6
$\begingroup$

I posted this Q&A question on meta, for the sake of sharing knowledge.

After posting, a lot of different iterations and improvements came on the answers. I build upon one of the answers, made some improvements, and built a (at least to my purposes) perfect solution. I wanted to share this knowledge, so I had two options: I could either post a new answer, or edit my current one, which was very insufficient at this point. I chose the latter.

So my question is: Is this the preferred way of going about it, or is it better to post a new answer?

| |
$\endgroup$
  • $\begingroup$ you should make clear that this is about a question on the meta site. $\endgroup$ – supinf Oct 28 at 6:52
  • 1
    $\begingroup$ I think posting a new answer is OK. $\endgroup$ – GEdgar Oct 28 at 11:44
  • $\begingroup$ @supinf I've edited so it's clear. However, I'm asking this question about rewrites in general, both on meta and on ME. $\endgroup$ – Luiz Martins Oct 28 at 13:38
  • 2
    $\begingroup$ Editing your answer is better since it avoids people wasting time reading an outdated answer. If you choose to post a new one I would add a note at the beggining of the old answer pointing to the new, better answer. $\endgroup$ – Ruy Oct 30 at 15:54
  • $\begingroup$ Personally, I think editing the current answer was a good choice. Posting a new answer would have also been fine. $\endgroup$ – littleO Nov 2 at 6:40
  • 1
    $\begingroup$ there was a secondary effect to making the edit (minor in this case), the comments on your answer don’t really make sense / had their meaning changed. on Main i would flag them as no longer relevant / conversational but not sure what to do now $\endgroup$ – Calvin Khor Nov 9 at 1:23
  • $\begingroup$ @CalvinKhor That was one of my concerns. Feels similar to how it's not acceptable to rewrite a question if there's already an answer to it. $\endgroup$ – Luiz Martins Nov 9 at 1:30
2
$\begingroup$

One should post a new answer if both answers can stand independently and both answers contribute distinct ideas (i.e. it is valuable to have both). One should use editing to improve an answer without changing its basic content - including to build upon an idea, to rewrite things more clearly, to simplify arguments, and other similar tasks.

On the main site, this tends to play out as saying that if you find simplifications (or errors) in a proof in an answer or you find a way to build upon a proof, you should probably edit those in even if the result is that not many of the words of the original answer survive. In the opposite direction, if you post an answer using Fourier analysis and later discover a way to do it via a combinatorial argument, you should create a new answer - even if you feel that the new answer is superior to the old one.

Mostly, I suggest this because it often is the case that having both answers is better than having just one. The system of sorting answers via up/downvoting also tends to assume that answers don't change too fundamentally.

In the given case, I think your old answer contributed something valuable: it showed how to use the array environment in a simple way to accomplish the given effect - and this is something I believe would be useful to maintain, especially since people may be more likely to use array than to search meta for some specific macro definitions. Your new answer builds upon this original, showing how to encapsulate this basic idea with macros. In my opinion, the best choice in the given situation is to keep most of the original answer and then additionally include the refinement - my inclination here would be to just take the old answer and include it at the start or the end of the new one (possibly separated by a horizontal line), although there are surely more elegant ways to ensure that no value is lost in the edit (e.g. one could write it more as "here's the basic idea and here's a more advanced extension" or "here's a working system and here's how it works" - but such solutions can take some work to write and aren't strictly necessary).

| |
$\endgroup$
  • $\begingroup$ I agree with you on this one, perhaps there was merit to my old answer. I think both of your ways (posting a new answer or adding to the start/end) have their place, and may be used in different occasions. In any case, I've decided to reinstate it and append the new one to the end. Thanks for your answer! $\endgroup$ – Luiz Martins Nov 9 at 2:16
  • 1
    $\begingroup$ A minor quibble about your opening: one should post a new answer if the new answer stands on its own and contributes something new. As you have written it, you seem to be suggesting that if an old answer does not stand on its own, then one should not post a new answer. ;) $\endgroup$ – Xander Henderson Nov 9 at 14:13

You must log in to answer this question.

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