COMPUTING RESEARCH POLICY BLOG

Since about 2001, the computing community – through CRA and others, and with lots of mention on this blog – has aired concerns about policy changes at the Defense Advanced Research Programs Agency (DARPA), the Defense Department’s leading-edge research arm and arguably one of the two most important agencies in the history of computer science. In particular, we’ve been concerned with a set of policies that discouraged the participation of university-based researchers in DARPA-sponsored research – policies like the use of “go/no-go” decisions without regard to the realities of fundamental research, the use of prepublication review on basic and applied research, and an increased use of classification of research that precludes participation from most researchers in the university community.

With the change in Administration and a new DARPA Director (Dr. Regina Dugan) appointed, we have been hopeful that these problematic policies would be reviewed and reversed. We were considerably encouraged when Dugan selected CRA’s former Chair, Dr. Peter Lee, the Chair of Computer Science at Carnegie Mellon University, to head a new office at the agency chartered, in part, to reengage the agency with the university community. Both Dugan and Lee have been making the rounds to university campuses over the last year listening to the concerns and pledging to address them.

Last week, Dugan testified before the House Armed Services Committee and addressed this need to change explicitly. Here’s some of what she said:

Over the last few years, the University community has articulated concerns about DARPA’s commitment to basic research. There was much said on both sides about the veracity of these concerns. As I described previously, one of the elements of DARPA’s success is the Agency’s commitment to work at the intersection of basic science and application, so-called Pasteur’s quadrant. The tension created in Pasteur’s quadrant arguably serves as a catalyst for innovation. DARPA is not a pure science organization, but neither are we a pure application organization. We sit firmly at the intersection of the two and, to be successful, we need the minds of the basic scientist and the application engineer, those in universities, and those in industry. And we need them working together, often on a single project, in the cauldron created by the urgency and technical demands of Defense. This is almost a unique characteristic of DARPA projects, which are often multi-discipline, multi-community, and multi-stage.

University Outreach.

Upon arrival at DARPA, we were determined to understand and repair the breach with universities. We discovered the following: Between 2001 and 2008, DARPA funding to US research university performers did decrease in real terms, by about half. But, as importantly, a noble and recent focus in the Agency on solving nearer term problems for the Department had resulted in some additional, perhaps unintended, consequences. The nature of the work changed, from multi-year commitments, to those with annual “go, no-go” decisions governing continued funding, which made it difficult for universities to commit to graduate students. A later stage focus resulted in more work done by universities as subs to prime contractors responsible for integration efforts, and the resulting flow-down of restrictions on the use of foreign nationals, export control, prepublication review, among others.

We assessed that we could address many of the concerns identified. So last September I traveled to five universities – Texas A&M, Caltech, UCLA, Stanford and Berkeley – to meet faculty, deans, and presidents, graduate students and undergraduates. The goal was to speak honestly and directly with them. We laid out the concerns, as we understood them, and the changes we had made or intended to make. We asked for their feedback. And we asked for their renewed commitment as well. For researchers to renew their commitment to working on Defense problems. For university leaders to clear obstacles and encourage their best and brightest to serve in Government. This service is, of course, in our shared self-interest because the quality of Government research sponsorship goes directly as the quality of the program leadership.

We continue to work on the issues: by educating our program managers to include basic research as an element in their programs, where appropriate, and to protect the integrity of this work under the provisions afforded fundamental research. The Agency has instituted new processes to ensure the necessary elements of academic freedom in basic research are balanced with the responsibilities of national security concerns. And we have increased transparency so that researchers can quickly determine whether restrictions apply to their work.

Since September, we have visited additional campuses across the country and spoken with university representatives to include Virginia Tech, Georgia Tech, MIT, and others. Our dialogue continues with more than 100 schools. We have more work to do, on both sides, but so far, it seems as if the breach is healing.

The full testimony is online and worth reading. This change at the agency is enormously positive, not only for the computing research community – which will gain (regain?) an important funding source and a different funding model than NSF – but for DOD and the country as well. After all, one of our biggest concerns with DARPA’s disengagement from the university community over the better part of the last decade was that it meant that some of the best minds in the country – indeed, some of the best minds in the world – were no longer thinking about defense problems. DARPA’s policy changes should help reclaim some of that mindshare, and in the process, better serve our warfighters and protect our country.