10
$\begingroup$

I have a little dilemma. There is simple question that came up in my actual research that I'd like to ask, but I think it's forbidden by stackexchange rules, because it has more than one correct answer. The question (from exp. particle physics) would be something along:

What data-driven methods are there to estimate the QCD multijet background?

There are two, max. three methods that are well-known and can be described succinctly (and with references). But there might be more methods I don't know of, and these would probably be even more useful to me. I don't know all the techniques - otherwise I wouldn't be asking - so I don't know the total number of answers. But it's not open ended, in the sense that there is a small, finite number of answers.

Usually, I'd just reformulate the question slightly and pretend I didn't know the question had multiple answers:

How to estimate the QCD multijet background from data?

How does the technique work?

That's of course silly, and it makes the question less useful because I have to play dumb and can put less information in it. Someone will post an answer saying I have a misconception and explain two methods, and somebody else might post the third, so I'll eventually get what I want. I see this on Stackoverflow all the time.

(A third option: demand that the correct answer contains all techniques known to mankind. This is true to the word of the law, since there is only one right answer, but it's even more silly.)

I believe this question and its potential answer(s) are useful, yet I fear it will be closed and downvoted quickly. So, what should I do?

$\endgroup$
  • 2
    $\begingroup$ While the number of possible ways is non-trivial there are only a few usual ways---ones that won't bring on a hail of questions at a talk. Adding the work "usual" would, IMNSHO, be a good idea. $\endgroup$ – dmckee May 7 '13 at 18:54
  • 1
    $\begingroup$ Of course, questions about unusual methods are one topic as well, but you'd be best focusing on whatever property of the analysis you are trying to optimize. $\endgroup$ – dmckee May 7 '13 at 18:54
14
$\begingroup$

There's nothing wrong with a question having more than one answer. Many questions on SE sites do. The problem is with open-ended, soft questions, where there's a large and expanding or unbounded set of possible answers, and there is no metric to judge one against another.

I would recommend asking your question like this:

How to estimate the QCD multijet background from data?

(in question body)

I know of method X, Y, and Z, but these don't work for me because (details). Is there a method that gets around that problem?

or

Is there a more accurate way to estimate the QCD multijet background from data?

I know of method X, Y, and Z, but those give large uncertainties. (details) Is there another method that does better?

or some such thing - basically, identify why you're not satisfied with existing methods and ask for a better one. The trick is to phrase it so that you're looking for a solution to a problem you have, as opposed to phrasing it as though you're just trying to collect a list.

Even if you were to ask

What data-driven methods are there to estimate the QCD multijet background?

it'd probably be more or less okay because there is a small set of methods which could be described or summarized in one answer. Note that you don't have to require that an answer contain all known methods. Someone can provide an incomplete answer listing some methods, and other people can then post their own answers that build on that one.

$\endgroup$
  • 8
    $\begingroup$ In fact, having a small number of working solution from which the community can select the one most appropriate to the question is exactly the situation in most Stack Overflow question which are the model for the whole Stack Exchange network. $\endgroup$ – dmckee May 7 '13 at 18:59
  • 4
    $\begingroup$ Good answer! However I think in this case asking the general question might be more useful than addressing concrete problems in my work. The '(details)' are likely to be very obscure and won't help anybody in the future (a la "The foo fit in the 2nd iteration of my procedure does not converge when using bar jets"). Also I don't want to be falling into the XY problem, where my concrete problems would be fixed by using a completely different approach. $\endgroup$ – jdm May 7 '13 at 19:22
  • 1
    $\begingroup$ Wow, this is exactly the reasonable attitude and good judgement distinguishing legitimate questions which can naturally have multiple answers against (too big and not useful) list questions I appreciate and agree with too, +1 :-) $\endgroup$ – Dilaton May 7 '13 at 21:57
  • 2
    $\begingroup$ My thoughts exactly. A question can have many correct answers as long as they are indeed testably correct. As a matter of fact, those questions are probably the best ones: where you can look at things in a few different but fairly equivalent ways; this makes for better insights and more creative solutions. I think the problems mainly arise from questions that can have multiple answers, but none of them are objectively more correct than the others. $\endgroup$ – Wouter May 8 '13 at 0:18

You must log in to answer this question.

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