7
$\begingroup$

The question Generating inputs for random-testing graph algorithms? was asked yesterday. My answer explains how to create random graphs with property X. Let us say I would also like to answer how to create graphs with property Y. Maybe tomorrow I would like to explain something about property Z, and so on.

Should I edit everything into one answer? Is it okay to add multiple answers? Why or why not?

I don't have a strong opinion on this. In a way, adding multiple answers makes it easier to structure things and also keeps answers short. On the other hand, there and titles and so on to structure a single long answer. I wonder if this is somehow wrong or abusive since the system asks for a confirmation if you try to add another answer.

I tried googling for the answer, but all I was able to find was Should I edit my deleted answer or add another answer? which is not what I'm looking for. In some questions (usually big-list questions) the asker specifically wants only single point per answer. On CSTheory, there are examples of both single long answers and of the type one point per answer.

$\endgroup$
0

4 Answers 4

3
$\begingroup$

Almost never.

This comes up now and then on the main meta (e.g. 1 2 3). It is pretty rare for adding another answer to be the right thing. Most of the time, if you have two solutions for the same problem, it's best to include them in one answer, and discuss their relative merits.

A case where it would be ok to post two answers is if you have two solutions which can't usefully be compared. For example, a slow algorithm that works in all cases, and a faster algorithm that requires an additional hypothesis that is common but not specified by the question.

One-point-per-answer (“list questions”) are proscribed on most Stack Exchange sites, for good reason: what you get if you do that is not answers but items.

In this case, I think it would be best to have a single answer, with an introduction explaining what properties you think are interesting then sections for each property.

$\endgroup$
4
  • $\begingroup$ I disagree. For example, the "How to show L is not ..." questions work much better with one answer per method; each method is a self-contained answer to the question. In the question Juho links, the case is not as clear as the question underspecifies the problem to some extent. There, one answer outlining several alternatives is probably better. $\endgroup$
    – Raphael
    Aug 1, 2012 at 21:40
  • $\begingroup$ @Raphael I think you misspelled “I agree”… This example would fit in “two solutions which can't usefully be compared”. $\endgroup$ Aug 1, 2012 at 21:43
  • $\begingroup$ Well, they can: different methods can be applied differently well to different languages. One method is even a generalisation of another. So yes, they can be compared, but it's useless to have some of them lumped together. Either all in one answer, or each in its own; mixing is weird. $\endgroup$
    – Raphael
    Aug 1, 2012 at 21:47
  • $\begingroup$ Non-reason: a user who goes to the lengths to present multiple methods should be able to receive rep for each. $\endgroup$
    – Raphael
    Aug 16, 2012 at 8:34
4
$\begingroup$

My opinion: If you have two different answers, it should be okay to post them as different answers.

  1. The function (several answers by a single user) is supported by SE.
  2. More answers gives the list of answers more structure.
  3. Say, you present two different proofs of a theorem, than the community can vote which version is preferable.
  4. Commenting works better.
$\endgroup$
1
$\begingroup$

Part of the beauty of SE is, in my opinion, that the community and the asker get to rank and choose the best answer. This doesn't happen at all if there is only one answer.

For someone who regularly follows questions in the community, comments hold relevant information, but for someone who has a question and is google-searching to see if it's been answered on SE, comments often make it hard to find the relevant information.

Alternately, multiple answers provide a clear indication that there is an element of subjectivity present, that there might not be a single "right" answer.

$\endgroup$
0
$\begingroup$

You can edit all of your answers into one answer. But I wouldn't edit other people's answers.

$\endgroup$

You must log in to answer this question.

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