2005 CATCS Report

The SIGACT committee has been holding biweekly conference calls. The following are the main problems it sees with TCS funding.

  1. Low grant sizes in TCS, and too few of them. Grant sizes are now $70K/year.  A more viable grant size (paying for say summer salary+one student + computer/travel) will truly raise the effectiveness of researchers, allow grad training to continue, and lower various overheads for both researchers and NSF/CISE.  Researchers would write fewer proposals.
    In recent years there has also been a severe problem with very low numbers of funded proposals. In 2005 this was  ameliorated a bit since CISE made a special effort to raise the funding rate in TCS this year (both by increasing the TCS program budget and by reducing grant sizes). But the underlying fact is that the total budget of the TCS program is essentially unchanged since 1989 (which means it has greatly decreased in real dollars).
  2. TCS’s position in the CISE hierarchy is too low which causes CISE leadership to miss its importance. CISE’s view of TCS (a sibling of numerical and symbolic computation, information theory, geometric computation, etc. in the TF cluster) seems out of accord with the view in most research  departments (viz., TCS as a major subdiscipline of CS on a par with AI, systems, software systems, etc.). Note that both AI and Networking and Systems are two levels higher than TCS in the CISE hierarchy.One recent statistic to support this: this spring six of the top 10 CS depts —Stanford, CMU, Cornell, UW, U. Wisc (Madison), UIUC– made offers to six different junior TCS people (five of which were accepted). This may greatly exceed the tally for any other leaf of the CISE tree, or the combined tally for the rest of the Theoretical Foundations Cluster.
  3. Apart from the dedicated TCS program, few NSF programs support long-term, basic research. There is a pressing need for new NSF initiatives that support long-term, basic research  and which welcome TCS proposals. The TCS community also needs to be proactive too. Whenever NSF proposes new crosscutting initiatives –e.g., the new networking initiative– the TCS community needs to help delineate ways in which it can contribute.

Visit by Karp and Lipton to NSF (Sept 2005)

In early September, Richard Karp (committee chair) and Richard Lipton  visited Washington DC and met with top CISE officials, CRA staff members, and  two congressional staffers. The general tenor of the meetings was supportive and constructive. Karp gave a 1-hour talk at NSF about “Key Challenges in  Theoretical CS” that surveyed the vast spectrum of past TCS innovations and outlined some key challenges for the future. The talk was very well-attended and well-received. Slides appear here. 

A few concrete action items were formulated as a result of the meetings.

  1. The TCS community will give a writeup to CCF director Mike Foster suggesting the proper budget for TCS together with a detailed justification (including recent breakthroughs, future challenges etc.).  Foster will use it in his budget planning in future years. (But no concrete promises were made.) The details of this document may be discussed at evening meetings in the upcoming FOCS and STOC conferences.
  2. The TCS community will propose and help develop a new NSF program devoted to “Theory of Networked Computation”  (the name is provisional) . This could be loosely linked to the the new GENI initiative in networking research  at  CISE. The exact structure and research goals for this proposed ToNC program would  be worked out over a couple of NSF-funded workshops in Spring 2006, and a detailed proposal will be ready in time for the budget process in Fall 2006. The SIGACT committee will arrange the workshops and write a SGER proposal to get funding for them. It is anticipated that “Theory of Networked Computation” will be defined fairly broadly.
  3. The SIGACT committee will actively work with CISE  officials to  get  TCS  researchers into  advisory positions at CISE. It will also help recruit TCS candidates for  positions that open up at NSF. Currently, Mike Foster is looking for two program officers in the Theoretical Foundations cluster, and encourages TCS researchers to apply.It is very important to get more TCS researchers appointed as program officers. As part of its effort to promote “riskier” long term research, and to make up for DARPA’s departure from funding academic computer science, NSF  plans to  dedicate as much of a third of its funds to research projects chosen by program officers. Thus program officers will play an important role in setting research priorities (as they used to at DARPA).
  4. CISE AD Peter Freeman would like to encourage the TCS community to publish a pamphlet through CRA summarizing what TCS is,  what its fundamental problems are, and how it contributes to the nation’s science and technology. (NSF is willing to underwrite the cost.) Other scientific disciplines do such things to maintain a higher profile and thus work themselves into the nation’s funding priorities. A few other subareas of CS have done this successfully as well (see, e.g. this pamphlet).
%d bloggers like this: