What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 130 Stack Exchange communities.

You probably get sick of hearing this suggestion, but...(a quick search didn't reveal a similar thread).

In instances like How to add an image to a JPanel? it seems that there is a "community preferred answer" (which, we may be a "better" answer, since it satisfied the answer for many more people--see also the long list of examples, some of them seem to have obvious "better" answers, which are not listed first).
However, when users run into that site from google, the answer they are "first presented with" is not that one, but instead the one originally accepted by the asker. This seems logical at first, but it causes grief to users because they naturally see the first answer, it has some votes so they implement it, only to be surprised that there is another, better answer just below.

There are other instances, for example, where the accepted answer did once work, but has not become out dated, so is no longer relevant: How to mixin and call link_to from controller in Rails?

Feature request: if some answers have many more votes than even the accepted answer, then set the answer order to be.

  • one with most votes
  • accepted answer
  • one with second-most votes etc. etc.

Other possible options:

  • If there are answer with higher votes than the accepted, but the accepted is shown first, automatically add a little text comment near the accepted answer of "note, there are other answers that are as least as popular as this one which may be of use to you." to warn them?
  • If a non accepted answer has 20+ votes and is "at least double" the accepted answer, place it first (does anybody know what the current algorithm is? I sometimes run into posts where the accepted answer is not listed first, but sometimes it is? It seems odd.)
  • Motivate askers to go back and "choose" a different answer, later, somehow, for example, display a message to users telling them "hey, there is another answer that seems more popular than the one you accepted, why don't you go and select it instead!" or something ro motivate...
  • do something to encourage users to "post mortem" switch their accepted answer, for instance a message "your answer appears to be less useful, would you consider accepting this other one that is more?" if it begins to appear to be out of balance...or some other means of encouraging them.

To me, "upvoting" an answer means "worked for me" or "that's the right answer" ("it was useful, because this was what worked" in many cases), if that helps any..

Thoughts? Any other options that would work here? May

Related: Why do incorrect answers keep getting "accepted"?

What is the etiquette for correcting old questions with incorrect answers? seems quite related

Other examples:

How can I get the current network interface throughput statistics on Linux/UNIX?

HTML Code for text checkbox ''

SQLite3::BusyException

How can I find the location of origin/master in git, and how do I change it?

Linker order - GCC

Determining if a variable is within range? (Ruby) (has since changed order to be "better", but used to show wrong order--anybody know what's going on here? Why does it show the "better" answer sometimes but not others? )

How do I clone a generic List in Java?

(Ruby) Getting Net::SMTP working with Gmail...?

Convert to/from DateTime and Time in Ruby (this one has comments on the question warning people not to trust the top answer!)

Why does git diff on Windows warn that the "terminal is not fully functional"?

http://stackoverflow.com/questions/14762/please-recommend-a-java-profiler

Regex (grep) for multi-line search needed (with this one, if you're not careful and scroll down, you'll be mislead).

How can I see the SQL that will be generated by a given ActiveRecord query in Ruby on Rails this one makes it look initially like SO doesn't have an answer...

Full URL with url_for in Rails a poor answer is first

How can I get Maven to stop attempting to check for updates for artifacts from a certain group from maven-central-repo?

http://stackoverflow.com/questions/1015525/why-use-buildr-instead-of-ant-or-maven

How to exclude all transitive dependencies of a Maven dependency

share|improve this question
    
Some of your "other possible options" seem less controversial, so you might want to post them as separate feature requests. Now, prods to get the original questioner to change their accepted answer may not work very well, since they may not be in the right "head space" to choose the best answer if it is much later, but giving those notifications are not as big a change as changing answer order, and you are right that there should be some solution. People going out of their way to vote for something other than the answer at the top with the big checkmark is a saying something. –  A.M. Jul 22 '13 at 18:53
    
Personally, I would upvote your 1st "other suggestion", at least for cases where the accepted answer takes up a lot of space and prevents readers from seeing the next one on the list without scrolling. –  A.M. Jul 22 '13 at 18:54

2 Answers 2

The thing is, if things worked the way you request, then someone would post a request like:

In instances like … it seems clear that there is an "asker vetted answer" (which, we may presume is a "best" answer). However, when users run into that site from google, the answer they are "first presented with" is not that one, but instead the one most upvoted by the community. This seems logical at first, but it causes grief to users because they naturally see the first answer, it has some votes so they implement it, only to be surprised that there is another, better answer just below.

Feature request: if the accepted answer has fewer votes than the most-upvoted answer, then set the answer order to be first the accepted answer, then the other answers ordered by score.

The reason why the accepted answer float to the top is that, supposedly, the asker has tested that answer and has verified that it works. Upvotes usually means that an answer looks good, but an answer that has actually been tested in the scenario it was meant for is really the best. This is especially true if the accepted answer came later and had less time to gather upvotes.

Of course there are cases where the asker accepts the first answer that came along. Pressure towards accept rate exacerbates this, but there are other features that lessen this effect, such as not being able to accept an answer in the first few minutes after the question was posted.

share|improve this answer
1  
I guess there's a tendency to, once you've accepted an answer, to not go back and accept a "different" answer. I was suggesting that if there is one answer that is overwhelmingly preferred, that it be shown at the top. It's just that several times I've dived into implementing some fix, then discovered "hey wait a minute, there's a community concensus here, and it's not the top answer". Maybe there could be a note added to the top, near the accepted answer "note: there is an answer that has received more votes than this one has" that might work well. –  rogerdpack Nov 20 '11 at 1:59
    
I think another thing that exacerbates the "effect" described is that if there are a lot of comments and a reasonably long answer, then the "community upvoted" answer is farther down, so harder to notice. I am tempted to write a "self" answer, and accept it, as a "live demo" of the problem, but I'm fairly sure you understand my initial concern, so not won't. –  rogerdpack Nov 20 '11 at 2:18
4  
Just found another example stackoverflow.com/questions/3498681/… I think the problem boils down to the "user accepts the first answer that works" then a better answer comes along, and it has no way of getting to the top. –  rogerdpack Nov 22 '11 at 14:09

It appears that, indeed, judging from the number of times I've found this odd behavior, that something should be done about it...

share|improve this answer
    
Or, it could be that you just need to consider the knowledge you already have about how the sorting works, rather than imagining that your inability/unwillingness to understand the sorting means there is a general problem. –  Andrew Barber Nov 26 '12 at 21:03
    
How many times this happens is irrelevant. What is relevant is whether this happens more often than the opposite (the accepted answer is better than the most upvoted answer). –  Gilles Nov 26 '12 at 21:35
2  
@Gilles but why is that relevant? The classical case for this is an abandonded question in which the OP has accepted the wrong answer, with another, correct answer having been voted to the top since. That's bad enough even if it doesn't happen in 3 million / 2 = 1.5 million questions. I think this feature request has some merit. –  Pëkka Nov 26 '12 at 22:58
    
@Pekka We have two proposals here: always put the accepted answer on top, or always put the highest-scoring answer on top. Either way, there will be cases where this is the wrong choice. Adopting this feature request means making the sorting on some questions better and the sorting on other questions worse. The total number of questions made better is not important. What's important is the relative magnitude of the two numbers. If the change means making 1000 questions better and 100 questions worse, it's good. If it means making 1000 questions better and 10000 questions worse, it's bad. –  Gilles Nov 27 '12 at 20:45
1  
@Pekka If the numbers are about the same, I'd be much more in favor of some way to override an accepted answer, or some blanket permission to thoroughly edit a generally-considered-wrong accepted answer. Sometimes the community votes are wrong, or the most-upvoted answer is out of date, and accepting an answer is the only way for the asker to make the answer he's tested come on top. –  Gilles Nov 27 '12 at 20:47
    
Ok @Gilles I added a few more proposals to the original question. Indeed there were some flaws with my original proposals, but that doesn't mean that addressing the problem doesn't have some merit, I suppose... –  rogerdpack Nov 28 '12 at 17:50
    
@AndrewBarber I did revise the question and add a question about current sorting order, perhaps you could explain/clarify it for me a bit? –  rogerdpack Nov 28 '12 at 17:51
    
I just mean that the number of times you have seen the behavior you describe does not necessarily mean that it's a problem that they are ordered like that; it could instead simply mean that you don't understand/agree with why they are sorted like that. –  Andrew Barber Nov 28 '12 at 17:55
    
@AndrewBarber yes that's right. Hence my question proposing slight modifications to the sort algorithm. I have also seen instances where the "accepted answer" is shown second, however. Perhaps you could explain what the current behavior is? (I don't agree with it, but I'm also not sure what exactly it is...) –  rogerdpack Nov 28 '12 at 18:03
    
The accepted answer is always on top, unless the accepted answer was also written by the person who asked the question, in which case simple vote sorting is applied. –  Andrew Barber Nov 28 '12 at 18:04
    
Ok, I've revised my original question, any feedbacks? –  rogerdpack Jan 17 '13 at 17:07

You must log in to answer this question.

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