2
$\begingroup$

In this question the user asks an interesting and natural question: "A computer has produced this list of numbers. I want to know how it does this." Unfortunately, this question cannot be answered without using the word "probably". In fact, the answer to the question as posed is "No, you can't". However, simply saying "No, you can't" is not good enough. It should be "From a mathematical point of view no, you can't and here is why, but from a real-world point of view this is what they probably did". Personally, I would be intrigued to see the answer of someone who actually knew their stuff, some who can say "They probably did this, because that is the standard approach", or "Some people do this, others do that. Going by your samples, it is probably this third, obscure option." Note that the approaches would all be mathematical, and so such an answer would be well within the remit of this site.

However, no such expert appeared before three people said "Oh, well, they maybe did this"...and so the question has been closed. Because if you try to answer it as a purely mathematical question it only admits opinion based answers (unless you say "No, you can't! And here is why..."). Personally, I think closing it is unfortunate.

*My point is, if we are here to help people understand stuff then closing this question is pointless - it admits an answer, just not the kind of answer the OP wants.**

My question is: what should be done in this and in similar cases? You see, the three answers are all useless and it is not unlikely that we will never get someone who knows enough to answer this question satisfactorily. So is just closing it better? This would stop spammy internet-point-seeking answers...but that isn't the point of this site.

*If we are here for internet point, then sure just close it.

$\endgroup$
3
$\begingroup$

There are some issues with this post on Maths.SE. It could possibly be improved to narrow it down to make only the type of answer you mention a good fit for it.

Then, it allows for the nice answers you refer to. But, I consider such "reverse algorithm design" questions to be better suited for some other site of the SE network (perhaps just SO, or CS.SE).

I say this because the question will not, at its core, be about mathematics. It will be about the algorithm.

And while there is some overlap between these two (e.g. prime factorisation algorithms) I don't think this question fits into that realm (which for me broadly ranges over algorithms built on solid understanding of graduate-level mathematics).

So even though I agree with you that this question admits, and to some extent deserves, better answers than "Hey this randomly fits your small data set", I don't think that Maths.SE is the place for this question with those better answers.

$\endgroup$
  • $\begingroup$ I agree that Math.SE is not the best fit, but then of course the solution would be to migrate the question rather than close it. However, migration will yield similar problems to the ones that we have here, just at a different SE network. So in a certain sense this isn't a solution. (What I mean is, yes your answer is perhaps the best fit here, but it side-steps the actual question of what we should have done in the first place, because it could have been answered here and it does fit, what -I choose to believe- you are saying is that there is somewhere better. But it still fits here...) $\endgroup$ – user1729 Oct 3 '13 at 18:00
  • $\begingroup$ (Does my comment make sense? Please say if it doesn't.) $\endgroup$ – user1729 Oct 3 '13 at 18:00
  • 1
    $\begingroup$ @user1729 Your response makes sense. I however contend that, in the most optimistic view, this question is on the border of on-topic for MSE. Personally, I feel that it's not a question for MSE. On those other sites, the "we can't be sure"-card is less likely to be played so prominently. The only other thing that crossed my mind for attracting better answers is to put a bounty on the question. However, I'm sceptical about the number of MSE contributors that could actually provide the answer you want. In fact, that's (one of) the main reason(s) why I consider it off-topic for MSE. $\endgroup$ – Lord_Farin Oct 3 '13 at 21:16
0
$\begingroup$

Have you looked at the history of the question? I'm pretty sure that some of those close votes were cast against revision 1 or 2, which asked people to extrapolate from one sample (although the fact that that's what was asked is much clearer with the hindsight afforded by later edits - "Unclear what you're asking" was perfectly fair).

The question has been improved by the addition of a dozen more samples, and it's not only a bit clearer what the OP is looking for, but it's feasible to do some analyses and potentially come up with an answer that really deserves "probably" rather than "possibly". I've just cast the 5th reopen vote.

In short, this seems to be the system working as designed. The question is closed, improved, and reopened.

$\endgroup$
  • $\begingroup$ I disagree that adding more samples improves the question. It would allow someone to get a better approximation, but my point is that mathematically that is all we can get - an approximation. The question needs dragged into the real world by a good answer (and so perhaps MSE isn't the place for the question) rather than a better way of posing it. $\endgroup$ – user1729 Oct 4 '13 at 9:26

You must log in to answer this question.

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