7
$\begingroup$

It occurred to me that creating a contest, where (1) a statistical challenge or problem was completely defined, (2) judging criteria were unambiguously defined in terms of the winning answer, and (3) contest rules like "ending date," "no clear winner," etc. were unambiguously articulated might be a Good Idea.

I felt that such a contest could be a fun way to publicly express different approaches to solving problems in statistics, with the added advantage of identifying promising "best practices"—at least within the constraints of the contest rules. Something along the lines of Code Golf, but delimited by statistical interests and approaches.

I note that highly regarded questions and answers may present a statistical challenge, and wonder if we might add a [contest] or some similar tag, along with appropriate desiderata for posing such a contest.

I feel that the [puzzle] tag is getting at a different motivation, because it implies the OP knows the answer, whereas what I envision as the use for a [contest] tag implies the OP as criteria for judging a best of acceptable answers within a time frame, but does not know the winning answer a priori... indeed is hoping to see creative, possibly unorthodox, and certainly non-canonical answers.

| |
$\endgroup$
  • 1
    $\begingroup$ My first impression is that this isn't a good idea. For one thing, there is a real question about whether the ideal you describe can really be counted on to be implemented over time. It also seems like these questions would end up being the kind of publicity-seeking threads we've been (correctly) mixed on & have largely moved away from. $\endgroup$ – gung - Reinstate Monica Nov 8 '17 at 2:36
  • 3
    $\begingroup$ I find the idea of challenges attractive -- even intriguing (+1 for raising the idea); however I wonder about the extent to which a nontrivial problem could be completely defined and have unambiguous criteria. Would these be theoretical problems or data analysis problems or something else? The thread here that you link I think gives an excellent example of just how muddy such a problem may be (e.g. consider Russ Lenth's well-aimed disagreement with my answer there; while I was able to offer some justification in response, that's a good illustration of the kind of issues that may arise) $\endgroup$ – Glen_b Nov 8 '17 at 2:47
  • 1
    $\begingroup$ @Glen_b There's really good points... (a) I wonder if excluding specific types of contest a priori would be a reasonable way to approach (like we exclude some types of questions), (b) perhaps there could be sub-categories of contest, (c) I think there'd have to be a commitment on the part of the contest poser to judge by the criteria... perhaps contests might even require a bonus? Dunno... $\endgroup$ – Alexis Nov 8 '17 at 3:08
  • $\begingroup$ @gung Thank you for the intuitive response! :) I wonder if "counted on to be implemented" could be encouraged by appropriate standards in contest rules? (I.e. enough time, don't ask for a resolution to P v. NP or similarly tough problems, etc. :). As to publicity-seeking, might we not establish standards for high quality/unacceptable contests, just as we do for other kinds of questions? $\endgroup$ – Alexis Nov 8 '17 at 3:11
  • 3
    $\begingroup$ The Stack Exchange principle remains that the OP gets to decide, or not, on which answer is accepted. Others can express contrary views through voting, comments or answers. So no principle seems to be raised by someone posing a question to which they know the answer. On a different level, it is perfectly acceptable for people to answer their own questions: whether their answer is known at the outset is immaterial. So, I think any disagreement is over whether the wording contest or competition is congenial, or the flavour so implied to everyone's taste. Just do it and avoid those words? $\endgroup$ – Nick Cox Nov 8 '17 at 8:19
  • 3
    $\begingroup$ It sounds like a duplicate of stats.meta.stackexchange.com/questions/4464/… $\endgroup$ – Tim Nov 8 '17 at 8:52
  • 1
    $\begingroup$ Possible duplicate of Are puzzles on-topic on CV? $\endgroup$ – Andre Silva Nov 11 '17 at 0:57
  • $\begingroup$ +1 I like contests. Mainly because the questions are fun and educative. They sharpen my knowledge and understanding in a wider area than my own field. For this reason I always try to solve questions of a quarterly mathematical magazine (winner is either the most elegant or if unclear based on a lottery), which are very approachable. And now since a while I answer questions here, for the same reason. $\endgroup$ – Sextus Empiricus Nov 14 '17 at 18:08
  • $\begingroup$ Currently the bounty system already works a bit like this, creating attention for interesting questions. And I believe it has an effect. Additional tags parallel to the bounties, e.g. pointing at competition/difficult/unsolved problems in statistics/etcetera, might help. However, I wonder if it would not move too far away from the formula of this website. But then again, I do not understand what the stack exchange network wants to do with the growth and development of their websites, and how this idea would fit their policies, desires and perspective of sustainability of the platform. $\endgroup$ – Sextus Empiricus Nov 14 '17 at 18:16
  • $\begingroup$ @MartijnWeterings I am less interested in "creating attention for interesting questions," than I am in asking and answering questions in a form that invites creative, rather than canonical approaches to narrowly-posed statistical problems. But maybe that's already within the remit of CV... so your feedback (and others') is good to mull over, because maybe I am just interested in something we already do. :) $\endgroup$ – Alexis Nov 16 '17 at 21:33
  • 1
    $\begingroup$ I had not seen your question like that before. I believe it makes it stronger. Certainly, creative (or elegant) answers are to be appreciated, and promoting them is good. This fits well to what I mentioned in my first comment. In my second comment I worried about the stackexchange format, and I still do. The Q&A format is very powerful, but also very rigid. Possibly a meta website might work well. A website that communicates with the information on stackexchange, but being separate does not interfere with it. Then stackexchange could become more like a backend. $\endgroup$ – Sextus Empiricus Nov 16 '17 at 23:09
  • 1
    $\begingroup$ OK, thanks. This has been lingering in the close vote queue for a while. I'll vote to leave open. $\endgroup$ – gung - Reinstate Monica Nov 20 '17 at 19:12

You must log in to answer this question.

Browse other questions tagged .