7

There are many questions like this one:

Identifying a story about a girl named Sarah who finds out she is a robot

The pattern is the same:

  1. User (usually new one) asks a question.

    Often, it's their only question on the site

  2. Someone produces an unequivocally correct answer.

    "Correct" as evidenced by one or all of the following:

    • High upvotes

    • Agreement between many high-rep users on chat/meta

    • Special formal vote on Meta

  3. User disappears from SF&F SE forever (as evidenced by "Last seen" timestamp), without accepting the answer as correct.

    More often than not, the account creation date==disappearance date.

We are left with a question with no accepted answer that nevertheless is 100% (as much as we can tell) correct, but is somewhat mis-leadingly not accepted.

Questions:

  • Does anyone else feel that this is detrimental?

  • Is it desirable to have the community be able to cause "Acceptance" on such questions?

Clarifications:

  • This is only intended to apply to questions where one answer is crystally clear correct, based on some fairly stringent pre-defined measure.

  • The exact mechanism by which this can be achieved (if it's deemed a good idea) is out of scope for this question

  • The question of whether this is even possible to do (if it's deemed a good idea) is out of scope for this question

  • The linked example question is not the absolutely best example of "100% sure that the answer is correct" - merely the first I found.

  • You raise some interesting ideas, I've day-dreamed about auto-acceptance on those questions. It's a tough call. – Mark Rogers Jan 19 '12 at 23:38
  • 2
    Auto is not a good idea because the reason for lack of acceptance could be that the answer just isn't good enough. – DVK-on-Ahch-To Jan 19 '12 at 23:42
  • Yeah, I agree that's the unacceptable drawback of auto-acceptance. It would be nice if a very high rep user could accept for an unaccepted question that is x months old as a gain-able power. – Mark Rogers Jan 19 '12 at 23:44
  • 2
    I think after six weeks or so, there should be a mechanism to auto-award an answer. To do otherwise can discourage answering such questions from newbies. – Tango Jan 20 '12 at 0:16
  • If this were possible, would it also be possible to have the answer un-accept should the user ever return to the site? Maybe with the site sending a message to the user's network inbox, reminding them to review the answers? – Xantec Jan 20 '12 at 0:41
  • @Xantec - that's an implementation detail (says my Architect hat) – DVK-on-Ahch-To Jan 20 '12 at 1:21
  • Around the January this year, I vaguely remember in Chat that we were talking about one of these seemingly abandoned questions accepting an answer around a year after it was posted... – Izkata Jul 22 '13 at 23:29
  • 2
    Suppose someone comes along and posts a question which is identical to a seemingly abandoned question, and suppose the new question is answered and the answer is accepted. Could the original question then be deleted as a duplicate? Or do the rules require that the newer question be deemed the duplicate? – user14111 Jul 23 '13 at 5:59
  • @user14111 - I think the rules are flexible and even if the newer is closed as duplicate, this can be flagged/disputed on Meta and fixed. Generally, I saw a trend to have the superior Q&A left open and inferior closed as duplicate; independent of timeline; but it's not a hard and fast rule – DVK-on-Ahch-To Jul 23 '13 at 12:57
  • @DVK In that case, there is already a mechanism in place for accepting abandoned questions. What am I missing? – user14111 Jul 23 '13 at 20:06
  • @user14111 - hmm... plausible. – DVK-on-Ahch-To Jul 23 '13 at 20:16
  • Is it desirable? Yep. Will it ever happen? Nope. – Wad Cheber May 18 '16 at 9:19
15

This has been discussed many times on the main meta site, in particular:

  • 2
    I'm aware that this was discussed on MSO. However I feel that there are certain cases where SO differs from SFF - on SO, one (if not the main) reason against it is that each users' technical probems are likely to be so special, that only he can tell if the answer was actually most helpful. On SFF, we can frequently tell without consulting the original user. – DVK-on-Ahch-To Jan 19 '12 at 23:47
  • 8
    @DVK I don't see why SFF would be different. In fact, it's rather the opposite; there are many answers on SO that can be verified easily by running the code. – user56 Jan 19 '12 at 23:49
13

As one who specializes in story ID questions, I'd find it very useful to have all of the correctly answered ID questions so marked. DVK mentioned three kinds of evidence for an "unequivocally correct answer". Let me mention a fourth and even more unequivocal kind: when the OP accepts the answer verbally in the form of a comment, without formally accepting it. This has happened to me 3 times in the 3 months I've been a member:

Here, the OP came back and commented "That's the one... awesome, thanks!" and even posted a second comment "Yes, it was an older book when I read it. Some of my recall was a admittedly a little off. In any case, this has nagged me for years, so thanks again for your reply" without accepting (or voting but that's beside the point).

Here the OP commented "Thank you so much, that was precisely the one I was looking for. I had almost given up ever finding it." without accepting or voting.

And here the OP's comment "The interwebs are truly amazing. I wondered what the name of the story was for 35 years and you answered the question two minutes after I uploaded it. Thank you Jeff." was appended to the question and addressed to Jeff, who did not answer or comment on the question but who edited the title. Again, no acceptance or vote.

Perhaps, in cases like this, the moderators should be empowered to mark the answer as "accepted" without awarding points.

Or maybe, on the Ounce of Prevention theory, a better way could be found to inform new or one-time users about the custom and mechanics of marking accepted answers? One time (maybe on another stack exchange site) when I was answering a question for a brand new user with 1 rep point, I tried to take matters into my own hands and ended my answer by saying something like "if you find this answer satisfactory, you may consider accepting it by clicking on the check mark on the left" but of course someone came along and edited it "deleting hint to accept". Would a "hint to accept" in the form of a comment be kosher?

  • On StackOverflow, I've seen OPs ignore both the in-answer hint and the comment-hint, post a comment similar to what you quoted, then vanish. Not really sure it'll help. – Izkata Jul 22 '13 at 23:21
2

An important thing about acceptance: it doesn't mean the answer is right.

It only means, has only even meant, will only ever mean that the OP accepted it as right.

How do you know the OP made the right choice? You don't. Indeed there's a badge (a gold badge) for answers that are much better than the accepted one.

And if acceptance doesn't coffer magical correctness, then there is no pressing reason to insure that every questions has an accepted answer. Aside for the bit where I want my fifteen points, of course.


Acceptance also doesn't mean that the question is somehow over: questions don't close, but remain available for better answer to be added in the furture. Indeed there are a couple of badges just for that. Four if you included the ones that reward editing old posts under some assumption that this encourages improving old answers.

  • It's not about 15 points - for that, we can have other mechanisms if we even care. In fact, one of the implementations of proposed policy explicitly provides zero extra rep. – DVK-on-Ahch-To Jan 20 '12 at 1:23
  • 2
    It's about decreasing the amount of questions with no accepted answer, which IMHO is detrimental to the site – DVK-on-Ahch-To Jan 20 '12 at 1:25
  • 1
    You are mistaking acceptance for something it isn't: something important. It's meaning is defined by the the mechanism: it means the OP picked the answer. It doesn't mean anything else. – dmckee Jan 20 '12 at 1:29
  • 1
    Sorry, but YOU are mistaking the official party line of what acceptance is supposed to mean with my (perhaps correct, but perhaps not) assertion of what it LIKELY means to new users. Also, J&J are not infallible - I remember major MSO flamage about not valuing Qs as much as As - only to have them realize that people like me were right all along in stating it was the right thing :) – DVK-on-Ahch-To Jan 20 '12 at 2:06
  • This isn't a party line its a tautology. The OP picks some to accept and it gets the mark. So, the mark mean the OP selected it. That's all that's there. – dmckee Jan 20 '12 at 2:11
  • You can assign ANY semantics you want to anything, and the users will likely assign somewhat different semantics. Who's right, the designer or the user? – DVK-on-Ahch-To Jan 20 '12 at 2:13
  • So, you want to assign "correctness" semantics to the acceptance. What do you do when the OP chooses the wrong answer? Are you going to take away the only feedback mechanism available to users with less than 15 rep? What about a situation like Why do replicants have a short lifespan?? Wayfaring Stranger's answer is definitive in some sense, even if I don't like it. The semantic in the current code are unambiguous. You make this change and you remove that certainty. – dmckee Jan 20 '12 at 2:21
  • 1
    Nope. As I explicitly noted, (1) I am addressing the semantics of ABSCENCE of accepted answer [to be honest, "wrong" accepted answers are exceedingly rare even if I cared]; (2) I am positing that the semantics assigned to such by the users will SOMETIMES be wrong (e.g. they assume the answer is wrong whereas the careful examination shows nobody was around to accept) and (3) I'm proposing to rectify SPECIFICALLY those cases as deliberated by the community. If there's any doubt/disagreement, err on the side of NOT forcing the accept. But I saw many of those where there's no possible disagreement – DVK-on-Ahch-To Jan 20 '12 at 2:32
  • 1
    The question is, is the concept of 100% community supported forcing of an "accept" where in the "right" course of life it would have existed, going to make the site better or worse? I state that with proper controls - better. Even if it will violate some inner semantics that Joel made up 5 years ago, for SO :) – DVK-on-Ahch-To Jan 20 '12 at 2:34
  • 2
    @dmckee No, that is not all that the acceptance mark means. The "Community" user regularly bumps questions that don't have an accepted answer, because not having an accepted answer means that none of the posted ones satisfied the OP. Or at least, that's the intent, because OPs that don't accept an answer are cluttering up the system. – Izkata Jul 22 '13 at 23:23
  • 3
    Unless things have changed in the last few years, Community only bumps questions when they don't have an accepted or upvoted answer. So that's not likely to be relevant here; if you want to stop a question with an obviously correct answer from being bumped, all you have to do is upvote it. – Micah Jul 24 '13 at 2:42

You must log in to answer this question.

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