35
$\begingroup$

Stack Overflow changed the way they sort answers:

We no longer pin the accepted answer (with the green checkmark) to the top of the list of answers. By default, we now sort strictly by votes (descending order by highest score), and the accepted answer's order in the list is based on its score.

and it is asked on Meta Stack Exchange if we would like to have the accepted answer unpinned on our site:

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. ... We are going to collect feedback before the end of September 19th.

I am posting this here so that we can have a discussion of what is best for the Mathematics Stack Exchange site.

Update:

Based on the voting and the report on Meta Stack Exchange, the site setting has been adjusted to not pinning accepted answers on Math.SE anymore:

There's no need to do this on the sites that are on the list already; we have adjusted the site setting for them.

$\endgroup$
29
  • 37
    $\begingroup$ Judging by the edits, the real question is whether we are MSE, math.se, or Math.SE. $\endgroup$
    – Gerry Myerson
    Commented Sep 9, 2021 at 6:25
  • 20
    $\begingroup$ A soft option (at least in theory) would be to make the ✅ count for some $K\ge 1$ upvotes, say $K=5$. Then early on, the accepted answer will be up top, but if a newer answer can prove itself, it will beat the accepted answer. $\endgroup$
    – Calvin Khor
    Commented Sep 9, 2021 at 7:54
  • 1
    $\begingroup$ @GerryMyerson sir I think Math.SE is the right word $\endgroup$
    – jasmine
    Commented Sep 9, 2021 at 11:06
  • 8
    $\begingroup$ @CalvinKhor That is an interesting idea. I suggest that you follow up on Meta SE about it. However, the current state of affairs is that we have binary option: (1) leave things the way they are (accepted answers are unpinned, all answers are sorted by score), or (2) request that the SE overlords pin accepted answers. $\endgroup$
    – Xander Henderson Mod
    Commented Sep 9, 2021 at 12:45
  • 2
    $\begingroup$ @XanderHenderson: As I understand it, we have three options: (1) have accepted answers pinned to the top, (2) have answers sorted by votes, or (3) do nothing and accept the global default (and it is not decided yet which one that is). $\endgroup$
    – Martin R
    Commented Sep 9, 2021 at 12:53
  • 6
    $\begingroup$ If accepted answers are unpinned, then what's the point of the whole "accepted answer" concept? $\endgroup$
    – user1729
    Commented Sep 10, 2021 at 9:07
  • 1
    $\begingroup$ @PeterPhipps: As I understand it from meta.stackexchange.com/q/369568/196432 it is a “site setting” and I assume that our moderators can have it changed. I do not know what the proper procedure is, though, probably a new discussion on Meta. $\endgroup$
    – Martin R
    Commented Sep 10, 2021 at 9:16
  • 4
    $\begingroup$ "If accepted answers are unpinned, then what's the point of the whole "accepted answer" concept?" Per a comment I left below my own answer, accepting an answer is a way for a user to say "thank you" and award a small bounty. $\endgroup$
    – Xander Henderson Mod
    Commented Sep 10, 2021 at 13:46
  • 1
    $\begingroup$ @user1729 Do you have a similar objection to bounties? $\endgroup$
    – Xander Henderson Mod
    Commented Sep 10, 2021 at 15:16
  • 4
    $\begingroup$ Okay, so, again: the green check mark can be thought of exactly as a bounty is thought of. It is a way for one user (the original asker, in this case) to say "Your answer is the one I though was most useful". $\endgroup$
    – Xander Henderson Mod
    Commented Sep 10, 2021 at 16:26
  • 2
    $\begingroup$ @user1729 No, you are misinterpreting what I have written. I have not said that the green check is not useful for future readers. I have stated that I think vote totals are more relevant than the green check. My feeling is that the green check should be treated exactly as a bounty: it creates a visible flag on the post, which draws attention to it, and it rewards the answerer with a small amount of XP. Neither bounties nor the green check should be used (by default) for ordering the presentation of answers. $\endgroup$
    – Xander Henderson Mod
    Commented Sep 10, 2021 at 18:53
  • 4
    $\begingroup$ @CalvinKhor We literally honour those who provide significantly better answers than the accepted one: the Populist badge. That could be a guidance as to which answers should be placed above the accepted one. $\endgroup$
    – Hagen von Eitzen
    Commented Sep 12, 2021 at 14:38
  • 4
    $\begingroup$ @TheAmplitwist The status- tags are mostly used that the level of the SE overlords for tracking bugs, feature requests, etc. As this post is a "discussion" (not a bug report or feature request), I am not sure that the status-completed tag is entirely appropriate. We can still discuss this, after all. $\endgroup$
    – Xander Henderson Mod
    Commented Sep 29, 2021 at 13:50
  • 4
    $\begingroup$ It's funny that you accepted an answer. How ironic. $\endgroup$
    – Nike Dattani
    Commented Oct 3, 2021 at 22:17
  • 3
    $\begingroup$ @NikeDattani: Why do you think so? I accepted the answer because it has by far the most votes, and because the suggestion “sort by number of votes only” has been implemented for Math.SE by the site moderators. – This poll was not about whether to accept answers or not. $\endgroup$
    – Martin R
    Commented Oct 4, 2021 at 4:58

5 Answers 5

Reset to default
55
$\begingroup$

(Note: this is my opinion as a user; my moderator hat is off.)

I am in favor of ordering answers by number of votes only.

The green check is supposed to indicate that the author of the question found a particular answer more useful than others. It has been my observation that many question authors here (a) lack the mathematical maturity to judge a "good" answer, and (b) are inclined to give the green check to the first person who answers. As such, a lot of the accepted answers on Math SE are not as good as alternative answers.

Assuming that my anecdotal observations of the state of the site are correct, I think that it makes sense to order answers strictly by number of votes, and to ignore the green check.

$\endgroup$
39
  • 4
    $\begingroup$ (This is here as an answer rather than a comment so that it can attract both upvotes and downvotes.) $\endgroup$
    – Xander Henderson Mod
    Commented Sep 8, 2021 at 20:44
  • 8
    $\begingroup$ On a fence about this. Your description is accurate in many cases, and that sorting works better than the current. Against that there are cases like my second highest voted answer. It has more votes than the accepted answer, but it is building upon it. I just wanted to explain the observed phenomena a bit more, and the voters liked it better. Yet, the frequencies are probably on your side. My answer does refer to the accepted answer, so the readers should not be mislead too much, if their order is changed. $\endgroup$
    – Jyrki Lahtonen
    Commented Sep 8, 2021 at 21:02
  • 6
    $\begingroup$ @JyrkiLahtonen There is definitely a "rich get richer" problem, but if there is any consistent presentation of answers, that is going to be a problem. Either the accepted answer accumulates extra votes because it is on top, or the already most-upvoted answer accumulates more votes. Hard to say what is right. $\endgroup$
    – Xander Henderson Mod
    Commented Sep 9, 2021 at 1:13
  • 6
    $\begingroup$ @Jean-ClaudeArbaut I am in a camp that doesn't see a lot of value in the green check. I think that it might have some value on a site like SO, where the original asker can say "Yup, the code works now and does the thing I want it to do", but I think it has much less value here. I have come to understand the green check as a "bounty" which the original asker as the privilege of awarding. It is an extra "thank you" from the post author, and has little significance beyond that. $\endgroup$
    – Xander Henderson Mod
    Commented Sep 9, 2021 at 13:46
  • 4
    $\begingroup$ I agree. I would go further: voting is useful: the community decides which answer is best, in a rather democratic process (that's also debatable in maths, but it's better than the alternatives, on a web site). However, reputation points are not very useful: apart from the self esteem boost, they are only helpful in deciding if a user can be trusted or not. But actually reputation points tell nothing about the ability of the respondent on that particular subject (for instance, I may answer on numerical analysis or calculus, but I'm a noob in algebraic topology). The tag badges are more relevant $\endgroup$
    – Jean-Claude Arbaut
    Commented Sep 9, 2021 at 14:15
  • 5
    $\begingroup$ E.g. it has already been suggested that the green check be used as a tie-breaker. $\endgroup$
    – Xander Henderson Mod
    Commented Sep 10, 2021 at 22:14
  • 7
    $\begingroup$ @ContraKinta I would invite you to adopt a more civil tone. Calling others "stupid" is a violation of the SE Code of Conduct. $\endgroup$
    – Xander Henderson Mod
    Commented Sep 11, 2021 at 13:36
  • 10
    $\begingroup$ Responding to the substance of your comment: I do not think that the issue is that answerers are writing "mathematically mature answer that the questioner does not understand". Rather, I think that a lot of questions get (1) quick answers with a short computation that can be turned in as homework, and (2) longer answers which attempt to explain why something is true. The former is often accepted quickly because (1) it provides the asker with what they wanted (homework help), and (2) it came quickly. However, an answer which provides explanation is better for the site on the whole. $\endgroup$
    – Xander Henderson Mod
    Commented Sep 11, 2021 at 13:38
  • 4
    $\begingroup$ I am also somewhat shocked at the level of vitriol on display here. We are talking about a site setting which would change the order in which answers are presented. We are not talking about closing questions, deleting content, or suspending accounts. We are discussing whether or not the "accepted answer" should be displayed first, or if the most highly upvoted answer should be displayed first (assuming that these are not the same). I don't think that it is a huge deal in either direction---perhaps we can continue the conversation with civility? $\endgroup$
    – Xander Henderson Mod
    Commented Sep 11, 2021 at 13:43
  • 3
    $\begingroup$ Finally, because I reread the comment: the main goal of the SE network is not to provide the questioner with an answer. The main goal of the SE network is to connect questioners with answers (see here; search for the text "inverted pyramid" to get to the relevant bit). If an individual asker is helped, excellent! But helping future askers is even more important. Votes reflect how useful other readers have found an answer, including these important future askers. $\endgroup$
    – Xander Henderson Mod
    Commented Sep 11, 2021 at 13:49
  • 3
    $\begingroup$ First of all, I have not called anyone stupid. I said Accusing the questioner of lacking the mathematical maturity to judge a good answer is stupid. You are putting words in my mouth. Secondly, stuff like "the main goal of the SE network is not to provide the questioner with an answer. " is really off putting. Especially if the end result is an incentive to write answers that will rather please the mathematically mature audience of registered regulars rather than helping the actual questioner and all the registered/unregistered people on his or her level of mathematical maturity. $\endgroup$
    – ContraKinta
    Commented Sep 11, 2021 at 14:24
  • 3
    $\begingroup$ @XanderHenderson The concern with the voting system you propose, as I see from what Contrakinta is saying, is that if people see that questions are ordered by votes, they will not necessarily want to write for the OP, but rather for the general audience (on occasions where they know that a general audience will appreciate their answer more than an OP e.g. with a generalization). This should not lead to the preferred objective of an answer being to address a general audience rather than the OP itself. We have to preserve both objectives, and I don't think disregarding the green check is correct $\endgroup$
    – Sarvesh Ravichandran Iyer
    Commented Sep 13, 2021 at 5:40
  • 5
    $\begingroup$ In my opinion, the main goal of an answer is to be able to write both for the OP, as well as for the general population. This is at least what I strive to do. I get the feeling that this suggestion completely disregards OPs opinion of the answers they received as null and void : yes, maybe the majoritarian questioner behaviour is of the kind mentioned, but I don't think the situation is so bad that we must bin the value of the green check mark. It should count : I'd go with what Calvin says ideally , but in a forced situation i.e. the binary, I'd agree with you. $\endgroup$
    – Sarvesh Ravichandran Iyer
    Commented Sep 13, 2021 at 5:48
  • 3
    $\begingroup$ Given the binary option which has been presented to us, my feeling is that the default option should be to list answers only by total vote score. If you feel that the binary choice is not good enough, e.g. you want to give the green check the power of $n$ upvotes, then please have this discussion in the main meta thread, where your suggestion has a chance of being heard. $\endgroup$
    – Xander Henderson Mod
    Commented Sep 13, 2021 at 17:55
  • 5
    $\begingroup$ There are times where a higher voted answer is not as intuitive as the accepted answer. In that way, the second answer acts as a more rigorous expansion upon the first, which gets the idea across. However, I think on average, sorting by votes has more benefit for future questions. $\endgroup$
    – Mateen Ulhaq
    Commented Sep 16, 2021 at 3:21
30
$\begingroup$

I dislike this proposition, for one reason: For many questions, the highest voted answer is simply the oldest. This is especially true of questions that appear on the "Hot Network Questions" list. In such cases, often a really good answer will appear later and will never attain the same number of upvotes as earlier questions. The "accepted answer" in this case can highlight such really good answers, distinguishing them from a plethora of mediocre, earlier ones.

The argument for unpinning accepted answers is to address situations where the highest upvoted answer is better than the accepted one. However, my own experience since first starting to use stackexchange is that you learn quickly to always check the accepted and the most upvoted answers. On the other hand, in the reverse situation where an answer is accepted to highlight it among a throng of highly upvoted answers, if the accepted answer is unpinned, there remains no way to make such an answer more visible.

I have personally had several of my answers be accepted, and only well after acceptance did they reach the status of most upvoted. I think unpinning will cause a greater problem than it solves by effectively preventing this from happening.

$\endgroup$
8
  • 9
    $\begingroup$ Given how late I am in responding to this question, I wonder if this answer will suffer the same fate I describe, of not being visible for long enough to attain to the other answers... $\endgroup$
    – Yly
    Commented Sep 21, 2021 at 2:15
  • 6
    $\begingroup$ I think this answer makes a very valid point. Unfortunately, Xander has already said that "Mathematics SE is in favour of unpinning accepted answers", and so it does seem a little too late. This is why I would have preferred it if each individual user can choose how they the answers. $\endgroup$
    – Joe
    Commented Sep 21, 2021 at 19:41
  • 5
    $\begingroup$ This is a good point, that if you want to see both the top vote-getter and the accepted answer, then you can just look at the top two (by default) as it is now, but it won't be so easy to look for the accepted answer if we change it. It may be too late, but I think that I want to change my vote now! $\endgroup$
    – Toby Bartels
    Commented Sep 22, 2021 at 21:30
  • $\begingroup$ I have the same personal experience (i.e. late answer that gets accepted but few votes initially). However, I still agree with Xander's view because: (1) this is rare; (2) usually Math SE users who are not drive-by voters from the untrained HNQ crowd identify wrong answers that the asker cannot. Since there is a strong effect of biasing future voters toward the top answers, allowing a bad answer to stay on top by mere virtue of being 'accepted' is quite unacceptable. $\endgroup$
    – user21820
    Commented Sep 23, 2021 at 18:47
  • 1
    $\begingroup$ Hey, Maarten here, mod from Cryptography. I think this is a big problem with Stack Exchange and Q/A sites in general. But usually the accepted answer is also very old, so in that sense the reordering won't make much of a difference. It would only work if the org. poster changed the accept to a better / newer answer, but that's highly unlikely to happen in my experience. And a great accepted answer will still have the nice green mark beside it to distinguish it from the crowd. $\endgroup$
    – Maarten Bodewes
    Commented Oct 1, 2021 at 11:30
  • 2
    $\begingroup$ @MaartenBodewes That's what everyone says, but all the SEDE queries seem to show that in fact accepted answers are more likely to be later answers. The mathematics is against you! :) $\endgroup$
    – Araucaria - him
    Commented Oct 9, 2021 at 0:09
  • $\begingroup$ @Araucaria Interesting stats, I'll remember that. $\endgroup$
    – Maarten Bodewes
    Commented Oct 9, 2021 at 0:53
  • 2
    $\begingroup$ I totally agree to this answer. Thanks for posting another view to the question. I think there are many users (majority?) have the same opinion. I gave up for a long time to comprehend the voting- and listing-system here on MSE. $\endgroup$
    – callculus42
    Commented Oct 12, 2021 at 6:07
15
$\begingroup$

I think the best solution would be to allow users to choose between sorting strictly by the number of votes, and putting the accepted answer at the top. Right now, there are three ways of sorting the answers: active, oldest, and votes. I would like to add a fourth option that ignores whether an answer has been accepted or not. (What this option should be called is another question.) As before, the site should "remember" the option that you used last time.

$\endgroup$
4
  • 32
    $\begingroup$ The biggest impact of sorting on this site is how answers are presented to users who likely don't have accounts here, or who have very young accounts, and are not aware of the fact that posts can be sorted. For anyone who has already figured out that they can rearrange posts, I have no objection to a "strictly votes" vs "check, then votes" option, but I think that this is somewhat orthogonal to the real issue, which is about default presentation. $\endgroup$
    – Xander Henderson Mod
    Commented Sep 9, 2021 at 1:15
  • 1
    $\begingroup$ @XanderHenderson: I agree that the default presentation is an important issue that we need to discuss (personally, I haven't been convinced either way), but I also think that the point I'm making is relevant. Stack Overflow removed the option to put the accepted answer first, and I can't think of any reason why that would be better than just allowing users to choose how they want to sort posts. $\endgroup$
    – Joe
    Commented Sep 9, 2021 at 8:37
  • 2
    $\begingroup$ This doesn't answer the question as posed here. Math SE will have a choice of yes-pin or no-pin. Feedback on what other settings could be used belong on the Meta SE post, where they can be actioned or discarded, not here, where they just add noise to what could be a complicated enough discussion just in the yes/no part $\endgroup$
    – Nij
    Commented Sep 9, 2021 at 11:45
  • 9
    $\begingroup$ "Feedback on what other settings could be used belong on the Meta SE post" - Technically, the Meta Stack Exchange post also asks what the default behavior should be on the rest of the network besides SO: should accepted answers be unpinned network-wide, or not? ...It's just that many people have left feature requests there anyway. $\endgroup$
    – V2Blast
    Commented Sep 9, 2021 at 15:55
9
$\begingroup$

Yup I want accepted answers to be unpinned. However as I wrote my decision in meta here I would like that if 2 or more answers have same score out of which one is accepted then that accepted should be upper than the remaining other answers having same score.

Also it would be of immense pleasure if a user himself has the preference option in settings to choose whether in his question he want the accepted answer to be pinned or not.

$\endgroup$
-11
$\begingroup$

Tl;dr:

  • In Theory: Yes
  • In Practice: No

In Theory

In theory, as we know, posts should be voted according to their quality so that visitors can find the best ones easily.

So, when some answers are provided to a question, the best ones should be placed at the top of the answer list to help readers of the thread find more useful answers.

However, the questioner may accept some answer according to their need and understanding of the subject of the post; it is clear that such a choice does not necessarily indicate the best answer.

Thus, in theory, I agree that accepted answers should be unpinned.

In Practice

In practice, as we know well, poster are often voted, not their posts. Let us recall the following community misbehavior concerning the issue:

Questions are illigiitmately upvoted for reasons unrelated to the quality of a question. A very poor post answered by 5 separate users may very well get five upvotes. Friends upvote friends, admire certain users, etc.

(Although this a harsh but irrefutable fact, I think ignoring/evading it would not help us to find the best solutions for our community at all.)

Although such abuse may happen in acceptance of answers, its occurrence rate is likely much lower than that of post voting.

So, not to mislead people, especially novice users, it is strongly recommended not to implement the feature request until some people change their behavior.

$\endgroup$
6
  • 10
    $\begingroup$ -1: Quoting one random comment and call it "community norm" is ridiculous. $\endgroup$
    – Arctic Char
    Commented Sep 17, 2021 at 12:18
  • 6
    $\begingroup$ You are making some rather extraordinary claims here. Extraordinary claims require extraordinary evidence. What evidence do you have that (1) votes are based on who posted something, rather than the content of a post? and (2) that this is a "community norm"? $\endgroup$
    – Xander Henderson Mod
    Commented Sep 17, 2021 at 16:38
  • $\begingroup$ It is also interesting that you truncated the comment you quoted. The full quote is (the omitted portion is bolded): "Questions are illigiitmately upvoted for reasons unrelated to the quality of a question. A very poor post answered by 5 separate users may very well get five upvotes. Friends upvote friends, admire certain users, etc. To me those are abuses in voting far more often that retaliatory downvotes." $\endgroup$
    – Xander Henderson Mod
    Commented Sep 17, 2021 at 16:41
  • 4
    $\begingroup$ @XanderHenderson What evidence would satisfy you? If I showed you that my inactive posts received some downvotes immediately after my participation on Meta, if I showed you many high quality posts having much less scores than low quality ones, if I showed you many people complaining about the issue on Meta or in other places, ... , then would you satisfy, or you would say that the community judged the quality of the posts by their own opinions? Even, you ignore some heavily upvoted comment from some experienced user. If we want to evade unpleasant clear issues, we will harm ourselves. $\endgroup$
    – Later
    Commented Sep 18, 2021 at 10:36
  • 6
    $\begingroup$ The claim that you are making now ("Sometimes I get downvotes after participating in Meta") is very different from the claim made in your post, above ("The community norm is that votes are cast for individuals, not based on the merit of posts"). Conflating the two is akin to suggesting that armed robbery is a community norm in the US because as a convenience store in East LA was robbed last night. Votes which target individuals are not a community norm here, and are an abuse of the voting system. We try very hard to crack down on that kind of behaviour. $\endgroup$
    – Xander Henderson Mod
    Commented Sep 18, 2021 at 13:06
  • 1
    $\begingroup$ @XanderHenderson Yes, you are right. I confess that I was wrong; I should not have used that term. I edited the post. $\endgroup$
    – Later
    Commented Sep 18, 2021 at 14:06

You must log in to answer this question.

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