Wednesday, December 27, 2006

Foundations and Impacts

As we start thinking about our Theoretical Foundations proposals, a few related items from the theory community.

Joan Feigenbaum and Michael Mitzenmacher have written a report "Towards a Theory of Networked Computation" available on the group's web page. The report is still under revision and Joan welcomes your comments.

SIGACT Chair Richard Ladner writes in the latest SIGACT News about the importance of the required Broader Impacts criteria in NSF proposals.

One way to think about the Broader Impacts criterion is that when we receive money from the people of the United States through NSF, the people would like to know ahead of time of what benefit the research may or will be to society. If there is little or no benefit then why should the people continue to support NSF? When NSF goes to Congress to ask for money, it is going to the people's representatives, who ask for justification to spend the people's money on scientific research. Basically, NSF's funding, and ours indirectly, depend on the belief by the public that broader impacts come from our research. Some people have said to me that a focus on Broader Impacts is a move away from basic research to more mission oriented research, or research with strings attached. If we look at the ways that we can satisfy the Broader Impacts criterion, they are very general, and relate to education, broadening participation by underrepresented groups, and other benefits to society. Please read the representative activities for concrete ideas for how to include Broader Impacts in our proposals.

As SIGACT Chair, I am trying to help increase the funding for computer science theory research. The best way to increase funding for research is to convince people it is important to them and the people around them. There is a difference between "important" and "useful". Artists are able to convince people to buy art, not because it is useful, but because it inspires them. Astronomers convince people to pay them to study the stars, not because they are useful (except for our own star, the sun), but because the stars are fascinating in their own right. Understanding the birth and possible death of the universe is of no practical value, but is just a fundamental question.

All this said, I am a firm believer in serendipity. Often, research leads to unexpected results and unanticipated applications. Unfortunately, this phenomenon is quite rare and probably not common enough to convince people to provide large amounts of research money. The best approach is to have a great story about the benefits of theoretical computer science research and its promise for the future. This will generate enough money for all of us so that rare serendipitous events will happen naturally in the course of doing our research.


  1. Yawn. Didn't see many talks on disributed memory algorithms. Wonder how many of these guys are back in PRAM land.

  2. Unfortunately, the "broader impacts" section of NSF proposals has become a joke, as anyone who has ever written or reviewed a proposal likely knows. Almost everyone says essentially the same meaningless blather: how we are going to make sure to educate students (especially women and minorities!) and how furthering understanding of [insert proposal topic here] is both of fundamental importance and likely to lead to useful results. Bah.

  3. Yes, it is possible to say bland and uninspiring things in the "broader impacts" section. It is also possible to do more and make that section less of a joke. Some of the things I've heard of that people have done include developing educational material for high school students, volunteering their expertise in computer science for the benefit of local organizations, and getting involved in policy issues related to their research. These are concrete things that go beyond just vaguely promising to educate more students.

  4. By the way, the NSF highlighted list-decoding as one of the breakthroughs coming from NSF-sponsored research. Are there more theory-related such highlights? Are there similar results that should be highlighted but have not yet been? That might help with crafting a general story for theory, as Lance suggests.

  5. Hmmmm ... kind of a bland & cynical thread so far!

    Just to ask a non-bland and non-cynical question, how many people would agree that in order to dramatically increase the NSF's broad impact on society, the NSF should adjust its 2008 budget to allocate support 50%-50% between fundamental scientific and mathematical research on the one hand, and practical engineering research on the other hand?

    At present, the NSF Engineering Directorate receives approximately 10% of the NSF's overall budget (summary here, details here).

    Isn't this engineering choke-point, arguably, the main obstacle that prevents the NSF's (fiscally much larger) investment in science and mathematics from achieving a broader impact on society and the economy?

    To put it another way, if an adjustment along these lines isn't made, every engineering department in the country is going to have to start teaching Mandarin as a required language, because the Mandarin-speaking community will complete its rapidly accelerating program of becoming the dominant language of system engineering in the 21st Century.