15

UPDATE:

Having gotten mixed feedback on this change across the network, SE has decided as a default to keep the "old way", but sites can have a discussion like this one to request the "new way":

Unpinning the accepted answer from the top of the list of answers

Thanks to everyone for the feedback. Seeing that some sites do not want to unpin the accepted answer, we decided to move forward with the status quo and not to change the default behaviour on existing sites. If you think unpinning the accepted answer on your site makes sense, please do the following:

(meta discussion, share with SE staff via status-review, log decision)

It doesn't seem like there's going to be consensus here in favor of a change, so it seems we'll stick with the default option of how things were. We can re-start the conversation in the future.

I updated Academia.SE's consensus to be "50/50" since the voting is quite even between the two options.


On Academia.SE, all other StackExchange sites, and until recently Stack Overflow, the answer that appears first is the accepted answer (if there is one), the one that the OP decides is to be marked "correct". It is followed by all the others in order of votes by default (users can also change to sort by Active or Oldest).

As a result of user feedback, especially about "accepted answers" that have become outdated, and a test that suggests users on Stack Overflow benefit from the top answer being the one with most votes rather than the one that is accepted, this behavior has now changed on Stack Overflow.

Also quoted from that Meta announcement:

We can change the way the engine sorts answers in site settings. We would like to hear from you all if it is something you want to see on your site. (Please let me acknowledge in advance that we will not be able to run a test on each site.)

Currently we are planning to move forward with one of two scenarios, based on your feedback:

  1. Unpin the accepted answer on all SE sites by default and pin it back on a few sites that ask us to do so.

  2. Keep the accepted answer pinned on all SE sites by default and unpin it on a few sites that ask us to do so.

Please let us know what you think will work best for your site! If you can discuss this question with your community it would be awesome. We are going to collect feedback before the end of September 19th.

So, what do y'all think? Whichever way the default ends up going, how would you like things to behave on this site?

17
  • 2
    Here's a SEDE query to see posts where this change would have a retroactive effect (has accepted answer; accepted answer is not the top-voted answer): data.stackexchange.com/academia/query/1458833/… This does not include any voting effects of having the accepted answer at the top, however.
    – Bryan Krause Mod
    Sep 16 at 20:53
  • What portion of questions have an accepted answer? Sep 16 at 21:51
  • My impression has been that askers either accept the first answer when there are no other answers yet or they do not accept at all. Does the data bear that out? Sep 16 at 21:52
  • @AnonymousPhysicist You could definitely write SEDE queries to answer those questions, feel free to share the answers if you do!
    – Bryan Krause Mod
    Sep 16 at 22:05
  • Er, maybe I should revise that to say "probably"; I'm not familiar enough to know if all those things would be queryable in particular the exact sequence.
    – Bryan Krause Mod
    Sep 16 at 22:15
  • @AnonymousPhysicist As of now, 16,079 questions out of 36,700 have an accepted answer, which is 43.81%.
    – GoodDeeds
    Sep 17 at 9:47
  • Note that users have some options about the order in which to list answers: oldest, active, votes. A UI change could let the individual choose whether to pin the accepted answer or not.
    – Buffy
    Sep 18 at 12:41
  • 2
    I also note that in the current system, the topmost voted answer is easy to find, since it is always (first or) second if the ordering is by votes. It comes immediately after any accepted answer.
    – Buffy
    Sep 18 at 12:43
  • 3
    And, I'll also guess/predict that if the change is made, then the fraction of questions with accepted answers will decline.
    – Buffy
    Sep 18 at 12:44
  • I did notice that at SO, if two answers have the same (max) votes and one is accepted, the accepted answer isn't necessarily first.
    – Buffy
    Sep 21 at 19:54
  • 1
    @Buffy Related feature request on Meta.
    – GoodDeeds
    Sep 21 at 20:39
  • When will a decision be taken on this?
    – GoodDeeds
    Oct 1 at 14:16
  • 1
    @GoodDeeds SE updated the post at meta.stackexchange.com/questions/369568/… - after seeing feedback from many sites on questions like this, SE has decided to keep the old status quo but sites can request a setting change if they have a meta discussion supporting the change. The voting has gone a bit back and forth but the status quo currently has more sum votes while the change has more upvotes (ignoring downvotes).
    – Bryan Krause Mod
    Oct 1 at 14:30
  • 1
    @GoodDeeds The standard for meta polls tends to be count the upvotes only, but I don't think 20 votes vs 18 votes is sufficient meta consensus for a change and especially not when downvotes are generally pushing the opposite way. We can certainly revisit this as site in the future and opinions may change as users get experience with different SE sites that do it differently.
    – Bryan Krause Mod
    Oct 1 at 14:33
  • 1
    @GoodDeeds Yeah, that's correct. I'm open to other interpretations of how we should do things and I haven't discussed this with the other mods, but it seems like the process SE has laid out favors keeping the default unless there is support for a change.
    – Bryan Krause Mod
    Oct 1 at 14:44
17

The top-voted answer should appear on top, even if a different answer has been accepted (this is the new behavior on Stack Overflow).

14
  • 5
    This is my preference, because OPs do not always make reasonable selections (and the community / mods cannot change those selections, even when clear-cut). In some rare but noteworthy cases, the accepted answer has a hugely negative score, but the big green check mark may give casual users the impression that this is the "correct" answer.
    – cag51 Mod
    Sep 16 at 19:52
  • 4
    Yeah, I think @cag51 hits it for why this change could make sense here despite it being rare for questions here to follow the SO pattern of outdated answers, as those typically involve technology changes that happen at lot faster than changes in academia happen. I've also noticed several occasions where OP accepts an answer that fits the "what OP wanted to hear, not what they needed to hear" pattern, and that subsequently this validating-but-possibly-dangerous answer gets a bunch of upvotes via the HNQ, probably because visitors found it on top, even if it had a low or negative score before.
    – Bryan Krause Mod
    Sep 16 at 20:11
  • 2
    I counter that the OP of a question should have, and feel like they have, some respect. Odd that it is mods that like this. I agree that some questions are asked to get a particular answer and the OP is a bit incensed when they don't. I think "dangerous" answers get downvotes mostly. But a lot of such questions get closed in any case. I call out a few of those and absorb a few down votes myself for it.
    – Buffy
    Sep 16 at 20:18
  • I also, IIRC, have some accepted answers with negative votes. Not hugely. It isn't only the OP that can "want" a specific answer. And some questions are poison in any case.
    – Buffy
    Sep 16 at 20:21
  • 3
    And, FWIW, some of my most down voted answers are the ones I'm proudest of. Some things need to be said, even if they aren't heard.
    – Buffy
    Sep 16 at 20:31
  • 4
    @cag51 As a counterpoint, sometimes an excellent answer for a question already in the HNQ is posted late, and an accept by the OP gives it attention which it would otherwise have not received due to being buried under several others. (e.g. ref: academia.meta.stackexchange.com/q/4414/68109)
    – GoodDeeds
    Sep 17 at 9:52
  • 1
    Agree that this happens occasionally, and I would (in generalities) support a "smarter" way of ranking the answers than just votes. But the current system relies entirely on OP's judiciousness, and I think it's pretty rare that OP selects a better answer than the voters.
    – cag51 Mod
    Sep 17 at 16:24
  • 2
    BTW, this is a good example of an accepted-answer that I think should be "unpinned."
    – cag51 Mod
    Sep 19 at 0:27
  • 3
    @cag51, one example shouldn't drive a policy.
    – Buffy
    Sep 19 at 20:17
  • 1
    As any perusal of a well-voted-on page will confirm (HNQs aside) highly up-voted answers are primarily EARLY answers. LATE answers, however, brilliant, accurate or helpful they are will never be highly up-voted. The accepted answer feature allows for a correction to this pitfall, allowing useful, relatively 'late'- arriving answers to be put at the top of the page where they can usefully inform readers. An excellent answer given 48 hours after a more or less immediately provided one is highly unlikely to ever catch up. This is more or less when the views start dropping off massively.
    – user96809
    Sep 25 at 21:43
  • 1
    There are many bad examples; that's just one I happened to come across recently. Also: in all the discussion of the "tyranny of the early upvoted answer," no one has mentioned the "tyranny of the early accepted answer."
    – cag51 Mod
    Sep 30 at 0:09
  • 1
    In any case, RE late answers being overlooked -- there is a solution in the works! The announcement is here: they are developing a trending sort to address precisely this issue. It seems like everyone will agree this is a Very Good Thing (assuming it is implemented well). Personally, I think this (and not putting OP's choice first) is the correct way to handle the issue (though unfortunately, it could be some time before trending sort rolls out).
    – cag51 Mod
    Sep 30 at 0:09
  • @cag51 'the "tyranny of the early accepted answer." ' <--- There would be a good reason for that The stats show very clearly that those are comparatively rare. A rare phenomenon cannot be said to be dominating the system. In contrast, later answers compete in an apartheid environment in which they are discriminated against by the system.
    – user96809
    Sep 30 at 9:12
  • 3
    @cag51 Well, as Buffy says, we can kiss the accepted answer feature goodbye. No one's going to scroll down to find that, assuming any new users even notice it's there.
    – user96809
    Sep 30 at 9:15
15

The accepted answer should appear on top (this is the status quo).

12
  • 2
    This is my clear preference. It matters little if there are only two or three short answers, but searching a long page for the accepted answer is wasting my time. Alternatively provide a "click" to the accepted answer somewhere at the top of the page. It is clear enough when the accepted answer doesn't have the most votes without any changes at all.
    – Buffy
    Sep 16 at 19:47
  • 4
    It is also immediately clear that the question has an accepted answer without looking elsewhere.
    – Buffy
    Sep 16 at 19:54
  • 4
    The strongest argument against this is that frequently the OP is not the best judge of the "best answer". This recent question is a very good example of this... the OP is clearly looking for a specific type of answer, despite the fact that the community overwhelmingly agrees that a different answer is more appropriate. Look at the top entries on this query for more examples... this isn't that uncommon.
    – eykanal Mod
    Sep 17 at 14:37
  • 2
    @eykanal, I disagree with you on your example "recent question". I don't find the top voted answer very compelling, and I think the OP was truly lost, not trolling for a specific answer.
    – Buffy
    Sep 17 at 19:02
  • 2
    If the accepted answer is to be "deprecated" by not listing it at the top, why bother marking it at all? It is nice to get 15 rep occasionally, and to be publicly acknowledged for it, but the system would then be saying something like "ignore all that". It is also possible that a clearly wrong answer (ethically) might be the top voted one. Quite a lot of people here have little respect for copyright, for example.
    – Buffy
    Sep 17 at 19:30
  • @Buffy It'd still have the checkmark identifying it as the OP's preferred answer. In that way OP still has a privileged vote, as only they can assign the checkmark, they just don't have as much control over what other people see first. Academia.SE is indeed a bit different from other SEs in that many of the questions here are much more personal than elsewhere, but if SE is meant to be a repository of Q&A where other people can benefit, I don't quite see the logic of the person originally posting the answer deciding what the next asker sees first, rather than the voters.
    – Bryan Krause Mod
    Sep 17 at 19:36
  • 2
    @Buffy Whether trolling or lost, they still are not the best judge. I appreciate your second comment quite a lot... on StackOverflow, the checkmark is simply a visual indicator that the OP found this to work for them. At Academia, the checkmark is indeed much less useful. The whole point of these sites is to tap the wisdom of the crowds... why place such a highlight on one person's thoughts, especially when that person is the one who professed ignorance at the outset?
    – eykanal Mod
    Sep 17 at 21:12
  • 1
    @eykanal The checkmark can indicate what worked even in Academia (at least for some questions). "why place such a highlight on one person's thoughts": because they need not just be thoughts. The OP is the one facing the situation they need help with. They are the one in the best position to declare what advice they received on the site helped them. That is very useful information that is not captured by voting. Votes would indicate what should be the best answer based on the collective experience of the voters in general, and accepts would indicate (contd.)
    – GoodDeeds
    Sep 17 at 21:19
  • 1
    @eykanal (contd.) the best answer that helped the OP the most for the specific situation. So, to me, it makes a lot of sense that the top voted answer and the accepted answer are in the top two positions among all the answers. I agree they may not be the best judge, but my perspective is that their view is also important. I would not mind having the accepted answer in the second place, but that's not a choice here.
    – GoodDeeds
    Sep 17 at 21:20
  • 1
    @GoodDeeds Briefly; I guess I disagree. IMO, the OP's successful (or unsuccessful) application of advice in their specific situation is likely not that relevant to future readers. There are always specific nuances that affect any given situation, but the general advice—which is what this site is for in the first place—is almost always still true.
    – eykanal Mod
    Sep 19 at 18:48
  • 4
    I personally very much like being able to see the contrast between "this is what the OP liked" and "this is what the community thought", which comes through very clearly when the accepted answer is on top. When the accepted answer floats to an arbitrary place in the list, that's much harder to see.
    – jakebeal
    Sep 21 at 10:41
  • 4
    @eykanal, I think that "likely not that relevant" is far too strong a statement. And for a few topics the "general advice" can be toxic. I don't see this as a basis to choose.
    – Buffy
    Sep 21 at 15:00
10

I urge readers to vote to pin the accepted answer at the top of the page. This feature does a lot of silent good work whilst the very few annoying instances where an OP picks a (seemingly) obviously wrong answer are very scarce indeed, however memorable these are (see Cag51's post and the information therein).

Despite the terms voting and upvoted, there is a severe problem in terms of how democratic the voting system is. Early posts get voted on a lot, later posts much, much less so. Around 48+ hours after a question is first posted, the number of views by active voting members drops off significantly.

Pinning the accepted answer allows relatively 'late'-arriving but excellent and helpful answers to be recognised and pinned to the top of the page where they currently benefit readers. Without this feature some of the best and most helpful information on the site will languish unseen underneath a list of earlier mediocre answer posts. This current system still retains the benefit of having the highest-voted answer directly beneath the selected one. The new one will see many existing good posts vanish into obscurity.

Another benefit of the current system is that the Original Poster is the only member who is routinely alerted to new answers, especially those that arrive weeks, months or years after the question is originally posted. Because of this, they are in by far the best position to curate their own question page, and, if appropriate, accept a late answer. Certainly, the slew of voters on the original few answers will not be notified and will not get the chance to vote anew on the full range of answers.

Lastly, the current system affords some respect and agency to people who ask questions on the site. Whilst there are always vaguely annoying members in every aspect of the daily life of every SE site, we don't allow this to destroy or make us abandon useful and helpful features of the site. Where the odd muddle-headed OP might select the wrong answer, this is rarely anything more than an annoyance, and a rare one. In contrast if we in essence lose the selected answer feature, users will lose the benefit of many excellent posts and the helpful information that they provide. The vast majority of people asking questions here are sensible adults fully capable of making appropriate decisions regarding selected answers.

Some SE sites, for instance SO, get thousands and thousands of views by active voting members. So, for example, the highest voted answer on SO has over 33,000 votes. On these sites a very high number of votes over a quite sustained period may be the best indicator of the accuracy and helpfulness of an answer. Here, however, this is not the case.

The voting system on Academia SE is a good thing. However, it is not perfect for many reasons, including those detailed above. The 'accepted answer' feature helps provide checks and balances within the system. In particular it defends against the unintended and unwelcome tyranny of the early upvoted answer. Just like a healthy democracy, where second chambers and the separation of the legislature, the judiciary and the police provide safety in the form of checks and balances, the same is true of the accepted answer feature in its current form. Vote to keep it!

3
  • 3
    I agree with this view the most. The dilemma here is whether SE wishes to prioritize answers that are most suited to the OP's particular situation, or answers which are more suitable for general readers (those who upvoted). I personally think that the OP's needs matter more than those of the community at large.
    – djohn
    Oct 1 at 0:42
  • 2
    Real example: I provided an answer that was accepted by the OP. Then, another person downvoted my answer and said that my advice could be safely ignored, and since then people visiting the page have simply made a cursory read of the answers, and most proceed to downvote my post. The problem is that both the other poster and subsequent voters misread the context around OP's post, who was clearly a person just starting their career. The advice provided by them, while perfectly suitable for someone who already has an established career, is in my view risky for the OP's situation.
    – djohn
    Oct 1 at 0:46
  • 2
    Since comments only allow for a few lines, I ultimately I did not wish to start a nitpicking battle about how the examples provided by the poster, when looked at in close detail, actually prove my point. I understood that the other poster certainly had good intentions, and I just let it stay as it is without letting it bother me, but it made me realize that the accepted answer should take priority, unless there is some clear violation that requires a moderator to act.
    – djohn
    Oct 1 at 0:50
7

Among the choices given, I like pinning the accepted answer best. But my far preferred solution would be a change in the UI so that the individual reader can choose, either per-site or overall in their profile.

I note that it is currently possible to order the answers various ways, of which "by votes" is the default. But a checkbox to pin (or unpin) the accepted answer (or not) along with choice on the ordering would be simple enough to implement.

Note that with pinning the accepted answer, the top voted answer comes next by default, so it is trivial to find.

2
  • I'd say a box very specifically to unpin the accepted answer would work. Otherwise new users, who many not realise for weeks or months that they can order the answers, won't have a good idea of what the difference is. There's lots to lose by not seeing the accepted answer. However, all you face by having it at the top if you don't want it is mild irritation.
    – user96809
    Oct 4 at 13:37
  • @Araucaria, yes, that would probably be a better design.
    – Buffy
    Oct 4 at 13:43
6

Accepted answer should come first, because it is the one that answered the question. The person asking the question knows what kind of answer they need better than anyone else, so they should make that judgment call.

Here's an example of a question I asked on the Politics.SE where the top-voted answer turned out to answer something that I didn't mean to ask. I edited the question to clarify that, but the top-voted answer remains the top-voted answer, and it should logically not appear at the top.

1
  • 1
    Exactly. Especially with the more controversial questions, answers sometimes tend to be opinionated or differing from what was asked in the first place.
    – Sursula
    Oct 1 at 7:04
6

This post does not express a view either way, but provides some data (using queries from this Politics.SE post, which I reran for Academia with various modifications).

Findings:

  • We have 10.0K posts with 2+ answers and an accepted answer
  • Of these, there are 1,080 posts (10.8%) where the accepted answer has 3+ fewer votes than the top answer (on most questions, a clear difference).
  • For 93 of these 1,080 posts (8.6%), the accepted answer was written at least 3 days after the top answer (and so the discrepency could be attributed to an excellent late answer)
  • 89 of these 1,080 questions (8.2%) are closed
  • 19 of these 1,080 accepted answers (1.8%) have a negative score (and there is at least one answer with a nonnegative score).
3
  • 1
    At least one of those 19 is mine, actually. ;-)
    – Buffy
    Sep 19 at 15:07
  • 2
    I think the typical asker sticks with a question much less than three days. Sep 22 at 18:16
  • 1
    This info is not helpful for posts where there are 2 answers. These are both clearly visible! We need to know about the posts with 3/4+ answers, please! Any chance you provide that info? Also how does the 93 figure go when the late arriving answer was 2 days+? Excellent answers that arrive 48 hours after a popular immediate answer are extremely unlikely to catch up. (There is nothing 'democratic' about the voting in this respect. Early answers get voted on a lot, later answers less so!)
    – user96809
    Sep 25 at 21:48
2

I think the change is a mistake for Stack Overflow, because there answers are more clearly objectively right or wrong. I've seen many posts where an answer that simply doesn't work gets most votes, and the asker - who, unlike most voters, has actually tried the proposed solutions - has picked the answer that actually solves the problem.

Since Academia is a more subjective Stack, where answers more often represent cultural norms or personal experience, I think this is less of an issue. None-the-less, the asker is usually the one who is best placed to judge the answers and most invested in whether they are suitable or not.

It is natural to be irritated by those cases where a good quality, highly voted, answer appears below a low quality answer that is accepted, usually because it is the answer the OP was looking for from the start. However, these are - in my impression, at least - rare exceptions and it would be a mistake to change the system to account for the uncommon case rather than the common one.

(As an aside: my preferred system would be to show the top voted answer first if it beats the accepted one by some margin of votes, say 50% more or 100% more, and a minimum of 10 votes or something. I don't think that can be implemented by individual stacks though so it's by-the-by)

-3

My answer to this question has gathered a ton of down votes and is accepted. It is currently hovering in negative space. Perhaps it is a case study for the question here.

I stand by my answer there.

I answered late as I explained in the first sentence. My prediction seems to have been correct.

1
  • 3
    I would argue similarly, but in the other direction. I think this is a case where your answer sides with OP and tells them they are justified and doesn't sufficiently counsel them on seeing the circumstance from the perspective of others, and I think OP has accepted your advice at their own peril. This is not a case of a late answer not getting votes due to limited exposure; to the contrary, I believe there have been more downvotes than upvotes since it got the exposure of being listed first. OP came asking how to avoid offending others, and may have left with a recipe that will fail that goal
    – Bryan Krause Mod
    Oct 7 at 19:17

You must log in to answer this question.

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