10

The title says it all, really. Why I'm asking is because I've seen some questions with accepted good answers, but there are better ones -- more accurate, and probably with higher scores. This is particularly evident for the type of questions that recommend a certain package to do the job -- most of the times there's a package that supersedes another one, is better maintained, etc. Here is an example.

So, to reiterate:

Can somebody else besides the inquirer change the accepted answer, and if yes, what's the policy on that?

Update:

I found this discussion on meta.SO that is somewhat related to the issues of this question, particularly accepting wrong answers. Some opinions are worth reading.

10

Although I don't really like the proposed idea (moderators changing accepted answers), I do like the sentiment behind it. As Andrew has nicely spelled it out, this seems to be an issue of balance between the short-term goal of “I want an answer to my question, now!” (and acceptance of an answer means fulfillment of this goal), and the long-term goal of keeping a database of usable answers for people later searching-for/stumbling-with a page.

It seems that most people around think that the short-term goal is all that we should care, and the long-term goal is simply a “nice” by-product. However I'm one of the few who disagree.

At least for me, trying to build a mostly correct and up-to-date source of answers for (La)TeX questions is one of the reasons that got me motivated to participate and spend some of my time in this site. The web is already full with incorrect and outdated information about (La)TeX which is only kept alive because people searching for solutions on the web keep stumbling with the old and obsolete answers. From the beginning I thought that this site could finally be a solution to this problem, and that's why (for example) I tend to be more keen on heavily editing answers “for the future”.

I'm not suggesting, of course, that we should have to police around the site and have this long-term goal in mind all the time. Perhaps I would consider checking from time to time the most voted/viewed questions for sanity. But the ideal would be that we don't have to do much more for the site to be able to fulfill the long-term goal as well as the short-term one; and so, for the particular issue raised by the OP, this is my suggestion:

How about requesting a feature where the most-voted answer (perhaps after a certain threshold of votes) gets to be shown before the accepted answer?

  • 2
    "The web is already full with incorrect and outdated information about (La)TeX" -- I can't agree more. This was mainly what I had in mind when I asked the question, as I have spent quite some time hunting down obscure LaTeX bug, and stumbling on mailing lists, forum posts and blogs from 4-5 years ago -- and most of the time they don't work, or there are better ways to do it. – Martin Tapankov Sep 28 '10 at 18:55
  • 1
    I'd like to correct your second paragraph (in so far as it seems to be a comment on my answer). I may have given that impression, but it is the wrong conclusion to draw of my motives. The long-term goal is very important. However it cannot be achieved by itself. It has to come about as a by-product of the short-term goal. The short-term goal is what feeds the site. For people like us, the long-term goal is important to keep in mind. But it must not be something that Joe User has to think about. – Loop Space Sep 28 '10 at 19:54
  • 1
    That said, I like your final proposal. My variant would be that N days after an answer has been accepted, the default order is by votes. I'd also make it so that after M days, votes no longer count towards reputation (either positive or negative). – Loop Space Sep 28 '10 at 19:55
7

I think it's a good idea to think of the "accepted answer" as the short-term answer and the "most voted for" answer as the long-term answer. It's important to remember that someone asked the question wanting an answer, and often wanting an answer right now. Accepting an answer is the questioner's way of saying, "That's the one that worked for me when I asked the question.". But questions also have a long-term lifetime and are a useful resource for others (I've just directed someone to the question on "How to draw a torus"). The most voted for answer is likely to be the best that someone stumbling upon the question is likely to want.

But if our aim was just to build up a long-term resource, the site wouldn't work. We have to satisfy the short-term goals of the questioners and encourage people to answer the actual question that was asked. The long-term archive of good resources is a by-product of that short-term Q&A system. An extremely valuable by-product, and one that is really crucial to the sites' worth, but focussing on it to the exclusion of the short-term won't work.

  • But if the answer was accepted it means that the short-term goal has been fulfilled. What is wrong with, afterwards, also achieving the long-term goal by showing some better answer as the first one? – Juan A. Navarro Sep 28 '10 at 17:57
  • 1
    @Juan: nothing. But I'd prefer to do it without monkeying around with the fact that an answer has been accepted or not. Accepting an answer means one thing, let's not try to force it to do two jobs. – Loop Space Sep 28 '10 at 19:58
7

Mindcorrosive asks, in response to Jukka's answer: But what happens when the the accepted answer is not relevant anymore (for example, package changes, fixed workarounds, obsoleted packages)?

I've seen the following suggested on meta.so: ask a new question, referring to the old question and saying its assumptions don't apply anymore, and ask for a better answer. Then have the old answer closed as a duplicate of the new answer (really, it seems to me it should be "too localized", but that doesn't give you the nice link). This kind of thing can be coordinated here on meta.

  • 3
    This seems like a pretty decent plan. It doesn't make anyone mad/sad because "I lost my checkmark" or anything, it doesn't require cooperation from the original asker (who most likely doesn't care anymore), it avoids the confusion and clutter of having old and no-longer-valid answers mixed with new ones (and probably above them for at least a while), and it works with the incentive system, not against it. – SamB Oct 1 '10 at 16:29
3

I think nobody else should ever change the accepted answer, regardless of whether it's possible or not.

If you think the "wrong" answer was accepted, just upvote the "right" answer. (If the accepted answer is extremely bad, you can downvote it, too, and leave a comment explaining why it is bad.)

If you don't like the concept of accepted answers, just ignore it and always read the top-voted answer. It is listed right after the accepted answer.

  • 1
    I don't mind the accepted answers per se -- they are a good indicator which ideas are likely to work. But what happens when the the accepted answer is not relevant anymore (for example, package changes, fixed workarounds, obsoleted packages)? Then the accepted answer is misleading, and one may need to check other solutions. This ruins the whole idea of "accepted answers" -- "stuff that works". – Martin Tapankov Sep 28 '10 at 9:58
  • 1
    The problem is not with regulars of the site which can “just ignore the accepted answers”, but with people stumbling with the site, only reading the first (now obsolete) answer and spreading around that information. – Juan A. Navarro Sep 28 '10 at 18:22
  • 2
    Moreover, it also feels wrong just to downvote an answer because it got obsolete. Why should an user loose reputation for posting an (accepted!) answer, because (La)TeX decided to evolve and later provide a better answer? – Juan A. Navarro Sep 28 '10 at 18:25

You must log in to answer this question.

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