Sunday, May 18, 2008

Visioning Workshop

James Lee starts his guest posts from Seattle and Victoria.

On Saturday, SIGACT, in conjunction with the Computing Community Consortium, held a workshop on Visions for Theoretical Computer Science. The goal of the workshop was to produce "vision nuggets" about exciting research themes in TCS that could have a large impact in the future. In other words, to craft PR materials that advertise TCS outside the community (most importantly, to funding agencies). Some pre-workshop socializing started off a bit dangerously, with Anna Karlin explaining that Avi Wigderson should saber the champagne since last time she ended up in the emergency room…

     

The visioning began excruciatingly early (certainly before I could see clearly), but it started off with some good news from Sampath Kannan, the new director of the Computing and Communications Foundations (CCF) division at NSF:  We're moving up in the world (or at least in the new NSF bureaucracy tree).  CCF will be restructured into three top-level clusters:
  • Algorithmic Foundations
  • Communication and Information Foundations
  • Hardware and Software Foundations
STOC/FOCS/SODA/CCC-esque theory will fall into the first cluster.  Besides the hopefully inevitable consequences of getting us closer to the root, there were some more subtle ones, e.g. computational geometry and quantum computation will no longer be funded separately from the rest of theory (Sampath was careful to distinguish quantum computation from e.g. quantum information and quantum engineering which don't fall into this cluster).

Then we broke into groups to "brainstorm" the nuggets; the groups were arranged into categories based on nugget sketches submitted ahead of time:  computational complexity, data-centric computing, economics and game theory, natural science, parallel computing/networks/architecture, and security/privacy/reliability.  By lunch time, various nuggets emerged, with potential titles like "Debunking the privacy vs. utility myth"  (followed by an argument about whether this constitutes a double negative and should be replaced by "Bunking the privacy vs. utility reality"?).  Watch the wiki for polished nuggets appearing in the near (hopefully) future.

The workshop was not without controversy, with Leonid Levin and Avi diametrically opposed on the number of nuggets we should be creating.  Leo thought we should have 0 nuggets, since the future of science cannot be mandated by committee.  Avi, on the other hand, treated the nuggets much like crack (the more the better).  At one point, a group wondered "Should we merge these two nuggets into one?" with Avi replying (paraphrased) "But they're so fundamentally important, why not split them into three?"  In the end, we seemed to find a happy medium (especially once Levin realized that our goals were less as "Gestapo" and more as "PR firm").  In the mean time, the view out the window of the UW CSE department provided a calming distraction.



After the workshop, a large contingent of the participants boarded a seaplane for the trip to STOC 2008.  First Rocco Servedio loaded Karp's luggage.  Then he flew us to Victoria.  See you in Canada.

         

Thanks to the organizers:  Bernard Chazelle, Anna Karlin, Richard Ladner, Dick Lipton, and Salil Vadhan for all their hard work in designing a productive and non-too-painful day of workshopping.  Credits to Claire Mathieu for some of the pictures.

11 comments:

  1. This is very interesting. thanks for posting the update, James.

    Regarding the merge of CG with the rest of theory, was there any explanation for what happens to the (currently-with-CG) other kinds of geometry (symbolic methods, geometric modelling, etc) ? Do they get refactored into a different group under CCG ?

    Also, will the algorithmic foundations cluster still have different program managers, or will there be just one ? (because currently there's Richard Beigel and Bob Grafton for the theory+CG side of things) - although Bob handles other geometry as well.

    ReplyDelete
  2. One more comment. You said that theory is "moving up". But isn't this more of a lateral shift (from TF to the new AF cluster), while reclustering some of the areas ?

    ReplyDelete
  3. psycho-killer

    ReplyDelete
  4. STOC/FOCS/SODA/CCC-esque theory will fall into the first cluster. Besides the hopefully inevitable consequences of getting us closer to the root, there were some more subtle ones, e.g. computational geometry and quantum computation will no longer be funded separately from the rest of theory

    I'm sure it was just an oversight, but it's a telling oversight: since you say that computational geometry will no longer be separate, surely you should also have said "STOC/FOCS/SODA/CCC/SoCG-esque theory"?

    ReplyDelete
  5. So Lance, any personal favorites that didn't make the final cut?

    ReplyDelete
  6. Good job with the post, James!

    ReplyDelete
  7. I'm sure it was just an oversight, but it's a telling oversight: since you say that computational geometry will no longer be separate, surely you should also have said "STOC/FOCS/SODA/CCC/SoCG-esque theory"?

    On the contrary, I consider computational geometry to be STOC/FOCS/SODA-esque theory. As far as I remember, the call for papers of all three conferences explicitly mentions computational geometry.

    ReplyDelete
  8. One more comment. You said that theory is "moving up". But isn't this more of a lateral shift (from TF to the new AF cluster), while reclustering some of the areas ?

    Certainly it's not as simple as "up or down the tree." The general feeling seems to be that ToC has moved to being a child of CCF instead of a child of TF, but of course the actual reclustering is more nuanced.

    Probably the most significant change is that Jeannette and Sampath are very theory-friendly, and this empowers our community to make good things happen.

    ReplyDelete
  9. So you feel that anything that would be on-topic for SoCG would also be on-topic for FOCS and STOC? Would that all program committee members and all funding panelists felt the same way.

    ReplyDelete
  10. But in all fairness, I don't think that *should* be the case. One of the strengths of CG is its link to many application areas. But papers that look into geometric algorithms in specific areas are not necessarily appropriate for FOCS/STOC, are they ?

    ReplyDelete
  11. Draft nuggets are here. Very interesting!

    ReplyDelete