The joint investment announced yesterday by Microsoft and Intel in two university research centers (one at Berkeley and one at University of Illinois Urbana-Champaign) in order to work on solving the challenges of multi-core computing is all over the news, but there’s an aspect of the story that’s been hasn’t been highlighted sufficiently. The NY Times’ John Markoff picked up on it, however:
Both Intel and Microsoft executives said the research funds were a partial step toward filling a void left by the Pentagon’s Defense Advanced Research Projects Agency, or Darpa. The agency has increasingly focused during the Bush administration on military and other classified projects, and pure research funds for computing at universities have declined.
“The academic community has never really recovered from Darpas withdrawal,” said Daniel A. Reed, director of scalable and multicore computing at Microsoft, who will help oversee the new research labs.
[Dan Reed is also the current Chair of CRA.]
We’ve noted many, many times on this blog our concerns with policy changes at DARPA since about 2001 that have had the effect of pushing university researchers away from DARPA-sponsored research. As we wrote as recently as September 2007, shorter research horizons with an emphasis on go/no-go milestones at relatively short intervals and an increased use of classification at the agency has sharply reduced the amount of DARPA-supported research being performed in U.S. universities. In fact, between FY 2001 and FY 2004 (the last year for which we have good data), the amount of funding from DARPA to U.S. universities fell by half — and informal evidence suggests university shares are even lower today.
While it’s great news that two of the titans of the IT industry are stepping up to fill some of the gap left by DARPA’s withdrawal, their $20 million investment over 5 years represents just a tiny fraction of the DARPA shortfall. The difference in DARPA funding for university computer science between 2001 and 2004 was $91 million annually ($214 million in FY 01 to $123 million in FY 04 in unadjusted dollars), and anecdotal evidence suggests that shortfall may be even larger now. The Microsoft-Intel investment is a bold move and big commitment to address a key challenge in computer science that’s a primary concern for the two companies in the future. But it doesn’t represent a sustainable alternative to filling the hole left in the IT R&D portfolio created by DARPA’s absence.
DARPA has taken some steps to try to bring university researchers, especially younger faculty, back into the fold. In February, the agency also reorganized its IT office structure a bit — merging the Information Exploitation Office (IXO) with the Information Processing Technology Office (IPTO) to create a new Information Processing Techniques Office (IPTO) under former IPTO Deputy Chuck Morefield. There’s some indication that the office will have a technology focus (which suggests a research emphasis) in addition to a systems focus (which suggests a development-oriented emphasis), so there may be increased opportunities for university researchers to participate in DARPA-sponsored work.
We hope so, because while it’s great to see the IT industry step up and make some commitments to university-led research, the country (and the DOD, and the world) is probably better served by a DARPA that’s re-engaged with the university research community, supporting long-term, DARPA-hard research at a range of institutions on some of the grand challenges in computing….