6
$\begingroup$

I am thinking about submitting a feature request that will show us to "some" questions from CS.SE on cstheory from time to time (maybe with a small sign that indicates it is from CS.SE).

E.g a new question with complexity-theory tag and and "high quality" is shown among our questions every week or so.

We can discuss what we mean by "high quality". I was thinking of something like questions with +10 total votes.

To be clear, the question will remain on CS.SE, we will not have a duplicate copy on cstheory. If you click on the question you will be taken to the question on CS.SE. The only thing is that from time to time (e.g. when we are not getting too many questions) some "high quality" theoretical questions will be shown among our own questions.

The main benefit is that we will not miss interesting theoretical questions posted there, it will help questions on CS.SE get answered.

What do you think?

$\endgroup$
11
  • $\begingroup$ like the "thinking outside the box" but the reality is that se software is quite mature and not subj to much major changes anymore, even highly popular feature requests/ proposals as seen on the sites-wide meta. however, more mod attn/ cooperation about "promoting" (migrating) cs.se questions is quite feasible. another possibility is just making/ maintaining a list of "high quality (open) cs.se questions" with links & posting it in meta. agreed this is a worthwhile aim/ prj overall of bringing theoretical expert attn to some select cs.se questions. $\endgroup$ – vzn Sep 5 '14 at 2:15
  • 2
    $\begingroup$ Sounds like a good idea. Do you think this is a feature we could actually get? $\endgroup$ – Lev Reyzin Sep 5 '14 at 14:54
  • 1
    $\begingroup$ @Lev, I don't know, but I am hopeful. $\endgroup$ – Kaveh Sep 5 '14 at 19:41
  • 5
    $\begingroup$ It really depends to definition of high quality, e.g how does computer work, how to show that some language is regular, ... are perhaps high quality in cs.se with 10+ upvote but I don't think they are also high quality questions on cstheory. $\endgroup$ – Saeed Sep 6 '14 at 14:07
  • 2
    $\begingroup$ There are only 8 unanswered questions on cs.se with more than 10 votes. I've gone through these questions and feel that each question is better than an <i>average</i> question on cstheory.They certainly deserve more attention. $\endgroup$ – Jagadish Sep 12 '14 at 18:12
  • 1
    $\begingroup$ The existence of this proposal is concerning. Does Theoretical Computer Science not have enough questions on its own? That aside, I don't think that this "cherry picking" (with the particular metric you'd want) can be automated in any way. The best idea is probably for some users to monitor Computer Science and link interesting questions in Theoretical Computer Science Chat or tweet them at Theoretical Computer Science's Twitter. $\endgroup$ – Raphael Oct 28 '14 at 12:57
  • 1
    $\begingroup$ If/When you propose this to SE, keep in mind that the "elitist" scope definition and (nowadays) subsumption by Computer Science is not always well received. Proposals such as this may very well come across as attempts to save a flailing site by adding a new, "elitism"-driven feature (cherry picking). You want to avoid this impression. $\endgroup$ – Raphael Oct 28 '14 at 13:03
  • $\begingroup$ @Raphael, As I have written I saw some nice complexity theory questions on CS.SE remain unanswered and come up with this idea to help. I don't understand why you are concerned (and quite negative) about the idea, I think it can be a win-win for both sites compared to the current state. (And no, the idea that we should frequently visit CS is idealustic, it does not work and it is not going to work.) $\endgroup$ – Kaveh Oct 29 '14 at 2:36
  • $\begingroup$ ps: btw, did you come here as a result of a related discussion somewhere else? $\endgroup$ – Kaveh Oct 29 '14 at 3:02
  • $\begingroup$ @Kaveh I see, I did not quite catch that angle. The correct solution is of course to answer the question(s) on Computer Science. It's certainly true that complexity theory experts that visit only Theoretical Computer Science but not Computer Science won't even see the question. That's too bad. I still think that user action (be it via chat, Twitter or reposting) is a workable solution. I am working on a proposal for Meta Stack Exchange that might help here, but that's an even bigger one. (@PS I think the post was linked to me in chat. We were talking about the merits of crossposting, I believe.) $\endgroup$ – Raphael Oct 29 '14 at 6:38
  • 1
    $\begingroup$ @Raphael, :) btw, I said that the answer will be posted on Computer Science, there won't be a separate copy of questions at all, just a link on cstheory to the Computer Science copy. $\endgroup$ – Kaveh Oct 29 '14 at 14:43
5
$\begingroup$

I don't think this is the right way to go about it. A question on a theoretical subject getting a lot of upvotes on CS is not a good indication that it is a research-level question and therefore appropriate for CS Theory. Conversely, a question on CS having a small but positive score is not a good indication that it is not research-level: perhaps most of the people over at CS just thought it was too specialized to be interesting, or didn't see its significance.

If you're aware of a collection of unanswered questions on CS that would be appropriate here, why not just flag some of them for migration or talk to the CS mods about it? Sure, that requires manual work, which you wanted to avoid but, on the other hand, if some automated system shifts them over here, we'll be overwhelmed on the day it's switched on (OK, it could be rate-limited).

$\endgroup$
6
  • $\begingroup$ afaik there is an age limit on question migrations, not sure exact span. afaict k. is not proposing every question that fits superficial criteria such as voting. it would involve some subjectivity/ selection. aka "content curation". have noticed some/ few high voted/open theoretical questions on cs.se or which answers do not seem to cite any literature, have deeper theoretical elements etc. $\endgroup$ – vzn Sep 12 '14 at 15:11
  • $\begingroup$ @vzn If there's an age limit on migrations, removing that would be a much smaller change to the site software than the one proposed in the question. $\endgroup$ – David Richerby Sep 12 '14 at 15:13
  • $\begingroup$ as stated above, after long observation of this dynamics, dont think that any proposals to chging site features will fly with se or are the way to go, but like the general idea of someone further curating cs.se content for tcs.se beyond what mods have already done so far & think it has potential. its possible that mods could be part of that. part of this could be managed in chat rooms but which tend not to be very busy/ active esp on tcs.se where chat is mostly inactive. $\endgroup$ – vzn Sep 12 '14 at 15:15
  • $\begingroup$ @vzn I don't see how you propose for this content curation to happen without changes to the site software. Suppose a set of questions on CS.SE are identified that are interesting and relevant to people here on TCS.SE. Then what? $\endgroup$ – David Richerby Sep 12 '14 at 15:23
  • 1
    $\begingroup$ to me its just about (qualified/ expert) humans making lists that are of interest to other humans & bringing stuff to attn. there are many ways/ mechanisms to do this that are already feasible/ simple. meta, blogs, chat rooms, etc... have blogged about interesting cs.se/ tcs.se questions. note while feature-request is an official tag and se has sites like meta, one can see feature changes in the se software (esp major ones) are now rather quite rare, even if asked for by high rep users and/or mods. so se feature chg angle seems as "red herring". $\endgroup$ – vzn Sep 12 '14 at 15:29
  • $\begingroup$ @vzn That seems like a sensible suggestion. I recommend you turn it into an answer. $\endgroup$ – David Richerby Sep 12 '14 at 15:36
1
$\begingroup$

This is an intriguing idea. Cool! It seems like it would be a good experiment.

The tricky bit is to select the criteria for which questions would be shown. I can see two plausible possibilities:

  • Alternative 1: Have a set of tags which are eligible for this, e.g., algorithms, complexity-theory, computability, undecidability, etc. Any question with vote total of 10+ and with at least one tag in the whitelist would then be eligible to be shown on CSTheory.

    This might whittle down the list of questions to those that are related to theoretical computer science.

    However, it has the problem that many of the questions selected in this way will not be research-level. There are a lot of questions on CS.SE with 10+ votes that are not research-level questions in theoretical computer science, and they might be unwelcome on CSTheory. Take a look at the top questions on CS.SE sorted by vote total, or the top questions on CS.SE with the complexity-theory tag and you'll see what I mean. Only a small fraction of the highest-voted questions there would be a good fit for CSTheory; the majority would not be useful to display on CSTheory.

    I looked at the top 50 highest-voted questions on CS.SE with the complexity-theory tag. All of them have a vote total of 10 or higher. By my own highly personal assessment, about 10 of them look like they would be good to display on the CSTheory.SE front page. The others don't look like great candidates (they are elementary, of narrow interest, or were about proving NP-hardness of a specific problem and the question has already been solved). A 20% hit rate isn't great. But this is my own personal taste, and others may differ. I encourage readers to do their own count, see what they get, and post their own evaluation.

  • Alternative 2: Allow moderators to select which questions are eligible for display in this way. This could be indicated, for instance, via a moderator-only tag that's added to the question on CS.SE, which triggers it to be shown on CSTheory -- analogous to the "featured" tag.

    This would require some manual effort from moderators, but allows more control over which questions are displayed, and enables selecting questions which are research-level and which relate to theoretical computer science. (A minor difficulty with using a special tag is that, if the question already has 5 tags, it may not be possible to add a 6th tag.)

$\endgroup$
3
  • $\begingroup$ There might be a few unsuitable questions that might pass whatever filtering criteria we use, but as long as that is infrequent I don't think it would bother us much. The reason I am suggesting this is that I was looking at top-voted unanswered complexity theory questions on Computer Science and all of them seemed OK for cstheory. I am open to other filtering algorithms (or even machine learning). We can experiment with them and see how well they work. $\endgroup$ – Kaveh Sep 12 '14 at 0:30
  • $\begingroup$ @Kaveh, I've edited my answer to reflect that you did mention the idea of tagging in your proposal. Thank you. I'm not sure it's just a handful of questions that wouldn't be suitable. Can I suggest a little exercise? Take a look at the top questions tagged complexity-theory on CS.SE (link is in my edited answer), and do your own count of what fraction of them you would want to see displayed on CSTheory.SE. I'd be curious to see what count you get. I've edited the answer to include what I got, when I did that exercise. $\endgroup$ – D.W. Sep 12 '14 at 0:51
  • 1
    $\begingroup$ I agree that Alternative 1 is probably useless; depening on the upvote cutoff, you'd get either most questions on the site or only "hot" questions (it's rare that a deep question gathers more than five votes). The data explorer can be used for further investigation. <cs.SE mod>Regarding Alternative 2, I'm definitely not available for this. Users interested in CS content can monitor Computer Science just as well, with the means available.</cs.SE mod> $\endgroup$ – Raphael Oct 28 '14 at 13:01

You must log in to answer this question.

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