Monday, June 28, 2010

TheoryCS discussion site: A call for help.

Many of you are familiar with (MO), the discussion site where (mostly) professional mathematicians come to discuss research problems, get questions answered by experts in the area, and have very enlightening discussions on a host of topics in mathematics.

The conversation is at a very high level. The moderators (and the community at large) do an excellent job of pruning out the posters fishing for homework answers, the vague questions with answers that any google search could return, and off topic questions that are better served by a more amateur-centric forum. What remains are questions that are at least at the advanced graduate student level, and that frequently spur open-ended discussion among the advanced researchers in the area.

It's a dream site: it has much less clutter than sci.math (or comp.theory for that matter), and because of modern tagging and filtering technology, is a lot easier to navigate than the old Usenet.

While theoryCS questions are welcome at MO, they sometimes get neglected for a lack of critical mass in the audience. Many in the theoryCS community participate in MO, but it would really be ideal to have our own discussion site that's structured along the same lines with the same intent:
to be a site for professional researchers to come together and ask/answer questions about research.
If you're a student, this could be an invaluable resource to learn about different aspects of TCS, and get to know researchers in the area. If you're an active researcher, this is a great place to post queries about topics that you might not be familiar with but need in your own work.

We're in the process of creating such a site, courtesy of Anand Kulkarni (aka polybot). Stack Exchange is the software system and the organization that maintains discussion sites like this, and has built a system to create, discuss and vet sites to ensure that they have enough critical mass to sustain themselves.

The theoretical computer science site is here. The way it works is as follows:
  1. We start with a 'define' phase with prototypical questions that are right and not-right for the site. This has been completed, with a sufficient number of questions that people have voted for.
  2. We move on to the 'commit' phase, where we are now. Here, we need commitments from people (with their actual names attached) that they will participate in the site once it goes into beta testing. We have a number of commitments so far, but we need much more in order to move to phase 3, which is
  3. Beta testing, where the site goes active and we see if we can sustain it with questions and discussions for a while. If so, 
  4. The site gets created. 
 This is a great opportunity to create a site that can truly be our own, and that would be  a go-to location for all aspects of theoretical computer science, whether it be algorithms, complexity theory, quantum computing, game theory, programming language theory, logic, or whatever. Tagging and filtering means you don't have to get swamped in a sea of posts on topics you don't care about, and if you've used MO, you know how easy it is to have the site only display topics that are relevant to you.

What should you do ? Go to this site, and commit to participating in the beta. Committing costs nothing - it's literally one click after you authenticate. Then all you have to do is spread the word so we get enough people to move to the beta phase.

If you're skeptical about this (or maybe have wasted too much time on comp.theory in years gone by knocking down P vs NP cranks), do go to MO and see what a theoryCS site could become.

And if you're a theoryCS blogger with your own following, please spread the word ! you all know who you are :)

p.s there's some confusion about exactly how many people are needed to get to the next phase. The answer is that it's not based on numbers, but based on the reputation of the people committing (as measured by their activity on related sites - but not MO sadly). Most folks in our community are unlikely to have large reputation in the related (mostly programming) websites, so we'll need a good number of people (probably in the low 100s) to get there. Hence this appeal (note that if you commit and then use the 'share' link to get someone else to sign on, your "reputation" increases, and that improves the overall progress of the site in a cascading effect)


  1. Why is there a need for 1000 users to commit to the site, before it's initiation?
    It sounds too high for me.
    Maybe 300 would be enough.

    Do you think we'll get 1000, at all?

  2. I sincerely hope we wouldn't need that many. My understanding is that the calculations that determine beta readiness are evolving and also depend on user reputation rather than just counts. So maybe we'll get there faster.

  3. Are there even 1000 active theoretical CS people (with any significant reputation) online? That said, I think it's a very interesting initiative, and would benefit grad students such as myself.

  4. Well, not all the 1000 members should be professional researchers in TCS. Even 20% would be fine I suppose.
    (Since, there is a natural filtering mechanism there, this would work fine I think.)

  5. Are you sure a separate exchange is better than being a part of mathoverflow? Is this the majority opinion of the computer scientists already on mathoverflow? Personally, I believe that theory is a branch of math. Or that math is a branch of theory. Whatever. They should be more rather than less integrated.

  6. Is there a way to convince the people who operate this StackExchange site that the site would get substantial traffic even if less than 1000 people commit? At this point, I don't really think we'll ever meet the initiation requirement because as you point out, most TCS people don't have high reputation numbers.

  7. To those concerned about commit counts - we've managed to maintain a fairly high commit rate compared to proposals ahead of us in the queue. So don't lose heart. I'm trying to get more stackoverflow and mathoverflow types involved as well.

    For those who feel that we should merely absorb into MO, I'm not opposed to that if this doesn't work out. The MO folks are quite friendly. However I haven't seen major participation (in number of people) from the theoryCS side on MO as yet and a more focused home, if it works, might be more effective. Lets see how things play out.


Disqus for The Geomblog