-11

There has been a very large amount of confusion about self-accepted answers not rising to the top of all other answers. For a while, it wasn't possible self-accept answers. Then this blog post came out, and self-accepting was allowed, with three main restrictions:

  • You have to wait 48 hours to self-accept
  • No reputation is earned for anyone
  • It doesn't rise to the top

I think that for the last point, the problems outweigh the benefits. I think that self-accepted answers should rise to the top for a few reasons:

  • Accepted answers indicate the answer worked for the OP, no matter who posted it. Most people viewing a question want to see the answer that solved the problem best, which is what accepted answers are for.
  • It causes a lot of confusion and unnecessary bug reports. As mentioned above, there have been lots of bug reports about self accepted answers not rising to the top.
  • When the accepted answer isn't at the top, it appears that there is no accepted answer. It's really misleading and makes users think the asker hasn't solved their problem, and that the question needs more answers.
  • There are already a lot of things to encourage accepting other user's answers. For instance, you have to wait 2 days to self-accept so other users can add their own answers, nobody gains any reputation, self-accepts don't count towards badges, etc. If the OP is okay with all of that, then they probably have a good case for self-accepting and the answer should be most easily visible.

While this can be misused by making the OP's answer most visible, if it's a bad answer, it will attract more downvotes and the asker won't earn more reputation than they would otherwise. If it is a really good answer, then it's okay for it to be most visible and attract the most views.

  • A less intrusive change would be to change the color of the check mark to something else (maybe yellow), and/or having the tooltip identify it as a self accepted answer. – 1201ProgramAlarm Jan 6 at 3:53
  • 5
    A less intrusive change would be to not pin the accepted answer at all, but provide a quick link to it. All answers would be sorted by score, regardless of who wrote them. The accepted answer is one click away if you want it, but if you want to see what the community favors, just keep scrolling down. On most sites acceptance doesn't mean "this worked for me", because there isn't a testable problem to begin with. So the OP isn't in a better position to "certify" an answer. – Monica Cellio Jan 6 at 4:45
  • 1
    @MonicaCellio meta.stackexchange.com/questions/312597/… – Alex Jan 6 at 5:03
9

The biggest (from my point of view) problem with this request is that it can lead to abuse and becomes particularly problematic on meta.

Pinning has the side effect of getting the answer more attention. Even if the OP doesn't earn reputation by accepting the answer, they will still likely gain reputation because it's accepted and pinned and they're the one who controls that.

We encourage users to accept the best answer. Knowing that their own would get pinned and earn more votes might lead them to pick their own answer just for the visibility. It may be correct but not as clearly written as others, so assuming it would get downvotes isn't necessarily the case. Yes, this is always the case but with self-answers, the user has more incentive - the sweet, sweet 10 rep per answer upvote. As such, we require that the community recognizes that the answer is the best by voting it to that position.

Additionally, I've seen people write self-answers that are wrong or otherwise problematic. They generally get downvotes but they'd still be pinned. While it is possible to accept a bad answer written by someone else, it's at least less self-serving to accept it.

This is even more troublesome on Meta, where the content is more subjective. A user may have an idea for a feature or start a discussion and answer it. Accepting it makes it seem like the "correct" answer, which is confusing to people when they see that the answer scores lower than the others. Again, I recognize this is still an issue when the question is answered by someone else but that requires that someone else at least has that response. The pinning can be abused, even on child metas where there is no reputation.

For these reasons, I believe that self-answers should not be pinned.


To address your specific reasons:

  • Accepted answers indicate the answer worked for the OP, no matter who posted it. Most people viewing a question want to see the answer that solved the problem best, which is what accepted answers are for.

While many people think that the accepted answer is the best answer, we don't actually define it that way. The accept indicates the OP's preferred answer. Nothing more or less. That doesn't necessarily make it better in any way, it's merely an indication of preference; that of the OP specifically. In fact, it's confusing to many users who try to accept several answers because they assume it's the same as an upvote. If it truly "solved the problem best" we wouldn't pin accepted answers at all and we'd instead go by votes only.

  • It causes a lot of confusion and unnecessary bug reports. As mentioned above, there have been lots of bug reports about self accepted answers not rising to the top.

Yes, this does confuse people occasionally. I've been caught by it before myself. I'm not sure that confusion warrants this change. There are many confusing parts of the network. The solution is generally to do a better job explaining the situation than to change it how the system works. We could indicate why the question wasn't pinned to alleviate that.

  • When the accepted answer isn't at the top, it appears that there is no accepted answer. It's really misleading and makes users think the asker hasn't solved their problem, and that the question needs more answers.

Many people get to the question from the search results page, which indicates questions with accepted answers by filling in the answer box with solid green, so they arrive already knowing the question has an accepted answer. Additionally, there's no correlation between questions with unaccepted answers and lack of solutions.

Many people who ask questions never return to the site to see that they've received an answer and see if it's correct. On the other hand, on some sites, there's a problem with questions having no answers at all and yet being answered in the comments. The two can actually go hand-in-hand. A question that's answered in the comments often leads to the asker disappearing and never coming back, even when real answers come in later on.

I will agree that accepted answers often deter new answers on a post but even having an accepted answer doesn't mean that a question can't or shouldn't receive new answers. Questions can often use more answers. Many users are hasty in accepting the first answer they receive, regardless of how good it is.

  • I see what you mean, but most of the points you made could happen by accepting other people's answers as well as your own. The accepted answer isn't always the highest scoring one, so the community doesn't control the answer shown at the top even even when another user's answer is accepted. I just don't see how it would be any more likely for a user to self-accept their own bad answer than anyone else's. And even if they did, it would probably attract more downvotes than it would otherwise if it was pinned to the top. – Pikachu the Parenthesis Wizard Jan 6 at 0:22
  • You're assuming an answer that is outright bad or wrong. An answer can be okay, worthy of some upvotes. From what I understand, most people don't look past the first/accepted answer. If it works, they upvote and go on their way (assuming they vote). So, by accepting their own okay answer, they're likely to get more upvotes purely because it's pinned. Even if they get a downvote as some sort of "tsk tsk for choosing your own answer" (which we really don't want), they'll likely net more reputation than they lose since it takes five downvotes to match one upvote. – Catija Jan 6 at 0:30
  • 1
    I hate to say it but... some people are more interested in the game than the quality. If they can guarantee top billing by accepting their own mediocre but acceptable answer, they'll do that, even if there's much better answers at hand. We absolutely gamify the site but we have to balance the game with the content quality. If the self-answer is really great, it's going to earn the votes and bubble up just like a question without an accepted answer. – Catija Jan 6 at 0:32
  • If the answer is okay, though, it's fine for it to be accepted. It may need to be edited to get into good shape, but if it actually does solve the problem, I don't see why it shouldn't be accepted. Even if it's not the best answer, it's not always the best answer that get accepted anyway. And anyway, the highest voted answer will always show up second, so if the accepted answer doesn't solve the viewer's problem, they'll go to the answer considered best by the community. – Pikachu the Parenthesis Wizard Jan 6 at 0:34
  • if the accepted answer doesn't solve the viewer's problem - That's a big "if", though. We're talking about an okay answer that does solve the problem. The OP can prioritize their own content purely by accepting the answer. This is a huge problem. This is an action that I don't personally understand but I've seen enough people accept their own answers thinking it would pin them that I know it's something people do. – Catija Jan 6 at 0:49
  • I'm sure there are some times users will do that, but you can't assume everyone will abuse it. The OPs own answer may be the one that solved their problem the best. As you said, the answer at the top is the one most viewed, and in these cases the one that solved the problem wouldn't be most visible. I just don't think it's abused enough that the problems would outweigh the benefit of making the accepted answer more visible. – Pikachu the Parenthesis Wizard Jan 6 at 0:53
  • I understand what you're saying and disagree in this case. Part of the job of a feature request is to show why the request is an improvement. I'd argue that your first point is likely inaccurate in the case of self-answers. Your second point is, I think not really a concern. It takes little time to close new bug reports as duplicates of existing ones. Your third point is incorrect, as the search results show that the question has an accepted answer, and even if it does seem unaccepted, this is hardly unusual. Many very much solved questions do not have accepted answers. – Catija Jan 6 at 1:07
  • To address your edit: 1) See here: "It simply means that the author received an answer that worked for them personally." That means the accepted answer indicates the author has solved their problem with the accepted answer and that the asker does not need any more answers. 2) I agree this isn't that big of a deal 3) Not always. I very often find questions by searching on google search, which does not show if there is an accepted answer. – Pikachu the Parenthesis Wizard Jan 6 at 4:11
  • Re: 1 - The asker not needing more answers is irrelevant. While helping that one user is good in the short term, the main long-term goal is to have a repository of questions and answers that will help many people. This is why we generally avoid question types that only help the asker. – Catija Jan 6 at 4:16
  • In most cases, though, the answer that solved the OP's problem should be the one most easy to see. And I agree, the goal is to have a repository of questions and answers. I'm not against a question having multiple answers, but the one the OP accepted is most likely to be helpful to future visitors of the question. I understand this could be abused, but I believe it is much more beneficial to just show the accepted answer at the top, no matter what. – Pikachu the Parenthesis Wizard Jan 6 at 4:20
  • A lot of people will argue you on that. There are many people who don't think accepted answers should ever be pinned or we should have a way to ignore the accepted answer. I think there are even userscripts that resort purely by score. Accepted answer pinning is a much-debated topic. For active askers, accepting is a boon to questions in changing fields. If the answer actually changes, the OP can choose the newly correct new answer and pin it above the old, obsolete answer. With inactive OPS, the obsolete answer is permanently stuck. – Catija Jan 6 at 4:25
  • It's not really related that some people don't like accepted answers pinned at all. This request is to make all accepted answers appear at the top, rather than just some. There may be some people who don't like accepted answers being pinned, but that's a totally separate discussion. – Pikachu the Parenthesis Wizard Jan 6 at 4:28
  • You made the statement, "In most cases, though, the answer that solved the OP's problem should be the one most easy to see." That's what I'm responding to. You can assert that but it doesn't make it correct. The very fact that people don't want the accepted answer pinned refutes your statement completely. 547/-73... that's huge. meta.stackoverflow.com/questions/326095/… – Catija Jan 6 at 4:32
  • As I said, that's a totally different discussion. This is about making accepted answers consistent in where they appear, that's about unpinning them altogether. Askers can accept the wrong answers whether they self-answered or not. – Pikachu the Parenthesis Wizard Jan 6 at 4:37
  • 1
    Looking at "we don't actually define it that way", it seems like this kind of genericized fuzziness is a major part of the problem. The accept checkmark has essentially no actual semantic meaning, yet it has a huge effect on site functionality. Why is so substantial a feature still so ill-defined and subjective? That, I believe, is the main source of the confusion around accepted answer pinning. – Nathan Tuggy Jan 6 at 5:41
0

I don't think this is a good idea. Aside from the potential for deliberate abuse, I think it is unuseful in many cases.

As I mentioned in a specific-site feature request (which was followed up with a feature request here) in many instances the person asking the question is not the most well-equipped to determine the best answer. In fact, in many cases the person asking the question may specifically be bad at determining the best answer.

Sure, if the question was simply about some computer thing that didn't work then the questioner should be able to accurately tell us that a particular solution resolved the issue. But there are 170+ sites on the network, and many (if not most) of them cover subjects in which there is not a clear objective solution that can be definitively shown to work. At least for the handful of sites that I actively participate in, this is certainly the case.

If an answer can't be definitively proven to be correct, then the way to determine which answer is better is by the consensus of the other subject experts, which will be (loosely) represented by the votes. The person asking the question will often be the least likely to have a valid opinion, because the fact that they needed to ask the question in the first place is often indicative of the fact that they are decidedly not experts in the subject matter.

If the votes place one answer as significantly better than another, that should be the answer that appears at the top, because many readers never scroll past the top answer. They may end up just seeing the mediocre (or worse) answer posted by the questioner without ever seeing the better answer(s) posted by others.

Of course, many people will be honest enough to know that they can't make an accurate determination of which answer is the best; however, that won't necessarily stop them from selecting the answer that they prefer the most, and if they've posted they're own answer it's likely that they prefer it. And then there are the other people who will not be honest enough to realize that they are not in a position to judge the best answer, who actually think that their answer is the best (when the rest of the community disagrees).

In sum, I think it's bad enough that the person asking the question ever gets to choose the top answer; we certainly shouldn't exacerbate this by letting them choose their own answers too. At the very least, if this would be implemented, it should be on a per-site basis, so that the sites where the questioner's opinion is not very valuable would be able to opt out of this.

You must log in to answer this question.

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