5

I've been giving Let's move some negatively scored answers from the top spot (and the related posts it refers to) a little bit of thought recently, and I have a new proposal.

The problems with those previous ideas (according to discussion on the posts themselves, official responses, and/or my personal opinion, depending on the case) largely boiled down to two things: the new idea would increase complexity and the team is understandably reluctant to change a behavior that is extremely familiar/ingrained to even casual users. My proposal should be easier on both fronts.

Instead of changing the ordering algorithm, what about leaving accepted answers "pinned to top" and just providing an indication that an accepted answer is not the highest-score answer, where applicable?

For cases where an accepted answer is outscored, we could show, say, a yellow checkmark instead of the standard green, or a checkmark with an asterisk overlay, or something. I know that needs polishing (not great for the color-blind or on meta) but it makes the point for now.

This does introduce some new complexity, in that question pages would have to check relative scores, but it's not as bad as coming up with a new algorithm for score display. The bigger win is with user expectations. People will still see the post they expect, where they expect it, and the location of the icon they're already familiar with (and likely already using, too; subconsciously if not consciously) would provide a hint that a better answer might exist and it might be worth the user's time to scroll down.

It's been noted that accepted answers are their questions' highest-scoring answers in the vast majority of cases. In those cases, my proposal would change nothing.

  • Since the score of the other answers is not hidden, I am not sure that another mark would make any difference. Leaving the default sorting method (which is by votes), if the accepted answer is not the one with the higher score, the answer with the higher score will be the one right after the accepted answer. It's not that users have to look at every answer to understand that. – kiamlaluno May 1 '18 at 16:51
  • 4
    The scores of others answers may not be above the fold, though. One of the linked posts has an example where the accepted answer is hundreds of lines long. And it may also not be immediately obvious that an accepted answer is problematic, if it became obsolete over time instead of being horrible the moment it was posted. – SOLO May 1 '18 at 16:59
  • Fine, but you still have to check the score of the second answer, not every answer. The fact you need to scroll further down the page to find the second answer doesn't make the task more difficult, especially because the score is well visible and aligned to the left of the answer body. – kiamlaluno May 1 '18 at 17:09
  • 2
    That's where we disagree, then. A lot of this was based on my suspicion that some people (particularly less familiar users, I would guess) might decide to not bother scrolling down because the top answer is the best answer and it's no good. – SOLO May 1 '18 at 17:30
  • In that case, they could not even bother of the checkmark color, the negative score the accepted has (as it happens in some questions), or even the checkmark that marks the answer as accepted by the OP. What you propose doesn't resolve the problem of bad answers being accepted, which is different from answers that became obsolete, and which should not be considered bad simply for the fact they became obsolete. – kiamlaluno May 1 '18 at 18:17
  • Don't get me wrong: I agree it's bad that answers with a negative score are shown on top of the others simply because they are accepted. Even in the case of accepted answer with positive score, I still think it's bad that new, updated, answers with a higher score are shown after it. If the accepted answer would not be shown on the top of other answers, there would not be neither of those problems. – kiamlaluno May 1 '18 at 18:52
  • @kiamlaluno : I think it's bad that there's a presumption that an answer with upvotes is correct and an answer with downvotes is incorrect. I know stack exchange doesn't just deal in science, but don't forget the saying science is not a democracy. If science was conducted like stack exchange, there would be no scientific progress, because people would vote for their beloved epicycles and phlogiston et cetera. – John Duffield May 1 '18 at 19:29
  • 1
    @JohnDuffield Actually, the tooltip for the down-vote arrow says the answer is not helpful, and not correct answers aren't helpful. Then, you should not even presume that an accepted answer is correct too. How science work doesn't change how Stack Exchange works. They even have two different purposes. – kiamlaluno May 1 '18 at 19:44
  • @kiamlaluno : I don't presume, I check my facts. After many years of doing this, I like to think I can judge some answers well, and have some confidence that answer A is correct whilst answer B is incorrect. But then I see that answer B has more upvotes, perhaps because it's popscience, or perhaps because little Jimmy's friends upvoted it. If this happens a lot, stack exchange isn't building some kind of information base, it's building a _mis_information base. NB: I don't presume that an accepted answer is correct either. – John Duffield May 1 '18 at 20:27
  • @JohnDuffield How did you check users think that a higher score means this answer is the most correct? The tooltips say helpful for a reason. Without any contrary evidence, I like to think the answer with the higher score is also the most correct, but I don't pretend other users must think like me. – kiamlaluno May 1 '18 at 20:40
  • Anyway, this question is not about presuming that an higher score means the answer is correct but the fact the accepted answer is always shown on the top of other answers even when they are ordered basing on their score, and the accepted answer doesn't have the higher score, which the OP thinks is bad. I could think that an higher score means the user who wrote the answer has more aficionados, but that doesn't change the fact the accepted answer is always shown on top. – kiamlaluno May 1 '18 at 20:51
  • I really like the idea of a yellow check mark. As you say, if there's a green check mark there's a level of presumption that you don't need to look any further for other answers. While an answer marked as correct but having a negative score will likely result in scrolling down, if the answer marked as correct has a positive score, I think it's less likely that somebody would check the other answers . . . – Jason Bassford May 2 '18 at 5:04
-6

Instead of changing the ordering algorithm, what about leaving accepted answers "pinned to top" and just providing an indication that an accepted answer is not the highest-score answer, where applicable?

It wouldn't help much. People can easily page down and look at the other answers, which are ordered according to their votes.

For cases where an accepted answer is outscored, we could show, say, a yellow checkmark instead of the standard green, or a checkmark with an asterisk overlay, or something. I know that needs polishing (not great for the color-blind or on meta) but it makes the point for now.

The accepted answer is the one the questioner thought was best. He or she might be something of an expert. Meanwhile the answer with the most upvotes might be plumb wrong, and may have garnered upvotes from new inexperienced non-expert users. Just because an answer has a lot of upvotes doesn't make it the best answer. Your idea is kind of claiming that it does, and downgrading the accepted answer from a green tick to a yellow tick. I don't like that idea. It's kind of an insult to the user who asked the question and then accepted the answer s/he thought was best.

This does introduce some new complexity, in that question pages would have to check relative scores, but it's not as bad as coming up with a new algorithm for score display. The bigger win is with user expectations. People will still see the post they expect, where they expect it, and the location of the icon they're already familiar with (and likely already using, too; subconsciously if not consciously) would provide a hint that a better answer might exist and it might be worth the user's time to scroll down.

If users are genuinely interested they'll scroll down anyway. I don't think it's a good idea to insult the users who ask questions to pander to lazy users.

It's been noted that accepted answers are their questions' highest-scoring answers in the vast majority of cases. In those cases, my proposal would change nothing.

It doesn't really achieve much either. I don't like it I'm afraid. I wouldn't mind if your idea kept the green tick but added something else. For example there's plenty of dead white space on the left. A small "top scorer" message could be slotted in somewhere.

  • 4
    The concept of "higher score = better answer" is the basis of Stack Exchange answer ranking. The person who asked the question is more likely to not be an expert - that's heavily implied by asking the question at all, in the majority of sites (we can discount e.g. PPCG where the criteria for a better answer often rely on a single objective number). The answer with a significant number of upvotes, on technical and objective-based sites at least, and often on personal and subjective-leaning sites, is much more likely to be correct/better than not. Seems like you would optimise for a minority. – Nij May 1 '18 at 19:39
  • @Nij : I'm not proposing any optimisation. I'm saying leave the green acceptance tick as it is. I'm also reminding people that the answer with most upvotes might not be the best answer. Yes, the concept of "higher score = better answer" is the basis of the Stack Exchange answer ranking. But IMHO it's a dangerous concept, one that could lead to trivialisation wherein popularity counts for more than factuality. – John Duffield May 1 '18 at 20:12
  • We should also remind users that an accepted answer doesn't mean the user who asked the question thinks that answer is the best answer. It's just mean the user found that answer more helpful (as the tooltip for the checkbox for accepting an answer says). As a matter of fact, the best answer could not help the user who asked the question, who could find it more difficult to understand. – kiamlaluno May 1 '18 at 21:01
  • @kiamlaluno : that's fair enough. I wouldn't mind if every page said the accepted answer isn't necessarily the best or the most correct, and nor is the answer with most upvotes. But for me, it would have to be both. I don't like the way some people presume that the guy who accepted an answer is wrong because some other answer has more upvotes. However I don't think the stack exchange staff would like that. They seem to take the line that says the most upvoted answer is the best and the most correct, come what may, and that there are no issues of popscience/tactical voting/etc. – John Duffield May 1 '18 at 22:25
  • 1
    If they think that the user who accepted the answer is wrong simply because there is an answer with a higher score, they are wrong, or they simply forgot what accepting means. We always hope votes are objective, but there are cases they aren't, and that is true for accepting an answer too. I have seen some cases of users accepting an answer written from a co-worker, and I cannot avoid thinking they agreed doing so offline, just for the benefit of the co-worker account (who maybe also up-voted the question). – kiamlaluno May 1 '18 at 22:44
  • @kiamlaluno : it's an imperfect world, such is life. – John Duffield May 2 '18 at 6:50

You must log in to answer this question.

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