Recently, I read a question on difference between stackexchange and overflow and I found out that overflow is more focused on difficult questions. Some time ago, I asked on a stackexchange a question (How many pure ternary operators are there?), that is probably too difficult to solve/answer and requires a lot more energy, focus and understanding than people on stackexchange are willing to put into it.

So, learning this new information, I considered asking the same question on overflow again. Would that be Ok or would that be considered a simple duplicate?

It is true that I could try to draw more attention to it by putting a bounty on it, but that does not guarantee results and must be paid upfront, so I would rather try other ways first.

  • 3
    $\begingroup$ No, I don't think so: it's not about difficulty. People ask questions equivalent to famous open problems at MSE (and the lemmings upvote those questions, slap-happily), just because they don't know they are famous open problems. Overflow is about research problems, people there can judge if a question is difficult (so I'd like to discourage your asking about "how many pure ternary operators are there", I'm not surprised you aren't willing to pay for an answer). $\endgroup$
    – user436658
    Sep 11, 2017 at 8:56
  • $\begingroup$ @ProfessorVector It is not famous open problem, it is designed by me and I could not solve it on my own or find any other published work on that topic. It might be open problem someone is dealing with, but definitely not famous. $\endgroup$
    – TStancek
    Sep 11, 2017 at 9:03
  • 3
    $\begingroup$ @TStancek I suggest you post a question on Mathoverflow's meta asking about the appropriateness of posting the question there. Include a link to your original post, relevant references, any further details that may clarify why you think it may be appropriate there. If the response you get is positive, then go ahead with the post. Questions here disappear quickly from the front page, so not having received an answer yet is not necessarily an indication that it is of the appropriate level. $\endgroup$ Sep 11, 2017 at 11:50
  • $\begingroup$ @AndrésE.Caicedo Great idea, thank you. $\endgroup$
    – TStancek
    Sep 11, 2017 at 11:52
  • 3
    $\begingroup$ How long ago is "some time ago"? If it's more than a week, it can't hurt to try it on MathOverflow, BUT be sure to link the question on each site to the question on the other site so everyone's in on it. $\endgroup$ Sep 11, 2017 at 13:07
  • 2
    $\begingroup$ BTW I suggested this at meta.MO some time ago - but it seems that most people disagree with that idea: Thread for asking about suitability of math.SE question on MO? $\endgroup$ Sep 11, 2017 at 13:12
  • $\begingroup$ I would doubt that a lack of time and energy acounts for the failure to get a response about the number of "pure" ternary operators. Perhaps if you linked to your Question, rather than merely mentioning it, we could better advise you. $\endgroup$
    – hardmath
    Sep 11, 2017 at 20:07
  • $\begingroup$ @hardmath math.stackexchange.com/questions/2337184/… $\endgroup$
    – TStancek
    Sep 11, 2017 at 20:45
  • $\begingroup$ BTW if you simply write overflow, some people might think that you are asking about Stack Overflow and some people might think that you are asking about MathOverflow. (I assume it is the latter - which is why I added the (mathoverflow) tag.) $\endgroup$ Sep 11, 2017 at 22:01

1 Answer 1


I've looked at the Question at issue, and I think it not in good shape for Math.SE or MathOverflow.

It asks about "how many" functions of three variables cannot be expressed as a composition of two bivariate functions in a particularly constrained way, namely that $T(x,y,z)$ cannot be expressed as either of these forms:

$$ f(x,g(y,z)) \;\text{ or }\; f(g(x,y),z) $$

So the Question is somewhat related to Hilbert's Thirteenth Problem, which also concerns the possibility of expressing a function of three variables as a composition of finitely many functions of two variables.

But in addition to restricting how the compositions are taken (preserving in particular the "order" of arguments above), the presently posed Question fails to say anything about the underlying domain, except to name it:

$$ T:A\times A\times A \to A $$

There are two versions of Hilbert's Thirteenth Problem, one for (real) continuous functions and one for algebraic functions, but by asking "how many" pure ternary functions (or operators, if you prefer), the OP is probably thinking about $A$ as a finite set.

Given the fact that Edits and a posted Answer have already occurred on the original Question, my advice is to post a new Question on Math.SE with the necessary gaps filled in about the domain $A$. Such a new Question can of course link back to the old one for context.

  • 4
    $\begingroup$ "Such a new Question can of course link back to the old one for context." I would say, such a new question must of course link back to the old one for context (and I would add that the old question must link forward to the new one). $\endgroup$ Sep 11, 2017 at 21:55
  • $\begingroup$ @Gerry: I stand corrected ! $\endgroup$
    – hardmath
    Sep 11, 2017 at 22:05

You must log in to answer this question.

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