What is the etiquette for fixing up questions which have an incorrect answer that is the highest voted answer, or a wrong answer is accepted?

Related: Why do incorrect answers keep getting “accepted”?

Also, here is a list of problem questions compiled using the SO data dump.

Return to FAQ index

  • 10
    I think the "accepted answer" feature is stupid fullstop. You are asking the person, who is by definition unqualified, to judge which answer is the most accurate. Best just to sort the answers by votes and leave it at that.
    – U62
    Commented Jan 30, 2009 at 11:20
  • 10
    No, you are asking the person to indicate which answer helped them the most, and rewarding the person who explained it in a way they could understand. It isn't, and never was, meant to indicate correctness. If it fixed the problem, though, isn't it by definition correct?
    – Pollyanna
    Commented Jan 30, 2009 at 21:22
  • 5
    I think it should be replaced with a "thank you" feature, that allows you to say thank you to 1-5 people who contributed to solving your problem. then have some algorithm for a once off distribution of points and a pretty thank you icon
    – waffles
    Commented Jan 30, 2009 at 21:30
  • @Adam (a) many (perhaps most) questions aren't of the easily tested "how do I do X?" type; (b) if that is the idea as you state, then whatever validity it does have, it gives far too prominent a position to answers that are often not the best except to the person that asked the question.
    – U62
    Commented Jan 30, 2009 at 22:16
  • I notice that you did not select a best answer to this question -- were you worried that you would select the wrong one?
    – Jay Elston
    Commented Jul 14, 2011 at 3:21
  • The worst part is the default UI is usually a green tick, which is synonymous with correct.
    – Walf
    Commented Jul 27, 2016 at 2:06

6 Answers 6

  • Answer the question
  • Include in your answer explicit reasoning why the other answers are wrong. Don't be personal or hostile about it, but give enough evidence to convince someone coming to the site in the future
  • Downvote incorrect answers
  • Leave a comment on each incorrect answer, either just referring to your answer or briefly refuting the relevant point

I had a really good example of this once - parameter passing in C#, probably - but I can't remember exactly what it was.

I'd actually regard it as slightly irresponsible not to do each of these things, if you're confident of your answer and can explain it convincingly.

  • 1
    Whats your take on attaching a warning to the question (by editing the body)? So many old questions are abandoned and only show up on google.
    – waffles
    Commented Jan 29, 2009 at 23:15
  • 1
    Also, how do you go about fixing up a serious issue where an incorrect answer has 20 plus votes, there is no way to catch up, Im thinking SO needs a social networking thing where I would ping my mates and let them know about these things
    – waffles
    Commented Jan 29, 2009 at 23:17
  • Hmm... not sure, to be honest. I might edit the question (or even the answer) if the answer is actively dangerous. However, I'd make darned sure it was obvious what was going on, and it would be an utter last resource.
    – Jon Skeet
    Commented Jan 29, 2009 at 23:23
  • Is it fair to down vote an answer goes way too far. ie achieves an answer to the question but forces the implementer to do a dozen or more other things that (s)he did not request? Commented Feb 11, 2009 at 20:52
  • Assuming that there are other answers that achieve the goal in far less Commented Feb 11, 2009 at 20:53
  • 1
    @Rory: It depends on whether it's actively unhelpful. I tend to only downvote answers which are actually going to cause problems, e.g. misinformation. If a very general answer is presented, e.g. one which works for IEnumerable<T> instead of just the array in the question, I think that's okay.
    – Jon Skeet
    Commented Feb 11, 2009 at 21:00
  • 1
    Ok fair enough... I'm gonna go with the "actively unhelpful" definition.. Therefore no down vote in the case I'm thinking of since it is simply "a lot less helpful" but still "helpful". Thanks. Commented Feb 11, 2009 at 21:50

If the accepted answer is dangerous edit the question and chuck in a warning. For example see this.

If the accepted answer is wrong but not too dangerous, post a comment on the question.

  • Interestingly, your third example has been fixed now.
    – derobert
    Commented Feb 5, 2009 at 5:45
  • yerp, I think we should keep on posting examples here
    – waffles
    Commented Feb 5, 2009 at 6:02
  • 2
    Except now the question in the third example is even worse since it warns you about the accepted answer (which is now a different answer), which might lead readers to use the highest non-accepted answer, which is the answer the warning was meant to be referring to in the first place.
    – Alconja
    Commented Jul 29, 2009 at 5:01
  • I just fixed that
    – waffles
    Commented Jul 29, 2009 at 7:09

Perhaps upvotes need to be weighted based on the applicable tags of the votees. IE C# experts with lots of points should be able to up or down vote stronger than those who have not proven themselves to be experts in a field.

I believe that if something is clearly wrong, then you have a responsibility to correct it, but I would leave the original content in. Post evidence as to why it is wrong in the upvoted post, and give the postee a chance to respond to your comment.


With the new recent activity feature it should be easier to get the question askers attention as they will be alerted to comments and edits to their posts, but it might also be worth downvoting the question to make sure they get the message that something is wrong, and also to reduce it's vote relative to any similar or duplicate questions so that when they are returned in search results or the Related Questions when asking a new question the questions with more reliable answers are given prominence.


What is the potential for that taking place in an environment full of professional developers? I suppose a question could get a few upvotes, and selected as the "correct" answer, but it would seem that the vast majority of developers having a solid understanding would either suggest the correct answer, or downvote the false-positive.

In most cases it would seem sufficient to provide a better answer, and then reference your answer in the comments of the accepted answer. If your answer is better, it should begin taking on some weight.

  • Really - see this question : stackoverflow.com/questions/108403/… 20+ upvotes for an answer that is not safe for concurrent use
    – waffles
    Commented Jan 29, 2009 at 22:30
  • 3
    You suppose wrong. I have come across questions where the most up-voted answer was clearly and totally incorrect. This tends to be the case when the answer is not easy to verify. Recently, the number of such catastrophic answer seems to have gone down.
    – Anonymous
    Commented Jan 29, 2009 at 22:31
  • @stefan - when you come across those, do you downvote the answer and provide your own?
    – Sampson
    Commented Jan 29, 2009 at 22:36
  • Yes, but usually get down-voted :D
    – Anonymous
    Commented Jan 29, 2009 at 22:39
  • 1
    Remember that the person asking the question may not be the best judge of what answer is correct! The potential is not huge, but not tiny either. And it depends on how many people have true expertise in the area being asked. If it's a niche, perhaps a higher chance of insufficient oversight.
    – Eddie
    Commented Jan 29, 2009 at 22:41
  • Jonathan Sampson: so your argument is that accepted answers should work, based on... not sure what reasoning; therefore all the people who see accepted answers which are clearly wrong are wrong :D
    – Anonymous
    Commented Jan 29, 2009 at 22:45
  • @Eddie - You're absolutely right. Perhaps "Correct Answers" should be declared by votes from users with enough reputation in relevant tags. I'd be in favor of that.
    – Sampson
    Commented Jan 29, 2009 at 22:47
  • @stefan - My argument is based on the fact the author found it helpful enough to accept as an answer. Would you accept something that didn't work? I understand that isn't going to always render true answers, but when dealing with people you can't ask for much more unless the system is changed.
    – Sampson
    Commented Jan 29, 2009 at 22:48
  • The problem lies in the fact that just because something is A solution, doesn't mean it's a good solution. For example, I could ask the question "What's the best way to do this flow control?" If I mark the 'use a goto' as the answer, it technically solves the question but there's better ways.
    – Robert P
    Commented Jan 29, 2009 at 23:08
  • I wouldn't; but there are people who do not know enough about their problem to tell the difference between good or bad. Also, the answer might not be easily testable. The problem is that you're insisting that there is nothing wrong with the "system", when there is proof of the opposite.
    – Anonymous
    Commented Jan 29, 2009 at 23:09
  • @stefan - I'm not saying the system is working. I upvoted this question because I agree - users are stupid. I will eventually accept an answer in the future that isn't "the best" answer possible. It would be nice if "correct" answers could be voted by people with high enough reputation.
    – Sampson
    Commented Jan 29, 2009 at 23:14

One recent subtle example is this question: What is a plain English explanation of "Big O" notation?

Currently, all the current answers except for starblue's are technically incorrect. Granted, the asker probably didn't ask the right question and probably those answers really helped him better understand that stuff.

  • 1
    looks like this puppy is now sorted
    – waffles
    Commented Feb 5, 2009 at 6:03

You must log in to answer this question.

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