31

After reviewing this question...

I'd like to make an argument for answers that have attained more votes than an accepted answer always be ranked above when answers are being sorted by votes (the default)...

Here is my example case...

Someone visiting this (older) question expecting a relevant answer would be assuming the big green check mark means that it is the correct answer today when in fact it's the correct answer when the question was originally asked. Allowing votes to override the asker's time bound response, it allows the best answers to remain at the top and evolve with our ever changing field without requiring the asker to maintain their questions.

This would be a step in the direction of the site's ethos in my opinion, trusting the community to determine the best answer to place at the top for prospective visitors.

8
  • I'm surprised this Question has such a low score and surprised that nobody has left a comment to explain why. This is a great Question. I found this because I was similarly wondering why SO wouldn't move the highest voted answer to the top accepted or not. Note that duplicate questions to this exist, which to me suggests that several others also agree that answers that are upvoted should appear first.
    – Sam
    Commented Mar 1, 2013 at 6:14
  • 3
    @Sam Votes are meta are different. It has a negative score not because it is a bad question, but because the crowd disagrees with the suggested course of action. Commented Mar 4, 2013 at 5:05
  • @dmckee Thanks for clarifying that. I'm on newb on Meta. :D
    – Sam
    Commented Mar 4, 2013 at 16:43
  • @dmckee So jondavidjohn is losing Rep because he asked a Great question that spawned a lot of debate, but they don't agree with his assertion? Thank stinks!
    – L_7337
    Commented Sep 25, 2013 at 14:09
  • @L_7337 Well, there has been a suggestion to change that usage, but frankly meta-rep is even less important that the same stuff on the main site. Commented Sep 25, 2013 at 14:16
  • 2
  • 1
    Thank you @MonicaCellio this question brings more pertinent and really interesting answers/comments. Commented Oct 17, 2015 at 11:26
  • I've just found this question: serverfault.com/questions/330127/… where the accepted answer doesn't answer the extracting a tar question, it answers the question of how to make the tar in the first place. Great, but in my case I'm not making the tar, I'm just consuming tars made somewhere else. I need to fix on extraction. The non-accepted answer has >200 votes to the accepted's 28. I think that an answer being accepted should be equivalent to around a doubling of votes for sorting purposes.
    – rjmunro
    Commented Nov 1, 2018 at 23:36

5 Answers 5

24

I think the accepted answer should not always be first because the community knows better. This page is cool because many people combine their knowledge and their opinions. In this case we put the opinion of one person over the community.

I think the order should be:

  • highest voted (if it has 5 votes more than the accepted answer!)
  • accepted answer
  • other answers

Does the OP know better?

No, he has no idea about it, that is why he is asking. Even if you OP knows it better then everybody else, this only works if the user rechecks the question for new/better solutions regularly. If he is not looking at every new answer or simply not logging in, it does not work at all.

Theory vs practice?

Some argue that only the OP tried what is the practical solution and what is only theory. This might be true for some of the 5 first votes from users trying to help. But if you find the question because you have the same problem, you up vote what practically works for you. Ask your self what you would like to try first: The accepted answer with 5 up votes or the other one with +20.

Is it correct because it works?

No. Just because it somehow works does not mean it is what you should do. For a lot of errors there a many different solutions. What worked for OP might not work for 90% of the others. Also think about best practice. The OP might prefer quick and dirty, the community does not.

For example I had a problem with SwiftMailer and my solution was to not use SwiftMailer at all. Completely right for me but no solution for others as I was told by many down votes. They were right!

Please also see my duplicated question on this topic and this answer with 11 up votes.

13
  • 2
    If voting were perfect this would make sense, but "you up vote what practical works for you" isn't true -- people upvote what sounds good or what they think is probably right, they very rarely actually try it first. The OP is theoretically actually attempting to use the answers to solve their problem Commented Jan 2, 2013 at 15:48
  • 2
    @MichaelMrozek As I said, this is true for the 5 first helpers but not for the 20 other people with the same problem. If the highest voted answer did not work for you, you can still use the accepted answer. But ask your self what you would like to try first: The accepted answer with 5 up votes or the other one with +20. Commented Jan 2, 2013 at 15:57
  • I looked at many questions with this problem and I considered to change my proposal to have 5 up votes more than the accepted answer. There are some low voted answers where the accepted answer is indeed better. Commented Jan 2, 2013 at 16:31
  • "Does the OP know better?" - well I disagree. He had no idea. But he applied answer and seen it works, and now he have pretty solid idea if it works, even if he has no clue why. Certainly more solid idea than voters who just stumbled upon a question. "Is it correct because it works?" - it doesn't matter, acceptance marks working solution, not most correct one. And again, what matters is that person most involved actually tested this one and found it working.
    – Mołot
    Commented Sep 25, 2013 at 13:16
  • 2
    @Mołot "voters who just stumbled upon a question" Why do they stumbled upon a question? Because they have the same problem. They are able to check the answer the same way the OP did. The whole problem here is who do you think is voting? You are talking about the 3 people checking the newly asked question. I am talking about the 100 people searching for problem they have. Commented Sep 25, 2013 at 13:53
  • @PiTheNumber if you would give the right to vote only to people who came to a question from search, I could agree with your comment above. Do you have any data that shows how high % of voters are from direct entry (probably search engine result) or search? Because you talk as if it was near 100% and I could bet it's not even near 75% (but that's my feeling, I know I don't have data).
    – Mołot
    Commented Sep 25, 2013 at 13:58
  • @Mołot At long term most of the votes are from those people. Who else would be looking up a question from a year ago? Commented Sep 25, 2013 at 14:02
  • @PiTheNumber anyone who would see it on "active" page? Again, do you have a data to support your claim of "most", or is it just a guess?
    – Mołot
    Commented Sep 25, 2013 at 14:04
  • 1
    @Mołot Why would a year old question be on active? You will only find it if you search for it. Commented Sep 25, 2013 at 14:09
  • @PiTheNumber Above statement is simply false. New answer or an edit, or even simple retag can bring question to front for all to see. Just now on the frontpage of Drupal Answers is an answered question from 2011 and that's a fact that can be easily tested at the moment of posting this comment ;)
    – Mołot
    Commented Sep 25, 2013 at 14:11
  • @Mołot well, and how do you get to this year old question to make an edit, retag or that ever? Sure you can stumble on it be many reasons like find a link somewhere. But do you read and vote any answer you just found by random? You need to have some kind of interest. Commented Sep 25, 2013 at 14:17
  • @PiTheNumber one person was looking and found it, many have stumbled on it without looking for it. Quite opposite to your claim that only those who are actively searching for it will see it.
    – Mołot
    Commented Sep 25, 2013 at 14:18
  • 2
    I agree with @PiTheNumber. The most pertinent answer have to be the top one and the one that concerns the most people. The later are not up randomly upvoting. Commented Oct 17, 2015 at 10:43
16

I think the accepted answer should always be first because

The asker has decided it works.

In your example (which is arguably a very narrow corner case), it would potentially make sense to have the most popular answer listed first.

However, 99.999999% of questions asked aren't about theory or hypotheticals. They are of the form:

I have this specific problem and/or error message from the below code. How can I fix this/get the desired outcome?

For those cases, what everyone else thinks arguably doesn't matter very much. What matters is what actually fixes the issue, which only the OP can tell us.

For your corner case, I think it should be sufficient to:

  • Post a new answer
  • Edit the accepted answer
  • Make a comment about the dated/inaccurate nature of the accepted answer
16
  • 1
    Nice, this is pretty much what I was going to say. I'd add only that this proposal would also not help in the many cases where late answers don't earn enough upvotes to overtake old accepted answers, which would make the old answers look even more correct.
    – Pops
    Commented Nov 7, 2011 at 16:20
  • @PopularDemand - Good point on late answers.
    – JNK
    Commented Nov 7, 2011 at 16:21
  • 1
    Again, I think the issue is that the decided solution to the said problem is that it is timestamped... In this field time (and technology) move very very quickly, especially for specific question/answers, I have happend upon many questions that the checked answer is in fact no longer relevant, and after failing, going back to see that there is a relevant answer a few scroll ticks down... Commented Nov 7, 2011 at 17:08
  • 11
    and I'm not sure that "it's correct (best) because it works" is really the mindset I'd want to propogate in relation to the field of programming. Commented Nov 7, 2011 at 17:09
  • 3
    @jondavidjohn - I still trust the person with the actual problem more than random people who read a question and have absolutely no vested interest in which answer is correct.
    – JNK
    Commented Nov 7, 2011 at 17:11
  • Interestingly, in the question referenced the OP changed the accepted answer to be the most current one. Guess this sorted itself out.. as it should.
    – NotMe
    Commented Jul 11, 2012 at 0:50
  • 1
    @JNK, so, what you suggest? When I see question with upvoted but not accepted answer, that I recognize as much better one and I do currently have this problem and vested interest, what exactly should I do to support better answer for other people who'd come later (remember, SO wants Q&A to be useful for future visitors, right?) I can't make it new accepted unless I open my own question, that must likely will be "duplicate". Just what to do? Commented Jul 11, 2012 at 13:08
  • @OlegV.Volkov Options include voting, adding a comment, editing the answer to make it better, etc. You have options.
    – JNK
    Commented Jul 11, 2012 at 13:13
  • 1
    Often someone else post a better solution than the currently accepted answer, and of course it's also working. For example in this one: stackoverflow.com/questions/2394609/… the best answer has 6 times more score than the accepted answer, and it's indeed the better solution.
    – Calmarius
    Commented Aug 30, 2013 at 10:49
  • 4
    @JNK - Consider (a fairly common case) where the accepted answer is really a workaround that a later answer actually Solves but for a problem that the OP no longer cares about and thus doesn't bother updating the post.
    – Ian Tegebo
    Commented Sep 12, 2013 at 16:39
  • @IanTegebo That's an edge case, not a reason to change the behavior of the site.
    – JNK
    Commented Sep 12, 2013 at 16:46
  • 1
    @JNK I agree that the default behavior should not be changed. My experience causes me to disagree about how much of edge case this is. While off-topic, I'd advocate for a preference setting.
    – Ian Tegebo
    Commented Sep 12, 2013 at 16:57
  • 2
    Should a negative-voted accepted answer get pushed down?
    – bjb568
    Commented Mar 10, 2014 at 22:54
  • What about two answers where one user pours their heart out and gives a great answer and the other who just gives a ho hum one? For instance, one answer says click here and there, while another says click here and there and this is why. Shouldn't people be rewarded for their effort?
    – user148298
    Commented Apr 20, 2016 at 16:14
  • 1
    You've got to understand that initially, the purpose of SO is to help the asker. OK great, they got their answer and moved on. But as time moves on and people refer to questions from google, then the question is also for them (i.e. it's meant to help followers as well) so I don't see anything wrong with making questions, over time, tuned to their utility (i.e. the higher upvoted answers...)
    – rogerdpack
    Commented Sep 7, 2018 at 20:10
5

I think there are very good reasons that the OP selected answer is not necessarily the best answer.

  • Technology changed since the question was first answered. The best answer today may no longer be the best answer today. I often am looking at questions from years ago. The original answers may have been the best for that time, but over time become obsolete.
  • Maybe a better answer is added long after the OP asked the question.

Often the OP does not come back to view newer answers since they have long ago moved on. Once asked, a question becomes part of the community. It is not solely the property of the OP. If the OP wants to remain active with their questions, that's great. But often they don't. I don't have time to review all my old questions. Do you?

And, since the selected answer may not be the best answer, I think SO should not give it the first answer position. The first answer may very well be the last one someone reads. It might be outdated -- even wrong.

How about this: order all answers by votes, and if the accepted is not the highest voted, then provide a link to it before the first (highest voted) answer.

I think this might age better. When the question is asked, the selected answer will typically be the top-voted. If that answer stays accurate over time, then it will probably stay top-voted. If that answer loses value over time, then it will fall down the list.

A very rough sketch:

question: blah blah blah

(linked to accepted answer)

top voted answer: blah blah blah

...

accepted answer: blah blah blah

...

1
  • …How about this: order all answers by votes…” The top voted answer may also be obsolete, it might have been posted contemporarily with the accepted answer. If the best answer today is an answer posted six months ago on a five-year-old question, it's more likely to be upvoted by a mere handful of individuals, so your suggestion will do nothing to highlight it. It's up to users' intelligence and acumen to decide which answer is more appropriate for their needs. Does everything need to be spoonfed nowadays? Commented Oct 28, 2018 at 8:24
4

I’ve written a user script so that you can at least change it for yourself: https://gist.github.com/Michael-1/5295ad559731c238a11a885328bedc5b

It is also listed on Stackapps.

2
  • The Stackapps link is broken :( Commented Jan 7, 2020 at 15:22
  • 1
    Yes, it seems to have been deleted from Stackapps. The Gist still works, though, for all the rebels out there. Commented Jan 8, 2020 at 12:16
0

I do agree on this matter. I have numerous of occasions where I mistakenly took the accepted answer while below there was a better answer having a higher vote. After a few times I tought myself to look further than the accepted answer.

In your question and within the answers there is however one important element missing. To my opinion there are two different types of answers:

  • Answers that come immediately after the question was posted.
  • Answers that come much later because a user happens to come across to it or because it was bumped to the top by a comment. It may also be possible that someone found the question, wasn't satisfied with the answer and found a much better answer (either elsewhere or by him-/herself) and wants to share it.

Basically an accepted answer may be best at that time but a much better answer may come even half a year later.

Update

One example I just encountered by accident that may be good to illustrate. Accepted answer has 50 upvotes and 5 downvotes. Another answer that came 2 hours later most likely after the answer was already accepted has 280 votes with 0 downvotes.

Another one is this one. Accepted answer has 37 upvotes and 9 downvotes. Other answer that came 2 years later has 111 upvotes and 0 downvotes. Non working answer is not at the top anymore.

Update 11-3-2014

Thankfully the SO team ignored the funny down votes by some people and finally sorted answers based on votes so people won't be misled anymore with wrong excepted answers. See Pass parameter to modal for an example.

11
  • 1
    I don't think a comment can bump a question to the top.
    – Marc-Andre
    Commented Sep 25, 2013 at 13:32
  • @Marc-Andre In any case practice shows many good answers come after the answer was already accepted. An accepted answer show what was good at that time. Other (better) answers may follow. See update for example. I have seen many more examples like these where a better answer came after the first answer was already accepted.
    – Michael
    Commented Sep 25, 2013 at 13:37
  • I know I was just stating that I think that a comment can't bump a question. Only an edit or a new answer can bump a question. But regarding the matter of this question, the notion of accepted answer is regarding the OP. You should see the accepted answer as the one that help the OP about his question.
    – Marc-Andre
    Commented Sep 25, 2013 at 13:51
  • @Marc-Andre true but OP didn't state about changing the accepted answer but sorting based on the amount of votes or at least giving the opportunity to do so by default imho.
    – Michael
    Commented Sep 25, 2013 at 13:52
  • Who are you to tell the question asker which answer helped them the most?
    – user7116
    Commented Sep 25, 2013 at 13:58
  • @user7116 still that is only according to that time. A later answer may have helped him better; changing an accepted answer is not possible after * minutes. On top of that, the accepted answer's green mark is the job showing what has helped the OP. I don't think changing the sorting or giving the opportunity to do so by default signifies that another answer has helped him better. It is only in the interest on the community since the amount of votes shows what has helped the community the best.
    – Michael
    Commented Sep 25, 2013 at 14:10
  • You can change the accepted answer at a later date.
    – user7116
    Commented Sep 25, 2013 at 14:14
  • @user7116 true, i was wrong about that. However you can see in my previous comment I was not aiming to change the OP's opinion but changing the sorting or giving the opportunity to do so. Even if the sorting would put the accepted answer down, everybody would still see it is the accepted answer.
    – Michael
    Commented Sep 25, 2013 at 14:20
  • @Michael: nobody has deleted this answer.
    – user7116
    Commented Sep 25, 2013 at 15:03
  • @user7116 hmmmmm i thought so because it was greyed out and possibly taken out of public view... I guess i was wrong then and just because of the downvotes :)
    – Michael
    Commented Sep 25, 2013 at 15:21
  • 2
    Regarding your edit, it was always like that. Self-accepted answer is an exception and is not affected by the "always on top" rule. Accepted answer written by different user will still always be above any other answer, there was no change whatsoever. Commented Mar 10, 2014 at 22:54

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