20
$\begingroup$

SE is making the way accepted answers behave configurable per-site and is looking for input from our side what our preference is.

Currently, accepted answers - answers that the asker of the question has explicitly marked as answering their question - always are shown at the top of the list of answers when sorting by votes, regardless of whether there is an answer with a higher score. In the future, it seems we will be able to decide that instead it should just be sorted by its votes, ignoring its status as the accepted answer.

So, what are your opinions on this topic? Would you like the answer to be unpinned, and if so, why? Would you like the behaviour to stay as it is? Again - if so, why?

Note that in the linked meta question there is a deadline by Sept 19th for SE to decide what the default on all sites will be going forward, but this post is about your preference for the behaviour of accepted answers on physics.SE, not about your preference for the default for all sites.

The setting can be configured separately for the main site and our meta site.

$\endgroup$
14
  • 2
    $\begingroup$ Will the same setting transfer over to our meta site, or can we choose to make them different between the main and meta PSE sites? $\endgroup$
    – BioPhysicist
    Commented Sep 10, 2021 at 13:03
  • 3
    $\begingroup$ @BioPhysicist's point is quite relevant. The change would affect around 100 posts on this meta, and could end up impacting canonical policy threads. The clearest example is here. $\endgroup$
    – Emilio Pisanty
    Commented Sep 10, 2021 at 13:13
  • 3
    $\begingroup$ It is not clear to me whether this affects meta, but discussion here seems to indicate that it would. I've left a comment asking about this specifically. $\endgroup$
    – ACuriousMind Mod
    Commented Sep 10, 2021 at 14:13
  • $\begingroup$ For easy reference, here's a link to the MSO post with the results of the A/B test on unpinning. The test gathered data on the effect of unpinning on votes and on copying code from answers. $\endgroup$
    – PM 2Ring
    Commented Sep 10, 2021 at 17:19
  • 1
    $\begingroup$ @alephzero This unpinning initiative is just the tip of the iceberg of a very large project which is tackling the problem of old wrong answers. See meta.stackoverflow.com/q/405302/4014959 As Anita explains there, they are focusing on SO first because outdated answers are a big problem there (and because it's the busiest site of the network), but they hope to develop things that will be useful across the whole network. $\endgroup$
    – PM 2Ring
    Commented Sep 13, 2021 at 15:27
  • $\begingroup$ Outdated answers aren't much of an issue here on Physics.SE, but they do occur, eg in answers on detection of the Higgs boson, gravitational waves, and neutron star mergers. Also in relation to the changes to the definitions of some SI units. $\endgroup$
    – PM 2Ring
    Commented Sep 13, 2021 at 15:32
  • 1
    $\begingroup$ As context that I think is lacking in some comments, on SO I think it's typically the case that the OP actually tests the answer that they ultimately accept. Obviously that doesn't always happen, but it's a very important use case for that site. Others coming to vote after aren't necessarily adding much to that if they are cruising through reading but not testing other answers. On this site, we probably expect fewer OP are or can validate the answers that they get, so that probably results in a different dynamic. Conversely, SO answers go out of date, where Physics answers don't as often. $\endgroup$
    – Brick
    Commented Sep 13, 2021 at 21:27
  • $\begingroup$ There's now an answer on MSE with a table of all the sites that are discussing this question. meta.stackexchange.com/a/369914/334566 Some of the verdicts have already been posted. $\endgroup$
    – PM 2Ring
    Commented Sep 18, 2021 at 8:52
  • 1
    $\begingroup$ As we currently have a majority in favour of unpinning, I have entered physics.SE's verdict as "Unpin" in the meta.SE post. $\endgroup$
    – ACuriousMind Mod
    Commented Sep 21, 2021 at 11:49
  • 1
    $\begingroup$ Unpinning has now been implemented. $\endgroup$
    – PM 2Ring
    Commented Oct 1, 2021 at 8:04
  • $\begingroup$ @ACuriousMind 21/17 is a close call, I'd suggest waiting longer. $\endgroup$
    – user1271772
    Commented Oct 3, 2021 at 22:20
  • 2
    $\begingroup$ @user1271772 1. At time of my edit the vote count was 19/12. 2. The pinning behaviour has already been changed. If the votes for the old behaviour end up accumulating more votes, we'll have to ask SE to change it again, there's no "waiting". $\endgroup$
    – ACuriousMind Mod
    Commented Oct 4, 2021 at 7:48
  • 2
    $\begingroup$ @ACuriousMind since you have a fairly strong opinion about it, perhaps you shouldn't have been the one deciding that 19/12 was a large enough margin to go ahead and change a network-wide tradition which has been one way for the last 13+ years. Also I see that you're including downvotes, which is okay, but I had to check the timelines to figure that out. $\endgroup$
    – user1271772
    Commented Oct 7, 2021 at 18:32
  • $\begingroup$ @user1271772 I didn't decide this based on the margin, but based on the timing - shortly afterwards, SE went through the meta post and changed the behaviour for all sites entered there. If we end up deciding we want to have pinning behaviour after all, we'll have to ask them to change it again, which is a bit more inconvenient, but of course still doable (but we shouldn't do that every week :P). Let's wait until the month of this post being featured runs out and see what the "final" result is. $\endgroup$
    – ACuriousMind Mod
    Commented Oct 8, 2021 at 8:42

6 Answers 6

Reset to default
22
$\begingroup$

On Physics SE the best answer is selected by the community. That, IMO, means the answer with the most votes. It is the answer of most use to the community and hence to Physics SE going forward.

For that reason I think it is valid to make the most up-voted answer the one at the top of the list. By logical extension, the order of appearance should be vote score. There is an argument for making the accepted answer at least the second one listed, even if it's vote score would not warrant that.

The accepted answer is to allow the asker to make a special reward to an answer they personally found most useful. However this answer need not be the most useful to the community and might not even be correct in terms of physics — I believe I have seen this happen, but cannot locate a specific example.

We should prioritize the answers that the community as a whole considers the most useful, not answers which may have more limited appeal.

$\endgroup$
4
  • 2
    $\begingroup$ As the site does not require professional qualifications of members answering, the sort of "landslide" effects which have often proven disastrous politically would infect scientific opinion unnecessarily, and the value of the "profile" information would be diminished. (I'm speaking as someone lacking professional qualifications, and as a resident of the United States, where unscientific and downright anti-scientific opinions have gone haywire--ironically, because of a development extremely scientifc, the engineering of the internet.) $\endgroup$
    – Edouard
    Commented Sep 11, 2021 at 13:19
  • $\begingroup$ @ZeroTheHero Agreed. It would require a completely different system (some kind of officially acknowledged expert audit ?) to guarantee correctness. Voting or acceptance cannot alone do that, as I'm sure we all understand. $\endgroup$
    – StephenG - Help Ukraine
    Commented Sep 11, 2021 at 14:27
  • 1
    $\begingroup$ @ZeroTheHero This assumes that the most upvoted answer is the one most preferred by the community. That's the intent, but it doesn't always work, because an answer's score is strongly correlated with how quickly the answer is posted. (Whether or not the community's preference reflects the answer's correctness is a separate issue, not the issue that I'm raising here.) $\endgroup$
    – Chiral Anomaly
    Commented Sep 11, 2021 at 19:54
  • 1
    $\begingroup$ @ChiralAnomaly It is true that an answer posted early will get early votes but this doesn’t prevent late answers from getting upvoted by the same users if a late answer is better than an early one. It probably doesn’t always catch up you’re right, but the most voted is very likely the preferred answer. I’m sure we can find counter examples…. but see my answer below. $\endgroup$
    – ZeroTheHero
    Commented Sep 11, 2021 at 21:01
21
$\begingroup$

Well, we already have an against pinning to the top answer, so I'll do for pinning to the top.

I personally like the idea of having some control over the ordering of answers in my questions. I can imagine many scenarios where this is actually a good thing:

  • In general I would say that the person who asked the question is probably the one who is most likely to really understand the topic. For example, a short and superficial answer may be easier to digest and therefore gather more votes than a long and detailed one. As the OP, I will probably spend more time and effort in going through the answers, and therefore my choice of "best" is better informed than that of the community.

  • Selecting a best answer by hand fixes issues introduced by the "time factor". An early simple answer is more likely to have more votes than a recent one, even if the new one is clearly superior. Specially in the case of HNQ posts.

  • In the same vein, selecting an answer helps counter the natural inertia of "most voted answer keeps getting more votes" even when it is not substantially superior to other answers.

  • When the question is somewhat controversial (or is disliked by the community for whatever reason), it is easy to game the system and post an answer that intentionally does not really address the core of the question, but instead dismisses it by either explaining why it is a bad question, or by focusing on an entirely different reinterpretation that will satisfy the casual reader. An actual answer to the OP will get fewer votes than the dismissive one.

  • Finally, the opinion of the community is not always the best.

Of course, I can also imagine many scenarios where this feature can be abused, but in my experience this has never been a real issue.

$\endgroup$
7
  • 5
    $\begingroup$ 2. There are many questions here asked by users why are entirely not qualified to "really understand the topic" - they're asked out of the idle curiosity of a layperson, not professionally motivated interest. While I welcome these questions, I don't think the claim that OP's choice of "best answer" will on average be better informed than that of the community is true. $\endgroup$
    – ACuriousMind Mod
    Commented Sep 10, 2021 at 7:28
  • $\begingroup$ @ACuriousMind I don't disagree. Still, I like to think that I am able to objectively select the best answer, so I like to have that option over my questions. They will not take away my guns, you hear me! $\endgroup$
    – AccidentalFourierTransform
    Commented Sep 10, 2021 at 7:43
  • 1
    $\begingroup$ @AccidentalFourierTransform The OP still can select what they think is the best answer no matter where it ends up in the ordering. $\endgroup$
    – BioPhysicist
    Commented Sep 10, 2021 at 9:50
  • $\begingroup$ @AccidentalFourierTransform Regarding your fourth point, people not answering the question and trying to game the system seems like it would be well-handled by our community moderation systems. Do you have any examples of this? $\endgroup$
    – Chris Mod
    Commented Sep 10, 2021 at 19:12
  • 7
    $\begingroup$ I'm for this option. Under the current system, the top-voted answer is second rather than first. This doesn't really seem like a major deprecation to me; it's not like it would be buried deep in a thread. And honestly, if you're only reading the first answer on any Stack then you're doing it wrong. $\endgroup$
    – Michael Seifert
    Commented Sep 14, 2021 at 2:11
  • $\begingroup$ @AccidentalFourierTransform Hi, may I steal some of your points for a community wiki on English Language Learners, please? $\endgroup$
    – Araucaria - Not here any more.
    Commented Oct 13, 2021 at 16:26
  • 1
    $\begingroup$ @Araucaria-Nothereanymore. Sure, no need to ask! $\endgroup$
    – AccidentalFourierTransform
    Commented Oct 14, 2021 at 19:16
13
$\begingroup$

Just to give some voice to the data: Nihar Karve created (and I helped refine) a SEDE query,

Questions with accepted answer pinned above highest scoring answer

which lists the threads affected by this potential change. Under the current dataset in SEDE, the change would affect 5505 threads, which seem to cover a decently thorough cross-section of the site's threads, at least in terms of score. I don't know how useful it is, but it's interesting to take a look at, if nothing else.

$\endgroup$
8
$\begingroup$

Personally I find the accepted answers to be more often helpful than the most upvoted question.

I think this is true often because the person who asked the question is often the person with the most interest in if the question is actually answered.

I think often wrong or not-fully-correct answers are upvoted because they seem right. Often answers that are quick and lazy that maybe only address a small part of the question are often upvoted. Communities are subject to herd-mentality, and I think often most upvoted questions often completely miss the point of a question that's been asked.

$\endgroup$
4
$\begingroup$

In my mind, the most upvoted answer should appear at the top. If this is not the accepted answer, the accepted answer should come in second.

At worse, this will encourage users (including the OP) to reflect as to why the accepted answer is not the most upvoted in situations where this occurs.

$\endgroup$
1
  • $\begingroup$ At this stage, Stack Exchange are just asking what we want as the default for the network: pinning or unpinning. Each site will be able to choose one of those two options. They aren't asking about other variations, although various suggestions have been made on the MSE & MSO posts. $\endgroup$
    – PM 2Ring
    Commented Sep 13, 2021 at 15:15
1
$\begingroup$

Suggestion 1: Why not present the answers in a random order?
It seems to me that the arguments in favor of pinning to the top the accepted answer or the most upvoted answer are essentially a question of personal convenience - so that one does not have to look through all of the answers...

... but this is precisely the problem with the most upvoted answers: there is often a snowball effect of the answer that is not totally wrong and which was posted rather early gets more and more upvotes, even though the later answers may be more complete, better research, and on occasion even proving that the most upvoted answer is incorrect.

Suggestion 2: Let users with sufficient reputation have the accepted answer pinned to the top
Again, the problem with pinning the accepted answers is that the OP author may be too unexperience to judge correctly the quality of the answer. On the other hand, reputed users are more likely to be capable of judging the good answer better than the average member of the community.

$\endgroup$
1
  • 1
    $\begingroup$ +1. Your suggestion #2 is what I suggested here. Unfortunately one specific user bludgeoned it in the comments which lead to a lack of support, but it hidden under the "net" score are actually 10 upvotes from the community despite all that bludgeoning. $\endgroup$
    – user1271772
    Commented Oct 7, 2021 at 18:26

You must log in to answer this question.

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