8
$\begingroup$

I am wondering about if it is generally good practice to encourage someone posting about verifying their own proof to instead format their post as a question where they also post an answer? I specifically thought of this after looking at this question On open sets and closed sets where the asker definitely has their proof correct and I can't imagine anyone coming up with a better answer then "yeah you are right" which could be left as a comment so the question may end up being "unanswered".

When I was looking around here on the meta site to see if there was a good post already answering this question I saw this post Is it inappropriate to turn a proof verification question into a self-answered question? and specifically the line in the answer "This improves the readability of the question for those who come across it later: users expect to find an answer is the answer box, not in the question." I think that this is generally true that proof verification questions would be more readable if the question and proof were separated out from each other. What do other people think?

Specifically I was thinking of leaving a comment encouraging whoever is asking about a proof verification to answer their own question or something to that effect.

$\endgroup$
1

2 Answers 2

6
$\begingroup$

My answer with

This improves the readability of the question for those who come across it later...

was given in the context of a question that was already thoroughly resolved, and the issue was how to archive it better.

For an active proof-verification question, I'd expect the question body to contain everything that is needed for an answer. Separating the proof to be verified into an answer box has the following disadvantages:

  1. Reader has to look at two text boxes instead of one; this is annoying. Especially if another answer is posted by a 3rd party: now one has to look among the answer for the answer that contains the solution to be reviewed.
  2. If someone upvotes the answer for any reason, the whole Q&A will appear "answered" (i.e., will be removed from unanswered tab) even if nothing whatsoever was posted about the proof.
  3. Some users filter out questions that already have an answer (I do, at least). So, if a question is posted with an answer to it, I will never see it.

As an aside, Code Review is a Stack Exchange site that serves a purpose similar to solution-verification questions. As you can see, questions there are used for the material what is to be reviewed, and answers are used to critique that material. Makes sense to me.

$\endgroup$
2
  • 1
    $\begingroup$ Of course, on Code Review the asker is instructed to be fairly confident in the correctness of the code before submitting it for review. $\endgroup$
    – user642796
    Nov 2, 2014 at 12:44
  • 1
    $\begingroup$ To me the Code Review model feels a little different from the current model of most "check my proof" questions. As far as I can tell the point of questions is explicitly "what can I do better?" (and they ask for the code to be working so it isn't really about fixing stuff that is wrong) whereas here on Math the point of most questions is "is this proof right" and not "how can I make this proof better" (though you are definitely right that you can still answer "is this proof right" questions with "it is but here is how to improve it". $\endgroup$
    – user171177
    Nov 2, 2014 at 15:05
1
$\begingroup$

If this were a website mainly geared toward participants of a similar mathematical maturity and/or discipline, I would agree with your proposal: after all, what's the point of writing a "correct" response if all you're typing is "correct"? Yes, it should be left as a comment, and unless the proof is incorrect/incomplete, one wouldn't expect an answer post.

However, Math.SE is not such a website, but rather a place where mathematicians of all levels (and disciplines) read, view, and answer the similar problems. Even if someone's proof is correct, it should be encouraged that others not only verify one's proof to ensure consistent logic, but also to provide different suggestions and ways to prove whatever statement the OP had sent out to prove. In other words: there are many ways to prove something, and it only adds to the value of this website (and every subsequent user) to see alternative proofs, more efficient proofs, etc.

These will definitely have room in answer posts, and answer posts can come at any time- in fact, I think there are several badges that provide a little incentive to answer questions that have gone "unanswered" for long periods of time.

Just my $2¢!$

$\endgroup$
5
  • $\begingroup$ It seems to me like a question is more likely to get a variety of proofs/methods if it is phrased as "How to prove this" and the poster puts their own solution up instead of phrased as "Please check my work". Do you feel like you often see good answers to "check my proof" questions when the posters proof is correct? $\endgroup$
    – user171177
    Nov 2, 2014 at 0:29
  • $\begingroup$ @Gage A question phrased as "How to prove this" is likely to get closed, since reviewers do not necessarily look at answers when voting on a question. Or it may acquire a bunch of different solutions, none of which address the issue of whether OP's proof was correct. $\endgroup$
    – user147263
    Nov 2, 2014 at 5:55
  • $\begingroup$ @Gage actually, yes, sometimes I see very interesting follow ups. The "yes, it's correct" statements are usually kept in the comments, as you suggested. $\endgroup$
    – daOnlyBG
    Nov 2, 2014 at 8:28
  • $\begingroup$ @justabrickinthewall I was assuming it would include in the question something like the context "I have this proof i think is right see my question" or something but I definitely understand that these questions are better than ones where the person copy and pasted a question out of a book. $\endgroup$
    – user171177
    Nov 2, 2014 at 14:53
  • $\begingroup$ @daOnlyBG That is cool! I was only asking because I felt like I hadn't (when the proof was correct) but I'll need to keep my eye out more I guess. $\endgroup$
    – user171177
    Nov 2, 2014 at 14:53

You must log in to answer this question.