48

I often see complaints about highly voted answers (an indicator of community respect) appearing below an accepted answer which may have far fewer votes e.g. Shouldn't the answer with more votes be above the accepted answer?

I think the meaning of the accept check mark should remain as being:

This helped me [the question asker] the most

I think it works well for the question asker to have 15 points that they can use to reward the person that they feel helped them the most, and that the accepted answer is given the advantage of being read/considered first, which if it remains useful will often lead to more upvotes too. It's the same effect as why politicians want to appear first on a voting card.

I also think it works well that the asker can change their mind about the answer they accepted, and move the accept check mark (and the 15 points) to another answer.

However, what seems to be contentious is when an accepted answer with few votes (even net negative votes) remains above one or more answers with many votes in perpetuity.

As an answer to this question points out:

There are three aspects to this proposal:

  • accepted-answer stickiness
  • accepted-answer reputation
  • accepted-answer labeling

Stickiness I would like to see the accepted answer lose its stickiness to the top slot at some set time period after the position of the acceptance check mark was last set. I'll use 12 months purely as an example, and perhaps this could be settable per-site in the moderator tools, but my inclination is to have the same term network-wide for a consistency.

Reputation I would like to see the 15 points remain with the poster of the answer that was last accepted i.e. no change.

Labeling I would like to see the green check mark turn to amber 12 months after it was accepted as an indication that it was accepted more than 12 months ago, and that the answers are now ordered by votes alone.

Green implies more of a go ahead on trusting this answer (because it helped the question asker the most, and they said so relatively recently) than amber which implies to only proceed with caution (because acceptance was a while ago and has not been reviewed recently by the person who originally gave it).

Under this proposal:

  • The asker still gets the good feeling of awarding the Accept status and 15 points to the answer they like, and even to change their mind about which answer was best, forever (see below).
  • For 12 months after the accept check mark was last set, even if the community thinks there is a better answer, the accepted answer still stays at the top to give it a voting advantage, but then the stickiness disappears, and ordering becomes by votes alone.
  • The community can still see which answer was the last one that it's question asker set as accepted, because it still has a coloured check mark next to it, but its colour is now amber instead of green i.e. it retains its special status, forever.
  • When the asker of the question leaves the site, with a poor accepted answer occupying the top slot, the community may resent seeing the poor answer at the top, and the extra time it takes them to scroll past (or read) such answers (and sometimes long comment trails as well!), before learning that there is one or more that the community thinks to be more useful beneath it, but they will at least know that it won't be there forever.
  • If the question asker is still on the site after the accept check mark has turned to amber, then perhaps they can receive a notification of that happening, so that they can (with a single click) choose to re-instate the green check mark. This could be a great opportunity for them to reassess whether they still think the answer they check marked is the best answer for their question. Either way the 12 month clock starts again, with the 15 points following the acceptance check mark, just like it always has.

In other words, the asker owns their question and gets to say and reward the answer that helped them the most forever (if they are part of the site's community for that long!).

Remembering that self-answers do not get special ordering, I think a self-answer should only ever get an amber check mark (i.e. proceed with caution because "you do realize they gave it to themselves, don't you?").

As an aside, I toyed with the idea of adding red check marks for massively downvoted accepted answers but I think that would give signals which are too mixed - besides when I see an amber check mark I'm already going to proceed with caution and looking at the voting score would be the first thing I do.

I think this solution respects the community sentiment at Do accepted answers still serve a useful purpose?, is minimally disruptive to the look and feel of the Stack Exchange network, and saves the time of users and visitors who will not as often have to read or scroll past answers in order to find the one that the community thinks is best.

It also seems to be inline with something @Shog9 mentioned in an answer when declining Can we exempt downvoted accepted answers from getting the top spot?:

And some of our devs have suggested that simply time-limiting the pinning granted by accept (say, pinned for 90 days then sorted normally) would at least prevent it from being an eyesore forever.

Would this proposal work to keep the special place that the Accepted Answer has long held and allow it to have a positional voting advantage for a (potentially long) period as a reward for being accepted, and at the same time prevent an answer that is out of favour with the majority of the community from being stuck to the top forever?

  • 6
    The only real issue I have with this is that, semantically, there's no obvious reason for the truth value of the checkmark as an indicator of the working answer to change over time, and therefore no obvious reason to turn it off. Can you elaborate on that, or push a little harder on why, even if there's theoretically no reason for this, it's an essential hack just the same? – Nathan Tuggy Oct 30 '15 at 4:18
  • 2
    I think I now understand what you meant by "there's no obvious reason for the truth value of the checkmark as an indicator of the working answer to change over time" and of course you are right. So in my refined proposal the check mark is no longer turned off, it just changes colour to indicate that it has been quite some time since the question asker made that their accepted answer, and might not be the same choice that they would make today with potentially more answers to choose from. – PolyGeo Nov 1 '15 at 11:06
  • 2
    I think amber may be too strong a signal. After all there is timeless answers. I'd prefer to lighten the green instead by making it lightly transparent, which is also an indicator it's not as "heavy" with the pinning as before. Just bouncing ideas though. – Vogel612's Shadow Nov 2 '15 at 20:36
  • 1
    @Vogel612'sShadow Light green rather than strong green is certainly something that I could support too. However, I think it could be too subtle and difficult to implement with enough contrast on all screens and sites with customised check marks. A change to the check mark after the stickiness lapses should only imply that considerable time has passed since the asker said "this helped me the most" and that judgement may not be timeless. – PolyGeo Nov 2 '15 at 21:16
7

This has come up before, more than once, as have variants on the same theme.

And I find myself thinking about it now after encountering yet another comment to the effect of "This should be the accepted answer". The comment was left by a user who had found an answer to be "better" than the one the OP accepted. Such comments, of course, don't take into account our established guideline:

Accepting an answer is not meant to be a definitive and final statement indicating that the question has now been answered perfectly. It simply means that the author received an answer that worked for him or her personally.

Yet it's easy to see what someone means by leaving such a comment. Indeed there have been times when I've felt like leaving just such a comment myself, despite our guideline and despite the fact that doing so may not have any positive effect. Quite simply, I want to add my voice to those affirming this answer, but I feel frustrated that an upvote doesn't prevent other users from being presented with the accepted answer first, as though affirming its correctness. I want to add my voice to those saying "this is a good answer, you really want to be doing this rather than what it says in the accepted answer, which may cause you problems, or may not actually work other than in the OP's late-revealed special case, and at best is going to waste your time"—I want to add my voice, and I can do that with an upvote, but no matter how many of us upvote this answer, the accepted answer is going to stay right there as the first thing presented to other users with the same problem. With its nice big green tick shouting "this is correct, this is the best answer!".

The point that I want to make builds on a principle expressed elsewhere in a single sentence:

IMO downvotes are more important than the OP... just because the OP thinks it solves their problem does not mean it is correct/helpful to others.

That sentiment was expressed by @hayd in a discussion about downvoted accepted answers remaining at the top. The current question is less specific, correctly so in my opinion because I would say the underlying principle is purely one of relative vote counts.

What I propose, then, is that we simply drop the practice of sticking accepted answers at the top. An answer that both boasts a big green tick and is presented as the first answer conveys the strong message, "We, SE, affirm that this is the correct and best answer." (Whatever we may say to the contrary in our guideline.) And this is as true for established users as it is for newcomers. When researching a problem and finding an answer on SE, I regularly find myself avidly reading the accepted answer first, forgetting the wisdom of checking the ratings as a preliminary step.

The guideline bears repeating:

Accepting an answer is not meant to be a definitive and final statement indicating that the question has now been answered perfectly. It simply means that the author received an answer that worked for him or her personally.

which it can do just as well lower down in the list, where it should rightly be, if another answer has been voted higher. Then, other users with the same problem who encounter this question can be presented immediately with the answer that has received the highest recommendation from the community, instead of an answer that may now be deprecated.

  • 1
    Yes. Sort answers by votes (only), mark the accepted one where it lies (as we do for self-answers), and provide a way to jump to the accepted answer if you want to, and we get answers ordered by votes while still letting the OP designate what worked for him. Seems like the cleanest solution to a whole host of acceptance-related problems -- bad answers, out-of-date answers, abandoned questions that got better answers, etc. – Monica Cellio Jan 2 '17 at 19:49
  • 7
    I accepted this because, although it is the lowest scoring answer and one which I do not overly agree with, it helps me the most by acting as another example of where a more useful answer that comes in later should not languish below it forever. – PolyGeo Jan 9 '17 at 7:39
19

(This was posted in response to the first version of the question and then edited after the question was revised once. It hasn't been updated for further revisions.)

There are three aspects to this (original) proposal:

  • accepted-answer stickiness
  • accepted-answer reputation
  • accepted-answer labeling (OP clarified in a comment that the intention is for the green check mark to disappear after some time)

I can see some value in removing the stickiness after some period of time on some sites. On sites about rapidly-changing technologies this makes sense; on sites that are more timeless, like Pets or English Language & Usage or Hinduism, I'm not sure it does. So this should be settable per-site, but that's a minor point. I think it would be better for accepted answers to never be sticky, but that's a bigger proposal.

Reputation should not suddenly vanish just because some time has passed. If it was a good answer and helped the OP, the author deserves to keep the rep. We even keep rep for deleted posts if they had enough votes and longevity, after all. It sounds like everybody agrees on this point.

I disagree with removing the visual indicator that the answer was accepted. The acceptance mark has always meant "the OP said this helped him the most"; that doesn't stop being true after a year. Sorting by votes means that better solutions (that the OP didn't pick or that came after he left) will have their deserved visibility. The green check mark doesn't harm that.

One might argue that having an accepted answer farther down the page is confusing, but that situation already exists. A self-answer is sorted by score, and there are many proposals to remove stickiness for downvoted accepted answers. I think it will ultimately be easier to teach users the rule that answers sort by score and (at least for now) acceptance is temporarily special.

So I don't think a green check mark down the page will add any confusion that isn't already there, but I think removing it by anything other than OP action would cause confusion.

In response to changes to the question:

Since I wrote this answer the question has been changed to suggest changing the color of the checkmark instead of removing it entirely. That still leaves room for confusion: should we apply it to self-answers too? What about massively down-voted answers? How do we set the time period?

The real problem underlying your question, I think, is that it's not clear that the OP accepted this answer some time ago and might not have reacted to changes since then (new answers, voting, new technologies that deprecate part of an answer, whatever). The tooltip for the acceptance mark actually tells you when it was accepted, but casual visitors to SE won't know to try that.

Instead of changing a check mark whose meaning isn't immediately clear to another symbol whose meaning isn't immediately clear, how about instead making the acceptance date more obvious in all cases and leaving it to readers to decide how important that information is? Maybe something like this, but with better graphical design:

proposed screen shot

This gives the reader all the date-related information in one glance -- when it was written, last edited, and accepted.

This would be in addition to the bigger green check mark next to the score. I'm not sure a check mark is also needed here; perhaps the text is enough. That might be something to test.

  • 1
    As refined, my proposal slightly changes, but does not remove, the visual indicator that the answer was accepted so it would still mean "the OP said this helped him the most". I would like to think that I have addressed the main point of divergence between the early draft of my question and your answer. – PolyGeo Nov 1 '15 at 10:49
  • 5
    +1 for coming with another (maybe even better) proposal that really makes the checkmark better than it currently is. – Patrick Hofman Nov 1 '15 at 18:14
  • I think my question now addresses the first three new questions that you added to your answer. Coming to your last question, I think adding the extra text on large screens of PCs/laptops/etc would be useful, but what would you propose for mobile phone users? I think the amber check mark works just as well on the small screen as it does on the big. – PolyGeo Nov 1 '15 at 21:57
  • 4
    @PolyGeo I'm not sure this is really being true to the way Q&A is supposed to work, even on meta. You've changed your question a few times now and I'm not going to keep making edits (which will cause you to make more changes, etc). I don't think you've made the case for adding the confusion of multi-colored check marks; I've offered something that addresses what I perceive as the underlying motivation for your feature request, but I don't know what else to do. Better than any of this would be for acceptance to never pin answers, and maybe it's better to pursue that. – Monica Cellio Nov 1 '15 at 22:02
  • I thought the idea of Meta Q&A (Main is different) was to try and achieve consensus, in this case on my idea for bi-coloured check marks. Isn't the strength of wiki-style Q&A that both can keep improving. My motivation for this question was seeing so many ideas about what to do to improve the concept of accepted answers (including discarding it) and thinking there was another one worth exploring. – PolyGeo Nov 1 '15 at 22:17
  • 2
    This approach seems to make the most sense and be the least disruptive IMHO. – Elder Geek Dec 7 '15 at 22:37
  • 3
    +1 for being more highly voted than the accepted answer. – fredley Feb 11 '16 at 8:37
16
+150

I beginning to think that you all have been poking around in my brain. As I've mentioned before, this problem has been in front of me lately and I've yet to come up with solution that has been accepted internally.1, 2

FWIW, I've suggested a variety of things, internally, to "unpin" accepted answers from the top spot, including limiting it to those that were negatively scored (<= -3) to minimize the scope of the answers impacted. My suggestions were overruled because the pinning of accepted answers is part of what makes our sites different from others - it's the indication that an answer worked for the OP - we don't want to lose the signal for those answers or hide them way down on a list of a ton of answers. Unfortunately, at this time this is not something that is going to be implemented.

I won't this because I'm still working on some possible ideas, but this is most likely a long shot.

  • 3
    Would an easy way to jump from the question to the accepted answer (wherever it is on the page) address those complaints sufficiently? (I know there's at least one proposal for that here.) Some sort of "go to accepted answer" link, I mean. – Monica Cellio Nov 2 '15 at 21:24
  • 1
    That sorta just raises the question of why, if it's important, it's unpinned, @Monica. There's an interesting design challenge here; ideally, we'd somehow indicate situations where one would want to read two (or more) answers regardless of whether one was pinned - consider the case where a new answer gains traction on an old question, or the top two answers are far and away more popular than the rest. – Shog9 Nov 5 '15 at 5:13
  • 1
    @Shog9 and that raises the question of what exactly we are trying to signal with acceptance in the first place. I know the theory -- the OP tested this answer, it worked for him, and he doesn't need any more answers (so people are more likely to go answer something else instead where the help is still needed); I'm just not sure it's true on the majority of our sites. I'll probably raise a separate question about this (and try to science it, if SEDE contains acceptance dates). – Monica Cellio Nov 5 '15 at 13:45
  • Followup question @shog and bluefeet: meta.stackexchange.com/q/268959/162102 – Monica Cellio Nov 5 '15 at 22:11
  • I awarded the bounty here to encourage you to not set status-declined while upvotes on the question continue to accrue (albeit more slowly than I would like). – PolyGeo Dec 9 '15 at 12:22
  • 8
    What's the evidence pinning is a strong differentiator? Seems to be lots of evidence it's a strong dissatisfier. On the help page, I'm told that "the best answers rise to the top." On the tour page, I'm told, "Good answers are voted up and rise to the top. The best answers show up first so that they are always easy to find." The system is biased in the favor of the incumbents, and not in the favor of the best content. Readers read like this: nngroup.com/articles/f-shaped-pattern-reading-web-content Older, suboptimal, and deprecated content should be more easily supplanted. – Aaron Hall Feb 11 '16 at 3:40
  • 2
    @AaronHall Pinning is the differentiator that the accepted answer worked for the OP. Tim goes into a bit of detail in his answer about it. We use the checkmark as the indicator that the answer right or wrong, worked for the OP. It's possible that the answer was right 5 years ago, but technology changed and it's now not a best practice type of answer. I'm still looking at possible solutions to this problem, I just need to find something that equally gets buy-in internally and externally. – Taryn Feb 11 '16 at 14:36
  • 2
    So if you don't mean it's a strategic differentiator, note that a lot of stakeholders are unhappy with it, including many that are voiceless here as they don't have accounts here. But these people have accounts on reddit, freenode, imgur, 4chan, and hacker news - and they attend tech meetups and teach CS courses. And these stakeholders greatly affect the overall image of the site. You need some temporal locality in your signaling and ranking. What worked in 2008 doesn't work today. You need to discount your incumbents, and put more premium on the externals. – Aaron Hall Feb 11 '16 at 15:22
  • @AaronHall As I said, I'm still looking at the issue. We're aware there is a problem, but we don't know how big of a problem it is - it's unknown. – Taryn Feb 11 '16 at 15:24
  • 1
    I awarded another bounty here to encourage you to not set status-declined while upvotes on the question continue to accrue (albeit more slowly than I would like). – PolyGeo Oct 5 '16 at 21:46
8

Just a wild suggestion regarding the accept button:

The accept button 'claims' to show the 'best' answer, well, that is how people who are not used to SE think it does. And that claim is wrong, it is merely the opinion of OP, who doesn't have to be an expert on the subject and is not the person who you expect to be able to pick the best answer.

What if the accept button would actually reflect it is the OP who picked that answer as 'his' answer? Why not show the avatar of the user instead of or combined with the accept button? That would make clear it was OP who picked that answer, not the community. It would make clear the purpose of that button.

  • 1
    I think the check mark is used near universally to draw attention that somebody approved this (e.g. supermarket labelling of products), whereas a gravatar, and there are many system generated ones that are near identical, would not be so instantly recognizable as having some meaning. – PolyGeo Nov 1 '15 at 21:52

You must log in to answer this question.

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