0

There are cases that the accepted anwser is not relevant anymore. I encounter this situation often lately, particular at stackoverflow.

For example, this question: How can I make my own base image for Docker?

Early versions of Docker made it diffulult to do what the user asks. I guess the accepted answer was correct at 2013 but not anymore. Later versions allow this functionality out of the box (you can check my answer at the page)

People usually take the aswers they find for granted, so a wrong accepted answer with many upvotes is very misleading.

I think a mechanism to dispute the accepted anwser would work nicely in this case.

Another solution would be to mark the question or answer as deprecated.

Update

The answers at question What to do when the correct answer to a question changes over time? do not address the problem, because this is from the perspective of the person asking the question.

My question is from the community perspective: What to do when we see an accepted answer with many upvotes that is wrong. How to inform the users that the accepted & upvoted answer may not be valid any longer.

2
3

An idea that we explored is allowing users to vote to deprecate answers. With that idea in scope, we have to consider any answer that was once correct, highly-scored, but has simply fallen out of relevance.

We then had to address a pretty tough question, which is given that people have the ability to edit, why isn't that sufficient? What is really blocking people from completely changing an answer that didn't age well so that it's relevant again? Well, we discourage any edit that changes the intent of the author, and in doing that, we've inadvertently exacerbated the problem of some content not aging particularly well.

Thus, deprecating an answer has to do one of two things:

  • Get the outdated information at the bottom of the sort order, in a manner that doesn't make the sort order even more confusing, and take away the check mark if the answer was accepted -- or --

  • Remove the barrier to editing, by applying the wiki status to the answer, which changes how the answer is visually attributed. It's no longer your words next to someone else's user card.

When you consider that some questions have almost a dozen outdated answers (PHP, I'm looking at you), this 'wikify and edit' strategy starts to fall apart. What you've got to do is just get the outdated stuff out of people's faces in some way. And then you'd have questions where almost every answer is deprecated, and you start looking at the question with a hatchet in your hand.

In every sense that deprecation isn't basically synonymous with deletion, UX complexities become pretty formidable barriers to an actual implementation.

The simplest strategy seems to be:

  • Allow 'deprecation' on questions over a certain age, which basically becomes a fast track to deletion when it makes sense to just let the question be asked again if it's still relevant rather than clean up a bunch of answers.

  • Allow 'deprecation' on answers to questions that are a certain age, which is also a fast track to deletion, but would allow the author to undelete provided that an edit was made

Moving stuff around in the sort order is a long tail of suck that we'd rather not try, it's better to just make it go away unless edited.

But, there are caveats - how do we gate deprecation votes? A 10k tool, lower, or higher? What are the surfaces for abuse? Should mods have single-vote power to do this? All questions I still have, which is why I haven't taken another stab at the problem after the last round that we tried.

But I'm open for anyone else to take a stab at it.

2
  • I would think that if you went the deprecation votes way, these votes (from non-mods, at least) would require a badge in a tag from the question (somewhat like dupehammer). Maybe we could look at silver? (I am not suggesting that a single deprecation vote would be sufficient for action.)
    – user642796
    Nov 11 '15 at 11:51
  • @ArthurFischer We looked at that, and that would involve a lot of calculating to render a single page, so much that it's pretty much out of the question. So we'd have to gate this on having a tag bade, rep, or both. It could be unlocked at the gold tag badge level (but require 5 votes), or at the silver tag badge level + a certain amount of rep (which makes more sense as far as making sure it's broadly available), but there are tags where less than 5 people even have a silver badge. You almost have to go on rep alone for total effectiveness.
    – Tim Post
    Nov 11 '15 at 12:06
0

No. This is a bad idea. The purpose of accepting an answer is to indicate which answer was most helpful to the one that asked the question. This is not necessarily the best answer.

If we change this, the entire meaning of accepting an answer changes. If the meaning of the accepted answer changes to what the community finds the best answer, it is effectively the same as the current voting. The highest voted answer is deemed the 'best' according to the community.

2
  • I do not agree with this. You are reffering to questions that for example can have 10 logical/valid anwers but for a specific use case, only one will work. This is not the case here. The question is a common problem that can have only one correct anwser. And this anwser can change over time. Regardless, of how you see the meaining of accepting and answer, or if this feature request is a good or bad idea, this is a valid problem that misleads the users.
    – Mike
    Nov 11 '15 at 9:57
  • The point is: the votes indicate the community consensus. The accepted answer is that of OP. We don't want to change OP's choice. Nov 11 '15 at 10:01

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