24

The title says it all. On TeX.SE, is it proper for the OP to post his/her first attempt at solving the problem as a self-answer? Or should the first attempt be included in the question instead?

Take Get the head, the tail, and the last element of a comma-separated list as an example. The OP posted a self-answer to his question immediately after posting the question, a self-answer which he self-confessedly found not entirely satisfactory.

I think this practice may confuse subsequent visitors. IMO, a self-answer should be reserved for a definitive answer, i.e. an answer that solves the original problem in a more satisfactory way than any of the other answers proposed so far.

What do you think?

  • Technically I said "not entirely unsatisfactory", though I can certainly understand the misreading! (I've re-worded it now.) Anyway. I think it's reasonable for self-answers not to be perfect. That's why SE stops you accepting your own answer within the first day or so, so others can improve on your answer. (Originally I had intended my self-answer to go in the question, as you suggested, but found that as I tinkered with it, it became sufficiently good that it warranted being an answer, albeit a slightly flawed one.) – John Wickerson May 23 '13 at 15:13
  • Since other answers are present and one of them is accepted, I don't think it's a problem. – percusse May 23 '13 at 15:19
  • @JohnWickerson Hi John, Thanks a lot. I think i will delete my comment at main site as you have clarified the issue well. Good luck. – texenthusiast May 23 '13 at 15:47
  • Great, no problem. :-) It was a sensible issue to bring up, as I had wondered myself about the right thing to do. Nice to have some clarification for the future. – John Wickerson May 23 '13 at 16:38
  • @JohnWickerson Thanks for clarifying, John. Just so you know, I didn't mean to single out your question :p It was only the most recent example I could find. – jub0bs May 24 '13 at 12:01
24

In my opinion, the main criterion is "does it answer the question?". If yes, then it is an answer, if not, then it can be part of the question.

In particular, if the question is about getting a working answer, a first attempt with dirty code or a hackish workaround is perfectly suitable as a self-answer (bonus points for mentioning the limits of the answer, so that others can improve it). On the other hand, if the question is about improving a working but otherwise imperfect solution, that first-draft solution belongs to the question, and answers should show what is asked for: improvements.

I don't see how it is confusing for future visitors to have a imperfect, but working solution, to a question. When better answers are posted, the voting/acception mechanism will automatically make them more visible. In last extent, if it turns out the answer doesn't work as well as it should, it can always be deleted.

And last, but maybe most importantly, if someone with the same problem visits the question before it gets another answer, I think he will be glad to find an imperfect solution, instead of no solution at all.

  • Great,agree. mentioning the limits of the answer explicitly +time frame to self answer are key points (one must atleast allow time to settle with 2 alternatives before self A)+announcing before hand a hack is being worked on, will be posted may be an option. – texenthusiast May 23 '13 at 15:28
  • 1
    I disagree with the time frame condition, the time frame for accepting the self-answer should be enough. Imo, any question should have an answer (even imperfect, especially if its imperfections are documented) as soon as humanely possible (which, here, can be as soon as the question is asked). The hypothetical visitor can always read the answer, decide that he can't accept its limits, star the question and wait for a more perfect answer. – T. Verron May 23 '13 at 15:34
  • Seems reasonable, just made some points of thought while reading other related Q. – texenthusiast May 23 '13 at 15:40
  • Yes, I see your point now. The other question is very similar, except that the self-answers are assumed to be without imperfection. The same arguments apply, but another parameter has to be taken into account: will people be willing to answer a question which already has a working and non-imperfect (it's not entirely the same as "perfect") answer. – T. Verron May 23 '13 at 15:43
  • +1 Thanks; your answer led me to somewhat reconsider my stance on self-answers. – jub0bs May 24 '13 at 12:02

You must log in to answer this question.

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