How should SO handle users who don't "maintain" their questions by selecting answers, or remarking newer/better answers as the chosen one?

Many questions...

  • Have no answer marked correct, even when one is plainly there
  • Are not reviewed by the poster to select a new correct answer

I don't mind taking the time to answer questions, but don't see the point when I don't get "reputation" for doing so since no one reads those questions, and reputation seems to be the point of the site.

I think the "remind to accept an answer" feature is there now, since I just saw it when coming back. I missed it the first time though.

  • 1
    Were you going for an ironic statement by not selecting an answer for this question?
    – alecmce
    Feb 28, 2010 at 4:17
  • Yep... life is good I guess. @alecmce, I know I should really select answers for questions that have been migrated to somewhere I don't have an account. Also, it was a community wiki, and I didn't think that was supposed to have an answer selected. Anyway, I don't know what a correct answer for a subjective question about how to change behavior is. If I mark one correct, can I expect it to be implemented :)
    – Andrew
    Mar 9, 2010 at 16:51

14 Answers 14


Accepting an answer means putting one's stamp of approval on it as answering your question. I try hard to accept an answer on all of my questions, but I have about 1/3 that were not answered in a way that I felt I could endorse. In one or two cases, I felt like my own answer was the best provided (at a later date), but there is no way to accept your own answer. In some cases, I'm waiting to get more responses.

Not every question gets answered to the OP's satisfaction.

I think there should be a gentle reminder, periodically (that can be turned off) to accept answers to open questions, but it should be left at that. I don't think others should have the ability to accept answers on my behalf.

EDIT: since a while it is possible to accept your own answers.

  • @tvanfosson: agreed
    – RedFilter
    Dec 18, 2008 at 2:29
  • Agreed there are 23 such questions on my profile, 3-4 of which I answered myself.
    – Unkwntech
    Dec 18, 2008 at 2:36
  • Then you shouldn't answer them yourself, just add your answer to the next-best one and accept that.
    – user14860
    Dec 18, 2008 at 2:49
  • 3
    And now you CAN accept your own answers, you just won't get rep for them and they won't float to the top.
    – user14860
    Jan 8, 2009 at 0:10

My vote goes to doing nothing.

This site shouldn't have tons of rules regarding behavior. We won't keep a large userbase if the site requires you follow a super strict guideline.

Requiring that users check old questions or accept answers lest they receive a rep penalty seems like a we are running a military regime.

  • 1
    I have to agree especaily since there are times when a question has not received a decent answer, just look at my profile there are 23 questions that I have asked that either don't have a good answer or, I answered it myself, so it appears to not have an answer.
    – Unkwntech
    Dec 18, 2008 at 2:35
  • If you want an accepted answer to your question and it's yours, then you should just add your answer to the next-best one and accept that.
    – user14860
    Dec 18, 2008 at 2:53
  • Some/Most people don't have that luxury though. (2k rep req.)
    – jjnguy
    Dec 18, 2008 at 3:05
  • 1
    Doing nothing is the best option. The highest voted answer will remain in place, so that should suffice for most purposes, There are always going to be a few cases where a really good answer fails to get accepted, but I don't think I've seen many good answers get no votes.
    – Mike Burton
    Dec 18, 2008 at 6:25

If you've come here for reputation points, that's the wrong idea. Reputation points are just an inhibitor to stop silly behaviour until you've proven yourself. The reason you're here (in my opinion) should be to provide useful questions or answers for the greater good.

In answer to your two specific queries:

1/ Many questions have no answer marked correct, even when one is plainly there.

I think you mean accepted, which is subtly different from correct. Accepted answers are totally under the control of the questioner. Correct answers, in terms of what the SO community thinks, are the ones with the highest net vote count.

I've had one answer that received a net 36 votes (the highest for that question), the accepted answer was -18 (a subjective question regarding breaks in loops so I didn't care that much).

Another accepted answer was clearly wrong to both me and the wider SO community (to do with SQL selecting minimum value across columns as well as within columns) - the accepted answer would not scale well and my answer would. However, I'm pretty certain I offended the questioner at some point by telling him his schema was rubbish and he may have accepted the other answer out of spite. Such is life, I should learn to be more tactful.

In fact, your situation is better in that the correct answer will bubble to the top if the questioner hasn't accepted a 'lowlier' answer (for want of a better term).

Basically I think we just need to understand the distinction between accepted and correct. I've often thought that the accepted answer shouldn't automatically be at the top but rather have (for example) 10 votes added, so that if 10 other people thought another answer was better, that would render the acceptance moot. It would stop ignorant questioners from choosing what's at the top of the list, once again returning that power to the SO community.

I'd put something up on uservoice but I'm not passionate enough about it. Anyone else want to give it a shot?

2/ Many questions are not reviewed by the poster to select a new correct answer.

Unless you've hacked into SO and can see the access logs, I don't there is no way you could tell this for sure.

I, for one, monitor all my responses to see if anyone's added new information which makes the old accepted answers worse. I've never found one yet and, to be honest, I'd prefer to see the accepted answer edited to provide the extra information since that guarantees it will still be at the top of the list.

Again, this comes down to rep-seeking. I have edited other peoples accepted answers (only to add at the bottom, not change the original) to add some pertinent information of my own and then deleted my answer that had that information in it.

That way, the best information stays at the top.

In addition, I've once bought another answer's info into my own answer when mine was the accepted one, again to ensure that the best answer is at the top.

If people really want an accepted answer, maybe a uservoice request to automatically accept the highest voted answer when the question hasn't had any activity for (e.g.,) 4 weeks would be the way to go. If the original questioner doesn't like that, they can always change it or un-accept it (although it'll accept again in another 4 weeks so they'll have to be vigilant).

Thanks for listening, I know it's been a long and not entirely painless experience :-).

  • Up-voted because of your "End of the Internet page". I'm not really convinced that this wiki aspect of SO will work, though it's a nice idea in theory.
    – RoadWarrior
    Dec 18, 2008 at 10:09
  • Ok, I can't see if someone looked, but a better answer obviously exists (it's a technical question) and is newer. For subjective questions I don't give a ... if one is "accepted". For a REAL question on how to do something, we really need to have an accepted answer, because one actually exists.
    – Andrew
    Jan 7, 2009 at 19:34
  • @Andrew, see my comments on "accepted" vs. "correct". Accepted is totally under the power of the original questioner. Correct is controlled by the wider SO community (and even that may be wrong). Those two both appear at the list top so should be easily findable.
    – user14860
    Jan 8, 2009 at 0:05
  • @Andrew (cont): if it really irks you, just add the correct info to the accepted answer.
    – user14860
    Jan 8, 2009 at 0:08

Better answers float to the top. The answer which would be chosen by the original poster may be the top answer, so it wouldn't really matter. Stackoverflow is about getting questions answered, not reputation/e-peen (though that is a secondary objective, and a primary for some people =]).


The ability to select an answer could be given to the community (those with a certain amount of reputation). However there could be a rule in place that only allows a user to assign an answer to a question that doesn't belong to them if it has inactive for a period of time.

  • Put this on uservoice I'd love to give it a few votes there!
    – Unkwntech
    Dec 18, 2008 at 2:37
  • I'd rather the system accept the highest vote (i.e., the entire SO community) after inactivity than leave it up to a single person.
    – user14860
    Dec 18, 2008 at 2:50

The Uservoice entry for Remind to "accept answer" in user's profile has about 600 votes and is ranked 3rd at the moment. On the other hand, it hasn't got any "we're going to do it" annotation yet.

  • I think it actually exists, since I see (or rather, saw) it for this question.
    – Andrew
    Jan 7, 2009 at 19:37
  • See, you have no objective basis for accepting a 2-point answer over one with 19 points. But you did anyway. Why? Because it best agrees with your opinion? Then you should just post "Do you agree with my opinion that ...." and accept the first answer that says "Yes".
    – le dorfier
    Jan 7, 2009 at 19:41
  • Yes, the reminder does now appear, and has caused some angst because there isn't a way to make it disappear other than choosing an answer. When answered in mid-December, the comments were accurate. On the whole, I like the feature, but I always chose an answer for my questions anyway. Jan 8, 2009 at 7:59
  • You can also now select your own answer as the best - after a period (48 hours), and with no rep gain. Jan 8, 2009 at 8:00

I think they should let moderators mark questions as being the answer if a certain amount of time has passed.

  1. Question must be at least 30 days old
  2. User Must have rep of 3K (essentially a moderator)

The caveat being that the originating poster can still unmark the question if they choose to do so.

  • Who is a moderator? What reputation threshold would you use?
    – Jason Jackson
    Dec 18, 2008 at 2:18
  • @[Jason Jackson]: I would recommend 100 points higher than Jon Skeet's, or Infinity, which will eventually be the same thing. ;-) Dec 18, 2008 at 2:22
  • @Lowe, You mean "which are the same thing."
    – strager
    Dec 18, 2008 at 2:25
  • This is a good idea though. I think the rep threshold should be high though. Doing nothing would be easier though.
    – jjnguy
    Dec 18, 2008 at 2:27
  • What would that do, really? In the absence of an accepted answer, the one the community likes most will be on top, and that's almost certainly good enough. The accepted answer should be the one that worked for the questioner, and how is a moderator to know that any of them did? Aug 31, 2009 at 21:01

I think the more relevant point is that these question stay in the "unanswered" list, even though they are answered.

Maybe we need an "Unresponded to" list so we can see which questions have no responses at all?

  • Just change the Unanswered tab to Unaccepted, that's what it really means.
    – user14860
    Dec 18, 2008 at 2:51
  • Pax, "unanswered" means the question has no accepted answer and that no unaccepted answers have received any votes. As soon as anyone endorses an answer (by accepting it or by voting it up), the question is no longer considered unanswered. Dec 18, 2008 at 3:16

To worry about the points for accepted answers is making this site more about ego than sharing knowledge. Let it go. If you don't have time to answer, don't.

  • Hey, wait. This is a community wiki answer! How many points do I get for that?
    – rp
    Dec 18, 2008 at 2:54
  • Thats not the point. The site is set to give social standing to reputation, and not giving it violations basic fairness. The popular kits get big $$$ for stupid questions, but real work goes unrewarded, just like real life ;) I would rather there were no rep system at all.
    – Andrew
    Jan 7, 2009 at 19:30

In response to some of the answers (I am the OP)

  • I would prefer there were no such thing as "reputation" at all. It is basically for who can ask the most "what is your favorite programming book" questions, and is a form of fake competition. Lame lame lame.
  • Can I say again, reputation is really lame.
  • If rep isn't the point of the site, why not have "good/bad/ugly" only.

I would like to see questions answered so that I know that my input was useful, and so that others can see if an answer was actually successful. I know it isn't required, I can just test the answer myself, but the site is setup to say "this one worked" by having only ONE that gets accepted.

I answer questions to help someone, just like anywhere else. It isn't drive-by for me, but it sure is discouraging to take my time to answer someone and then have it ignored. I can see not updating after something worked (is accepted), but this is like going to a forum and seeing a bunch of answers, and the OP never says "thanks, #7 works!". Its like having a conversation where the other person never says anything.

The way experts exchange does it seems to work pretty well, actually. At least you know that no one has responded for months, someone arbitrates, things get accepted.

Also, many of these questions sit in the back, don't get many votes, etc. Having only 1 upvote on the correct answer doesn't instill confidence in that solution, and it's not like people seem to actually be upvoting things. I usually don't, even when I should.


I accept an answer if it's the right answer. I also accept an answer if it's not the complete right answer, if it at least put me on the right track to get the answer. In which case I'll explain in the comments, or in my own answer if it's long.

But if I get several answers and none solve the problem, how can I accept one of them? As far as I'm concerned, in that case, the question is still unanswered.

  • Absolutely, don't accept an answer if it doesn't help. But I gave an answer to a question and the guy who asked the question left a comment on my answer saying "That solved it, thanks!" but didn't accept the answer. I think that's more what the OP is talking about. Jan 7, 2009 at 19:46
  • Well, obviously that guy was a big meanie.
    – user142148
    Jan 8, 2009 at 22:25

Many - maybe most - questions simply don't have a correct answer. If anything, it should be harder to award the "Accept" prize.

Example: "What is your favorite ..."

This question, in fact, has no defensible way to choose which answer to accept. It will be interesting if OP tries to do it anyway.

You can't say any answer is more correct than another. It might be more verbose, or amusing and entertaining, or agree more with my point of view. But that's no reason to award it Rep points.

I personally think preference should be given to answers with authoritative references; but that's apparently not a popular point of view, having been healthily down-voted and ignored. But there needs to be some more objective way to describe answer "goodness". Otherwise it's a race to the bottom; which is kinda happening now.


That's what the upvote and downvotes are for, right?

Plus the best answer for the poster may not be for you or that with more experience and knowledge it might change, so I feel the answer tags are sometimes subjective too.

So I'll leave it at that.


Unfortunately I think SO deals with this poorly, sometimes you may not be certain that your own answer is what they needed given their situation. And if they never accept, or comment "not really what I was looking for", you may never know.

My suggestion would be to add a few features:

  1. Every week a question remains unanswered, the questioner should be alerted. They could either accept one, or mark the question as having no answer.

  2. After 3 months if no answer is accepted, and it has not been marked unanswered, the highest rated answer will be accepted.

  3. If another answer gets higher than that one, it will gain the mark.

  4. A new privilege should be added that lets a user change which answer is accepted. Accepting your own will not give you REP, as if it was your own question. They would also be able to modify manual acceptances, in the event that a much better answer was provided.

  5. If an answer is accepted, but the score drops below -10, it should lose the check. Users from search engines may think the check means it is okay, despite the score.

While you can use voting to promote answers, the check mark of approval goes a long way to helping people coming in off search engines. Those not familiar with stack may be led astray. Thus the ability for the community, or prominent community members, to change the acceptance is important. We can already edit the posts of others, given enough REP, so we should be able to change accepted answer.

  • 1
    How about if we allow the community to select answers? You know, like by voting on the ones they think are the best? May 10, 2012 at 5:30
  • That's the idea, however they also need the ability to make an answer "accepted" answer. Both to deal with those questions with no answer, and those that have an accepted answer that is bad/misleading/harmful.
    – zeel
    May 10, 2012 at 15:05
  • Why is that necessary? The point of the "accept" feature is for the asker to indicate which answer (s)he found most useful. That doesn't necessarily mean the most correct answer. That's judged by the community, with votes. May 11, 2012 at 4:55
  • Because being "accepted" makes an answer float to the top, and that check mark makes people look at that answer instead of others, even if it has a crummy score.
    – zeel
    May 11, 2012 at 10:34

You must log in to answer this question.

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