} f(x) \ne \operatorname{arg\,max}\limits_{x\in B} f(x) \quad \text{in general, when}\ A\subset B $$

Related: OPs changing mind about accepting solutions.

  • $\begingroup$ Reading that question about OP changing accepted answer...it reminds that two years ago while I was new to SE, I did the same thing on a TeX.se question I asked...hopefully it didn't hurt someone... $\endgroup$
    – Shuhao Cao
    Jun 30 '13 at 6:33
  • 9
    $\begingroup$ This answer completely marginalizes the very unhealthy emotional bonds, which I am sure many people on the site have, to their actions here. $\endgroup$
    – Asaf Karagila Mod
    Jun 30 '13 at 6:44
  • $\begingroup$ Haha, love the comparison :). $\endgroup$
    – dreamer
    Jul 1 '13 at 16:05
  • $\begingroup$ I think a more appropriate analogy is maximizing over a set with a partial order. $\endgroup$
    – copper.hat
    Jul 1 '13 at 16:13

You can limit the potential hurt by taking your time before accepting any answer, especially if no answer completely answers your question yet.


The first time one of my answers was "unaccepted" I was surprised, not hurt. I mean, I guess at that time I was so eager to make one the earlier (in terms or reputation points) benchmarks that I felt a twinge of disappointment; but once I learned that whereas votes can't be changed, acceptance can (and guess how I learned that--by having the answer unaccepted!), I just "accepted" the situation.

It seems to me the easiest "solution" here is to educate users so their expectations are based on reality.

Another thing is just to wait awhile before accepting to let the answers roll on in!

But if you feel an answer is what you want and need, well, the decision is yours . . .

As to whether I care about having an answer unaccepted or not: well, I cared a lot more when my rep was under $1,000$ than I do now . . .

  • 1
    $\begingroup$ This is the most woke answer that I have read on this site $\endgroup$
    – Buraian
    Sep 12 '20 at 23:24

You must log in to answer this question.

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