If someone answers my question and I want to generalize it, how should I do it? Is it ok to gradually extend my question as answers improve? An answer to a question I wanted to ask may show me that I really wanted something else, whence asking just what I need is not always possible. (This is especially true of research related questions, which should be the ones we want here anyway.) Is there a good principle for deciding whether to extend an old question or to make a new one?

From the other perspective, if someone keeps extending their question, should I extend my answer or give several answers?

Something of this kind happened with this question that I answered recently and it left me thinking.

  • 19
    $\begingroup$ I would say: pose a new question. Don't change the question to turn answers into non-answers. $\endgroup$ – Gerald Edgar Aug 12 '14 at 16:15
  • 1
    $\begingroup$ I think its ok to add an uodate with a new version of the question. "should I extend my answer or give several answers?" Both options are OK. $\endgroup$ – Gil Kalai Aug 12 '14 at 19:43
  • 10
    $\begingroup$ It's cleaner to pose a new question, referring back to the old question as desired. $\endgroup$ – Todd Trimble Aug 12 '14 at 20:30
  • 5
    $\begingroup$ If you are the one asking the question, you can also put comments under answers at the first question giving a link to the new question. If there turns out to be sufficient common interest, perhaps you have found a collaborator, even co-author. $\endgroup$ – Will Jagy Aug 12 '14 at 21:33
  • 8
    $\begingroup$ If I am the answerer, and the questioner changes the question, I ask the questioner to go away and come back when he has figured out what question he really wants to ask. But others are less short-tempered than I am. $\endgroup$ – Gerry Myerson Aug 12 '14 at 23:33
  • 2
    $\begingroup$ Similar question under discussion at m.se, meta.math.stackexchange.com/questions/16571/… $\endgroup$ – Gerry Myerson Aug 12 '14 at 23:42
  • 4
    $\begingroup$ I think some minor changes to a question are perfectly acceptable. For instance, if it turns out that in your original formulation, your question admits some trivial counterexample that is orthogonal to the issues you are actually interested in, I would recommend editing the question to exclude the trivial case (or to ask for further examples) rather than making a whole new question. $\endgroup$ – Eric Wofsey Aug 13 '14 at 18:11
  • 1
    $\begingroup$ As an example, this question asks whether a category with a certain property exists, and the empty category has that property. If the only answer given had been that the empty category was an example, then it would have been perfectly fine for the asker to edit the question to ask whether there was also a non-empty example. $\endgroup$ – Eric Wofsey Aug 13 '14 at 18:23

You must log in to answer this question.

Browse other questions tagged .