6

Currently the answers are sorted by votes and the accepted answer is docked at top. The intended design is to "get your answer quickly". This does work well in most cases.

Problem

If I have a problem and search for a solution on Stackoverflow I often find questions where the accepted answer is wrong or not the best solution. Mostly, in this case the community up voted a not accepted answer far more often than the accepted one but still the accepted answer remains on the top. If the accepted answer is long the user might not even see that there is a better answer without scrolling. This problem accrues mainly at old question which people found by Google.

Everybody needs to waste time, trying the wrong answer first.

Some examples:

Find more: https://stackoverflow.com/search?q=score%3A5+is%3Aanswer+isaccepted%3Ano

Also note that order is correct as soon as the accepted answer has negative votes like here: How can I change a file's extension using PHP?

Why do I think this is a problem and you may not?

After many comments, up and down votes I feel the need to explain a key point:

Perhaps we don't agree it is a problem at all? I don't think the problem is prevalent enough to make warrant special-casing.

If you visit Stackoverflow regularly, answer questions, helping people this is not a problem for you at all. Why? Because you are mainly dealing with new questions. Those questions have a low vote count and the voters do not have the same problem. There are voting based on theory what they think is right. The OP clearly does know it better and is very likely accepting the right answer.

But 90% of the people are using SO like this: They have problem, google it, find a matching question and try the answers in order. If an answer does work, they might up vote this answer. So after some time the most votes come from people with the same problem and the accepted answer might not be the one the community liked. Still, every user has to try the accepted answer first.

I don't know the hit count on Stackoverflow but if I look at some questions counter I guess each day thousand of people wasting time because of this.

Yes, the community can be wrong, too. But who is more likely to be wrong: The OP or 10 other users with the same problem?

Why is the wrong answer accepted?

  1. The question is abandoned or not maintained (even if you visit this page regularly you might not check for every old question if a new answer solves it better).
  2. The owner does not agree with the community.
  3. There are multiple solutions and the owner has a rare case.

History

The problem is not new and there have been made several (high voted) proposals to solve this:

So far, all offered solutions have been rejected while the problem still exists.

How should this be solved?

25
  • 22
    Perhaps we don't agree it is a problem at all? Commented Jan 8, 2013 at 10:20
  • You have other ways to sort the answers. There are little tabs at the top to sort by active/oldest/votes
    – Oded
    Commented Jan 8, 2013 at 10:20
  • 2
    @MartijnPieters It looks like it but don't you want to get your answer quickly? I thought that is the wanted design. If I have to scroll down because the wrong answer is accepted it's not quick anymore. Commented Jan 8, 2013 at 10:24
  • 7
    The community is not beyond error. Often a late answer that solves the problem better than an out of date highly voted answer will be accepted by the OP, putting it at the top.
    – user200500
    Commented Jan 8, 2013 at 10:48
  • 1
    @Asad The OP is not beyond error, too. No one is. The question is what is more likely to be right? If you have an accepted answer with 1 votes and 100 votes on the next question. What would you try first? Commented Jan 8, 2013 at 11:04
  • 8
    @PiTheNumber I would try the answer that the OP has marked as a working solution first. If this does not work, I would look at other answers. It is important to note that the OP is the one actually facing the problem, and is therefore better equipped to decide whether the answer solves it. The casual voter can only comment on whether the answer is theoretically sound.
    – user200500
    Commented Jan 8, 2013 at 11:07
  • 1
    @Asad And that is where you are wrong. You are all stackoverflow "power" users. 90% of the questions you face are new and you don't have the same problem. Yes, in this case I am wrong and the problem does not exist at all. But most user have a problem, google, look at some old questions and like to have a quick answer. They all have the same problem and up voting the correct answer. Commented Jan 8, 2013 at 11:15
  • 1
    @MartijnPieters I am sorry but I really don't know how you can not see that there is a problem. There are already high voted topics about this: +20, +54. You really like to try the wrong answer first? You really think the OP knows better than 10 other people? Commented Jan 8, 2013 at 11:37
  • 1
    @PiTheNumber: I don't think the problem is prevalent enough to make warrant special-casing. Commented Jan 8, 2013 at 11:50
  • 1
    A problem here is that in your second example, the accepted answer is entirely correct (and the one I upvoted a long time ago :-). Who is to decide that a +20 answer is not good enough?
    – Bo Persson
    Commented Jan 8, 2013 at 12:02
  • On the other hand, on this question I can see why we should change the accepted answer. :-)
    – Bo Persson
    Commented Jan 8, 2013 at 12:03
  • 6
    @PiTheNumber What people "like" is not relevant, considering the fact that one answer is objectively better than the other. This is simply a case study in the fact that the OP is the only one invested enough in the problem to bother reading answers besides the first correct one.
    – user200500
    Commented Jan 8, 2013 at 12:55
  • 1
    I kind of agree with you. The accepted answer indicates only that it helped the OP whereas the votes indicate that it was helpful for the wider community so why should the answer that is likely more helpful for the greater number of future viewers be placed second? Of course sometimes the community upvotes an "answer" that on greater scrutiny doesn't answer the question satisfactorily though. Commented Jan 8, 2013 at 14:03
  • 1
    @PiTheNumber - How many time will you keep adding examples in the question? Numbers of example should be 3-4 and maximum upto 5-6 and here you are adding example daily? Why?
    – Himanshu
    Commented Feb 8, 2013 at 9:56
  • 1
    @PiTheNumber If you really think this needs additional attention, consider a bounty instead. Or make your updates more substantial. It doesn't need to be bumped by each individual new example.
    – Bart
    Commented Feb 8, 2013 at 10:04

2 Answers 2

16

I think the idea behind showing the accepted answer first is clear if we think of Stack Overflow as a site that tries to help the specific person who posts the question.

If s/he found the problem solved, s/he accepts that answer and so a clear link between the question and answer is drawn.

The problem you describe comes when seeing it from a bigger point, that - of course - the questions and their answers also can help people with same or similar problems. This is where those up-votes come from.

I think after all it is a tough decision which way should be preferred. These people coming across the question by Google might have actually only a similar problem and not the exact same one, so this could lead to somehow wrong up-votes.

One simple solution would be to display a small notice just before all answers:

There is an answer with a higher voting available - jump to that one.
10

The checkmark just means that the original poster deemed that answer as "acceptable" to his or her problem.

The votes can do one of two things (if there is an answer with greater upvotes than the "accepted" answer):

  1. Show that the community believe that the other answer is more correct (whatever that means)
  2. That the accepted answer is also correct and should be shown in that light.

or even:

   3. The community could be completely wrong in upvoting one answer over the accepted answer,
       and it was a fluke.

But all in all --

   You can choose to view the answers by vote count (yes the accepted answer will be on top, but is it
   that hard to ignore it if you wanted to?)

3
  • @PiTheNumber There is a reason to the way I posted like I did. ^_^
    – Naftali
    Commented Jan 8, 2013 at 14:43
  • ok, sorry about that. Commented Jan 8, 2013 at 14:45
  • @PiTheNumber sok.
    – Naftali
    Commented Jan 8, 2013 at 14:46

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