What is the right site on the Stack Exchange network for computer science questions?

None of the answers I can think of are remotely satisfying:

An almost identical question has been asked before, but that was before SE2.0 which has completely changed the scope.

On a closely related note, I've searched for a number of keywords where I would expect some activity on a computer science Q&A site. (Note: there is definitely a bias in the keywords related to my background — suggestions for an objective keyword list would be welcome — but I wrote the keyword list before I did any searching.) These are topics that are typically taught in undergraduate or graduate courses (so basic questions about them are out of scope for TCS.SE). I'm only considering hits in questions, since here I'm looking at the kinds of questions people ask, not what people might mention in passing in answers.

  • strongly normalizing (0)
  • lambda calculus (43)
  • theorem provers (debated on TCS): coq: (13), agda/agda2: (5), twelf (0), isabelle (0)
  • linear logic (1)
  • abstract interpretation (16 hits total, few in questions, though most are good CS questions)
  • model checking: (3, not particularly science-y; some questions without the tag are more science-y)
  • bisimulation (0; 3 answers)
  • basic block (71 hits, ranging all over the board; I haven't counted the questions)

What this tells me is that a lot of subjects that I learnt about in CS courses don't receive any questions. There seems to be a gap between the first two undergraduate years or so (mainly introduction to programming and algorithms), which get reasonable coverage at SO, and graduate students' first research projects, which is where TCS.SE starts.

Computer science has more applied topics that don't fit well on Stack Overflow either. For example, questions about network protocols don't seem to fare well.


I strongly think that an own Computer Science site is the only way to go. I have the following reasoning:

  • If you look at any computer science university course, you see that programming is about 10% and theory about 20-40%. The rest is somewhere in between.
  • SO is mainly used for small every-day programming issues, typically tied to specific languages. In contrast, computer science is more about general questions, e.g. properties that any type system with subtyping has. Therefore, most CS questions will be inherently out of scope on SO.
  • cstheory, the only live CS related site, is very narrow in its scope: is is only for theory (maybe 10% of computer science work/research?) and only for researchers (a really exotic breed). Therefore, it is not the right place for most computer scientists. (The question wether its narrowness is a good thing is of no import here).
  • A general computer science site could subsume other, more specific proposals that have problems taking off, at least until they have enough traffic to justify moving to an own site (what it maybe what should have been done pre-cstheory, but again, that is beside the point).

Note that I use the term "theory" as a synonym for theoretical computer science, a proper subset of computer science, and not in its popular meaning (i.e. the opposite of "of immediate practical import").

| improve this answer | |
  • 2
    The second argument is plain wrong. There're lots of not-so-every-day questions on SO, for example in [algorithm], [language-agnostic], [math], [programming-languages], [data-structures] and other tags. In general, nobody's gonna push you out, if the question can be understood by average CS graduate (and the ones that require special knowledge usually also stay). – Nikita Rybak Feb 21 '11 at 12:09
  • It doesn't mean there's no room for a new place (there probably is), but I'm yet to see a real CS question actively repelled by SO community. – Nikita Rybak Feb 21 '11 at 12:20
  • I have not been too active on SO, but its FAQ says: "a specific programming problem, a software algorithm, software tools commonly used by programmers, matters that are unique to the programming profession", none of which covers all CS - TCS. Of course "software algorithm" can be interpreted arbitrarily broadly, but at least hardware guys are certainly OT. Myself, I would not ask a "serious" CS question on SO, simply because the noise of nonsense even on simple programming questions is disturbingly high. You don't get CS people to ask if the felt signal/noise ratio is low, so no repelling. – Raphael Feb 21 '11 at 16:37
  • Perceived signal/noise ratio is another (maybe valid) issue, but people do ask and answer questions on data-mining, robotics, grammars, complexity, EBNF and other CS topics here. (Can't tell if they're sufficiently 'academic', but they're here) – Nikita Rybak Feb 21 '11 at 17:50
  • 1
    @Nikita: The CS proposal cites a few questions that were closed on TCS as too basic. Do you really think that “How to prove that Satisfiability is in DTIME(n^2)” is SO material? – Gilles 'SO- stop being evil' Feb 21 '11 at 19:00
  • @Gilles There're more than few questions on complexity theory and some thorough answers on SO already. My point is, nobody is pushing you out. It may be true that some CS professor isn't dying to be on one site with "filthy engineers", but that's not because those engineers are gonna kill him, eat his brain and close his questions. So, don't blame engineers. – Nikita Rybak Feb 21 '11 at 19:28
  • @Nikita: I'm not blaming engineers! But don't blame the CS professor for not having time to wade through a lot of questions that neither interest him nor particularly call on his expertise to find the relatively very few questions in his subfield of CS. – Gilles 'SO- stop being evil' Feb 21 '11 at 19:32
  • @Gilles That's actually a bad example: difficulty to search for Haskell or Lua or Java isn't a reason to create a new site for them. Not to mention, it can easily be (if not already is) addressed through stackapps. In fact, there're many apps which will allow professor to not resort to search at all. – Nikita Rybak Feb 21 '11 at 19:51
  • @Gilles Forgot to mention, tag sets were also created to address topic-following problem. Just in case you're interested. (Although I still find stackapps more convenient) – Nikita Rybak Feb 21 '11 at 20:02
  • @Nikita: Actually you're making my point for me: even Norman has a hard time, and his interests are a lot closer to programmers' than the average CS professor. – Gilles 'SO- stop being evil' Feb 21 '11 at 20:02
  • @Gilles That question is more than two years old, from the beta age of Stack Overflow :) Try posting it today. – Nikita Rybak Feb 21 '11 at 20:08
  • 2
    @Gilles That being said, I do think CS professors can have their own site (as I said in my second comment here). Just not becouse programmers bully them, or overflow with 'too engineering questions' - they don't. But because CS professors prefer to have their own company and do not consider normal programmers to be 'colleagues' or 'peers'. (At least, the ones I knew) – Nikita Rybak Feb 21 '11 at 20:10
  • Would you consider the guy screwing together a car a peer of his boss, or of a mechanical engineer, or of a physicist? This is the spread we are talking about here. I never claimed anybody "pushed" CS people "out" of SO, I only claim SO can not serve the purpose of talking amongst computer scientists. Programmers have different view and perspectives, that is proven time again. That is helpful if something has to be coded up, otherwise it gets in the way. – Raphael Feb 21 '11 at 21:49

Well, we pushed for cstheory to include these kinds of questions, but that community has basically rejected the initiative.

(visit http://meta.cstheory.stackexchange.com for background on this if you are curious)

So, for now we are saying that these sorts of questions should mostly work on Stack Overflow and we have an explicit migration path on cstheory back to SO for that reason.

| improve this answer | |
  • 7
    Your answer misses the point. There is too much room between SO and cstheory.SE, as shown by Gilles. The fact that the community on csthery.SE does neiter want to extend to non-research level nor to non-theory CS (theory in the academic sense) is true, but of no import here. You could as well demand that SO extend its scope accordingly, what would be similarly meaningless. – Raphael Feb 21 '11 at 11:38
  • 2
    As Raphael says, there is a large gap between SO ans CSTheory. Ok, it's your site, you can do what you like. But you're missing a likely audience. – Gilles 'SO- stop being evil' Feb 21 '11 at 19:02
  • 1
    @gilles the problem is the cstheory community rejecting this initiative. Go speak to them about it. They will be marginalized in our network if they continue down they path they are on, and I'm losing any sympathy I had for them, personally. – Jeff Atwood Feb 21 '11 at 20:48
  • 8
    @Jeff: How is a general CS site the problem of a community who wants a MathOverflow analog about CS? – Gilles 'SO- stop being evil' Feb 21 '11 at 20:53
  • 4
    Another CS community would be good. It is not a duplicate of SO nor TCS, in my opinion. – Dr Beco Apr 3 '11 at 22:22
  • 2
    Revisiting this, considering that in the latest podcast you seem to feel that “there are technical computer science questions on Stack Overflow that aren’t being handled correctly and have no other place to go”: is it still SE, Inc. policy that all computer science belongs on Stack Overflow? – Gilles 'SO- stop being evil' Jun 12 '11 at 22:23

(some get closed, which I think shows they're not getting the right audience)

I'm gonna give pretty much the same answer here that I gave in the question you linked to: prove it. Current stats show that about 8% of the [computer-science] questions are closed on SO - that's high, but not obscenely high. Can you provide an example of some CS questions that aren't getting the right sort of attention on SO?

| improve this answer | |
  • 2
    Have you looked at the questions tagged [computer-science]? They seem to mostly be tagged that way because they're asked by people taking CS classes, not because they actually involve computer science. – Gabe Apr 4 '11 at 0:12
  • @Gabe: what should I be looking at then? Where are the real CS questions that are being closed? – Shog9 Apr 4 '11 at 0:16
  • I'm afraid that real CS questions probably aren't tagged as such. The most recent question I remember was something like "How are fractional numbers most effectively encoded in lambda calculus?" but I can't find it. – Gabe Apr 4 '11 at 3:35
  • @Gabe: that question got bounced around a bit... It ended up on Math.SE. Note that the original question consisted of nothing but the title followed by, "Don't ask why I need it" - which is a pretty good way to get a question on any topic off to a rough start. – Shog9 Apr 4 '11 at 4:43
  • Yeah, it was a pretty bad question, but the point is that it is a CS question that wasn't tagged that way. – Gabe Apr 4 '11 at 5:32
  • 3
    That's a highly loaded request. For one thing, CS questions are hard to locate (even harder, I realize now, than I thought when I posted this question). Most CS questions aren't tagged computer-science, that's an overgeneral tag that tends to attract poor questions. I'm looking for a few examples now, but they'll be easy to dismiss as anecdotal. – Gilles 'SO- stop being evil' Apr 4 '11 at 21:48
  • stackoverflow.com/q/3886092 (looks valid but I don't know for sure, might in fact be CSTheory material; bounced to MO where it got no good answer either). stackoverflow.com/q/4440261 , though in general I think algorithms and data structures get better answers than other areas (e.g. programming language theory, complexity) stackoverflow.com/q/4623104 stackoverflow.com/q/5056498 (not a great question but deserves better) – Gilles 'SO- stop being evil' Apr 4 '11 at 22:02
  • @Gilles: thanks for the examples... The first thing that strikes me about them is that all but the last are reasonably difficult problems: this tends to limit the audience to those with the necessary skill and the necessary time. Which does lend some credence to your "wrong audience" suspicion, though perhaps the solution is simply to make it easier for the right users to find the right questions on SO itself. – Shog9 Apr 4 '11 at 23:39

I'm having trouble differentiating between Computer Science, Information Theory, and Software Engineering. I think the people who design academic curricula are also having the same trouble.

In any event, I think that for any of these, there needs to be a forum where in depth and extended discussion gets more highly valued than quick answers to quick questions. I think SO is biased towards quick questions and quick answers, and rightly so. So I think some kind of alternative forum is better.

At the same time, some of the "theory" discussion groups I've seen over the years tend to get clogged up with cranks, people who cannot accept that their ideas have failed to obtain recognition because the ideas lack merit. Unorthodox thinkers can certainly add value to theoretical discussions. But cranks do not. It may be hard to draw a line between cranks and creative unorthodox thinkers.

This isn't really an answer, but I hope it triggers one.

| improve this answer | |
  • You are right, computer science is still in the process of defining itself. Everybody agrees that SE and CS are different things (much like physics and mechanical engineering), but people are hesitant and unsure where to draw a line. – Raphael Feb 21 '11 at 16:40
  • 1
    In other countries, CS has trouble distinguishing itself from math. CSTheory and Math Overflow aren't invaded by cranks, I think the way SE works is very good at keeping them out. – Gilles 'SO- stop being evil' Feb 21 '11 at 19:04
  • Math is the handmaiden of the sciences, but it doesn't belong to any one of the sciences more than it belongs to the others. I have no trouble distinguishing physics from math, or biology from math, or linguistics from math. Why should CS be any different. – Walter Mitty Feb 23 '11 at 15:02
  • In yet other countries, there is another discipline that lacks a good name in English. Consider the French word "informatique". I'm not a French speaker, but I think this word conveys nuances that no English term conveys. It's much more subtle and pervasive than "computer science", but it isn't math or even linquistics. – Walter Mitty Feb 23 '11 at 15:02

You must log in to answer this question.

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