I just ran this query on the Stack Overflow data dump. It highlights all answers that are accepted which happen to have another answer which is voted higher and have less than a -1 score. (wow thats a lot of waffles to chew on)

Anyway going through the worst offenders (posts are linkable) these answers really do not seem to deserve the #1 spot below the question.

Instead if an answer is marked correct AND it happens to have a -2 / -3 score I think it kind of makes sense to just sort it normally and have it pop up at the bottom of the list.

That way when people Google it they will not see the bad answers

I vaguely recall that this used to be the case though I am not sure.


  • 1
    note this is an edge case, barely 200 question affected on SO – waffles May 12 '10 at 11:46
  • Interesting idea, but given that the answer will have a negative score it should indicate that it's not the best answer - shouldn't it? – ChrisF May 12 '10 at 11:52
  • @ChrisF: You'd hope so, but when someone posts an incomplete question then what appears to be a poor answer could be good. Or it might just be the one that the questioner is best able to cope with (some people have odd blinkers). – Donal Fellows May 12 '10 at 12:35
  • 2
    +1 because the posts I'm reading via this query are highly entertaining. – John Rudy May 12 '10 at 13:11
  • Is there a reason why this accepted-but-downvoted answer does not appear right below the question, but at the bottom of the page? stackoverflow.com/questions/435620/how-are-css-frameworks-used – Jan Fabry May 12 '10 at 13:22
  • @Jan: Answered by the same user who asked. In those scenarios, the answer is not given top-of-page prominence. – John Rudy May 12 '10 at 13:25

Another answerer claims 'shenanigans', I say: Never ascribe to malice that which can be explained by incompetence.

The entire reason people ask questions is that they don't know the answer. They may pick an answer for a variety of reasons:

  1. It's first.
  2. It fixes their issue.
  3. It's last.
  4. A solar flare caused an atmospheric disturbance that moved the position of their mouse (Chaos theory, without the butterfly).

Does this mean the 'accepted answer' mechanism is faulty? No. It seems to work in the vast majority of cases, and where it doesn't can be explained by statistical outliers.

Should we do anything about it? No.

  • would you argue that the user accepted self answers should also always float to the top? – waffles May 13 '10 at 6:47
  • @waffles Why would I do that? It's against the argument I made. – George Stocker May 13 '10 at 10:32

Just because an answer has a negative score doesn't mean it won't solve the questioner's problem.

I've seen a few questions on SO where certain constraints in the question were flat out ignored by the answers, and the only viable one got downvoted because some of the question constraints resulted in using bad, or less-than-optimal, practices.

If the question is sufficiently stated and someone comes along from search looking to solve their own problem, the reader would be daft to not look over all, or most of, the answers before transplanting a solution into their own problem, regardless of the accepted answer's score (unless it's like >= +20).

As always, the accepted answer is the answer that best solves the questioner's problem.

  • I agree with Jon. The Perl community uses the word TIMTOWTDI =. there is more than one way to do it. I would add to Jon's answer that things change; a better answer might not exist when the questioner marked her/his choice--example, a new feature in vs2010 might provide a better solution than one could have had in vs2008. Also, someone like Jon might add a better answer after the questioner has already chosen an acceptable solution; given that in such a situation the questioner is less likely to revisit the answers, Jon's answer might not get the recognition it deserves. continued ... – gerryLowry May 12 '10 at 13:16
  • Finally, AFAIK, SO allows only one right answer to be marked. Other communities have other paradigms. forums.asp.net allows more than one right answer. another forum (VMware?) allows one "best" answer and up to three (?) second best answers. Ultimately, one will likely benefit from Jon's suggestion about looking over all/most of the answers to the issue. – gerryLowry May 12 '10 at 13:20

Scanning through your list I'd say it's pretty obvious to even the casual observer that there are shenanigans going on. The votes for the question and the accepted answer are more than enough to indicate that people are either being belligerent or goofy.

You must log in to answer this question.

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