Since that happened to me a few times, I am wondering how to deal with the following question. A user asks a question, gets an answer, seems to be happy with it, and accepts it. Some time later, (s)he changes her or his mind, and wants some additional feature. (The "some time later" part is important here, I am not talking about some immediate requests.) Then the user would "unaccept" the answer, and only accept it again if the additional request is met.

EDIT: Here is a concrete example for such a living question. (I stress that I do not want to criticize the OP, who is a rather new user and asks very interesting and clear questions. However, what I do want to achieve is that the unacceptance of egregs nice answer gets reversed, and that the additional requirements become part of a new question. Egreg's answer in its current form will be useful for many.) In this example, a question got asked, an excellent answer was given, and accepted. So far, so good. The unfortunate part is that the answer got unaccepted later and the question got extended.

There are several reasons why I don't like that.

  1. The original question may be of interest for other users. Modifying it later to become considerably different will erase some potentially useful information.
  2. The user has lower chances that her or his new request will be addressed because updated question is not at the top of the list. Only the answerer will know immediately about this question.
  3. I can't see a reason why someone would do that. After all, the questions are free.

My question is how to deal with such situations. One way to go would be some sort of agreement that no one will answer such questions. I'm pretty confident that this would prevent the situations described above.

Main question: how should one deal with such situations?

  • 5
    Does this really happen so often (especially the unaccepting part)? But anyway, I think that a comment telling the person that they should post new question linking to their old question is the best way to resolve this. If the modification to the question is extensive I would also ask them to roll back the edit. Since we want to avoid edit wars, I wouldn't do that yourself, though. – Alan Munn Aug 7 at 16:04
  • 2
    @AlanMunn Not very often but maybe once in 1-2 months. And yes, I of course tried what you're suggesting, but this does not always work. – marmot Aug 7 at 16:06
  • 1
    I suspect that this is also a by-product of "just do it for me" questions... – Alan Munn Aug 7 at 16:08
  • 1
    @AlanMunn Of the "just do it for me" questions or the fact that they get answered? – marmot Aug 7 at 16:19
  • 6
    Well both, of course. But "just do it for me" quickly leads to "just do it for me until I'm completely satisfied" (since it worked once...). But marmots will be marmots and continue to answer them. :) – Alan Munn Aug 7 at 16:21
  • 9
    Related on main meta: Exit strategies for “chameleon questions” – Andrew T. Aug 7 at 16:38
  • 3
    A "just do it for me" question (from my perspective) is one which just posts an image and asks for the code. Such questions also tend to be super specific and therefore of dubious value to others, although I understand that any good answer can in principle be applied to other situations. Questions without MWEs are a much more heterogeneous bunch. But as you probably know, I'm not one for legislating behaviour on such questions, nor for demanding MWEs in all cases. – Alan Munn Aug 7 at 17:32
  • 1
    I don't think there is a lot you can do besides asking the OP in a comment to ask a new question. I have on occasions rolled back edits to questions to restore an earlier version, but only after a short discussion or if a new question has been posted (e.g. tex.stackexchange.com/q/442415). Usually, however, the OP either has not edited the question or rolls the edit back themselves and simply goes on asking a new question after a short nudge in that direction. Like Alan I would not want to risk an edit war, so I would not roll back without prior notice and acquiescence by the OP. – moewe Aug 8 at 13:58
  • 2
    As to conspiring not to answer the question: I don't think that will be necessary, nor do I think it is a great idea. Especially if the question is unclear or just not-very-good in the first place one can't really tell if the answer wasn't to the point or if the question was changed afterwards without intimate knowledge of the edit history and the comments. Plus, I don't think that there are lot of people lined up waiting to swoop down on old questions with existing answers to steal the tick... But then again I mainly answer biblatex questions where not that many people hang out... – moewe Aug 8 at 14:33
  • 2
    @moewe I have very occasionally flagged at that point. If the edit invalidates existing answers, roll-back is appropriate, but it is better to flag than start an edit war. – cfr Aug 13 at 0:45
  • 2
    @moewe See e.g. here for an example. The OP had accepted egreg's answer but then unaccepted it because it didn't fulfill additional requirements that were added afterwards. It has nothing to do with the clarity of the question, the original question was very clear, but with the fact that users sort of "exploit" their power of unaccepting answers in a way that I do not think is 100% appropriate. – marmot Aug 20 at 19:37
  • 1
    Yes, that is a very unfortunate case. As far as I can see, the OP did not actually notify the answerers in the comments, instead they just removed the tick and edited the question. Normally this would probably go unnoticed by the answerers and the general public. The bounty makes the entire thing more intricate, since the OP may now feel that they are more entitled to an answer. As I say there is not a lot you can do to prevent these situations (a warning not to edit old questions with accepted answer? a warning not to remove the tick? - that could go against the system of this site). – moewe Aug 21 at 8:40
  • 1
    ... And even after the fact there is not a lot you can do. Certainly I think you should feel free to ask the OP to roll back and explain why (as you did). But if they don't, there is no way to force them to re-accept an answer. I assume only higher authorities could persuade the user to roll back the edit if they don't want to - and still they might not like it. – moewe Aug 21 at 8:44
  • @moewe I guess the answerer with the accepted answer will know as there is a -15 impact on the reputation. (OK, in egreg's case that's a different story because he's reputation is increasing so quickly that he won't notice. ;-) And yes, I also don't know what one can do. That's why I was asking the question... – marmot Aug 21 at 9:18

You must log in to answer this question.

Browse other questions tagged .