This follows Amir Asghari's advice concerning my comment on the answer to Answering my own question:

Background of my question is, that I can envisage different reasons for reference requests on MO, e.g.
1. I want to learn more about some topic, but I don't work on a related problem.
2. I work on a problem and need further information for progress.
3. I have solved a problem and would like to know, if either the solution is already known or, if it has been solved differently.

My question is related to the 3rd point, namely that I solved a problem, but could not find anything similiar on the internet; to be sure, I finally post a reference request on MO. After having waited a reasonable time, in which no appropriate reference has been received as a feedback (answer or comment), I have essentially two choices:

  1. publish my results elsewhere and supply a reference to that publication as an answer to my reference request or,
  2. provide my solution to the problem as an answer to my reference request.

My personal opinion is, that option 2. should be acceptable if the result can be presented in a 'fairly concise' manner.

I'm aware that my opinion stated above, leaves a lot of room for interpretation; therefore I would like to learn of the opinion of others.

  • $\begingroup$ Why don't you just do both things, 1 and 2, if you feel that the result is worth publishing ? If the proof is too long, you can answer in MO with a reference to your preprint. $\endgroup$ Commented Sep 3, 2013 at 20:01
  • $\begingroup$ @AlexandreEremenko: what you propose is exactly option 1; or did you mean that I should publish a more elaborate version elsewhere and a 'compact' version on MO first and later add the pointer to the article? – $\endgroup$ Commented Sep 5, 2013 at 5:50
  • 2
    $\begingroup$ Yes. It is unreasonable to enter in an MO window a text which is longer than 1 page. $\endgroup$ Commented Sep 5, 2013 at 13:05

2 Answers 2


My feeling is that either (1) or (2) should be fine, depending on the result. Generally if it's concise enough for (2) then my feeling is that it probably isn't publishable on its own, though this may depend on your field. Of course, if I were you I'd also ask experts directly whether this was previously known before submitting it to a journal. Just because your question didn't get an answer on MO doesn't mean the result didn't appear elsewhere. Maybe none of our users knew about it or maybe the users that could have answered weren't checking that day. With the current rate of questions most only stay on the front page for a matter of hours.

  • $\begingroup$ my 'problem' is, that I'm not a professional mathematician, so there are no experts around which I could ask directly; asking on MO whom I could/should contact doesn't seem to be an option either, so the only option I see for me, is to be very careful when posting on MO and it is also for that reason, why I sometimes refrain from posting on MO. The concrete case is the LP/ILP formulation of integer factorization, for which I'm very certain to have a solution but don't know whom I could ask for cross checking. $\endgroup$ Commented Aug 30, 2013 at 17:02
  • 2
    $\begingroup$ @ManfredWeis: regarding your "LP/ILP" integer factorization: just implement it, and if it beats FactInt (cf. gap-system.org/Packages/factint.html) for sufficiently large and difficult integers, send me the code and I'd be more than happy to check. $\endgroup$
    – Stefan Kohl Mod
    Commented Aug 30, 2013 at 17:27
  • $\begingroup$ @StefanKohl: thanks for the offer, I had implemented a tiny example in Excel and it worked but I guess, my solution would be more of theoretical interest. And I doubt that checking an implementation can prove correctness. What I will do is to post my ideas (after some rechecking) as answer to my reference request and risk being downvoted. $\endgroup$ Commented Aug 30, 2013 at 18:12

I think an appropriate reaction to a failed reference request is to turn it into a successful one. Either provide a link to the appropriate published literature, or give a link to an available preprint. If the work is solid, providing an "abstract's worth" of information about the solution is acceptable in the answer. Taking option 2) which might be using the answer to publicize the solution seems inappropriate unless the intent is to narrow the reference request and ask if the solution is also in the literature. MathOverflow is a good place to exchange and (to a small extent) proffer ideas. It should not be confused with any form of peer review.


You must log in to answer this question.

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