As mentioned in this space on Wednesday, the House Science and Technology Committee held a hearing Thursday morning to review the federal Networking and Information Technology Research and Development program (NITRD — alternately pronounced “NIGHT-erd” or “NIGHTER-dee”), the 13 agency, $3.3 billion budget activity that represents the federal government’s investment in IT research and development. The hearing mainly focused on the recommendations issued last year by the President’s Council of Advisors for Science and Technology (PCAST) in their review of the federal IT R&D ecosystem, Leadership Under Challenge: Information Technology R&D in a Competitive World (pdf) (which we’ve also covered here). The hearing represents the first step in a process that will result in legislation next year that will attempt to once again amend the High Performance Computing Act of 1991 (most recently amended as part of the America COMPETES Act, passed in Aug 2007) to codify some of those recommendations.
(You can watch an archived webcast of the hearing and see copies of each witnesses’ written testimony at the House S&T Committee website.)
Testifying before the members were Chris Greer, Director of the NITRD National Coordination Office; Dan Reed, CRA Board Chair; Craig Stewart, Associate Dean of Research Technologies at Indiana University and representing the Coalition for Academic Scientific Computation (CASC); and Don C. Winter, VP of Engineering and IT at Boeing’s Phantom Works. Greer was there to talk about what the NITRD NCO is doing and intends to do about acting on the recommendations of the PCAST report; Reed was there as both someone who was deeply involved in writing the PCAST recommendations and who also has a strong connection to the computing research community; Stewart was there to speak for the academic HPC users and researchers; and Winter was there to bring a corporate/private sector perspective to the panel. All filled their assigned roles well.
Chairman Bart Gordon (D-TN) opened the hearing by noting his favorable impression of the NITRD program. From his opening remarks:
I believe the NITRD program has been largely a success. It has made a substantial contribution to moving computation to an equal place along side theory and experiment for conducting research in science and engineering.
In addition, it has developed the computing and networking infrastructure needed to support leading edge research and to drive the technology forward for a range of commercial applications that benefit society broadly.
The technical advances that led to todays computing devices and networks, and the software that drive them, evolved from past research sponsored by industry and government, often in partnership, and conducted by industry, universities, and federal labs.
Greer used his opening remarks to detail the efforts NITRD NCO has already undertaken in response to the PCAST recommendations (though he indicated that they would probably have embarked on the process even without a recommendation), including a strategic planning process that will produce a plan for NITRD for release in 2009. Greer also didn’t take issue with any of the PCAST recommendations — in fact, no witness (or Member of Congress) took issue with the recommendations in general — and largely agreed that the program needs to improve it’s interagency planning.
Reed emphasized a few concerns about the overall IT R&D ecosystem in his remarks, noting in particular his concern that the federal portfolio for IT R&D has lost a key piece of what made it such a success with the withdrawal of DARPA support for much university computer science research. Historically, the diversity of funding approaches and mission needs at both DARPA and NSF drove some truly innovative research in computing. With DARPA’s absence, university computing research has become a “monoculture” of research supported by a single agency: NSF. Indeed, NSF now supports 86 percent of federal obligations for computer science research in U.S. universities. As a result, Reed argued, the process has gotten more conservative — more incremental and evolutionary rather than revolutionary research proposals. This lack of diversity in approaches and mission-needs threaten to constrain the robust pace of innovation in the space, he noted. (Dan posts some additional thoughts on his testimony on his blog today.)
Stewart opened by endorsing fully the recommendations of the PCAST on behalf of CASC, but focused some of his remarks specifically on the workforce issues faced by the field. The declining interest of U.S. students in S&E — and particularly IT fields — represents a huge challenge for America’s future competitiveness, he argued. Programs that could increase the participation of American students in the STEM fields (Science, Technology, Engineering and Mathematics) should be a strong focus of the committee, and he commended the Members for their work in getting such strong emphasis on STEM education in the America COMPETES Act.
Winter really focused his remarks on the importance of the PCAST recommendation to emphasize cyber/physical systems (CPS) as a research area in the IT research and development portfolio. CPS are very important to the aviation industry, he argued, and the industry badly needs advances in technology development and tool development in the space and are reliant on the research community to get that work done.
The member questions tended to focus on how best to get NITRD agencies to collaborate on research agendas and how to set priorities given limited funding. Of particular interest to Chairman Gordon was how the NITRD program could embrace the PCAST recommendation that the program ought to be rebalanced to emphasize more high-risk, long-range research efforts. Would this require new money, he asked? Greer thought that through better coordination, the agencies could do a lot to re-prioritize existing funding, but that new money was also likely required. Reed noted that it’s not just an agency problem, there’s also a cultural component within the computing research community that needs to change, too. Researchers need to think more audaciously in their research proposals and reviewers need to be willing to reward those proposals that are high-risk, but potentially high-payoff. More funding would ease some of the pressure to award conservative proposals rather than risky ones, of course, but this still requires a mindset change within the computing community — something Reed said the community is starting to focus on.
Rep. Jerry McNerny (D-CA) raised a question related to Reed’s testimony about the undesirability of a research monoculture in the long-term part of the IT R&D portfolio. Wouldn’t a single agency, assuming it’s well run, manage and coordinate the long-range research better than if that research were spread among different agencies, he asked? Reed explained that, while its true that a single agency could certainly take on that piece of the portfolio by itself, historically, having a diversity of different funding models and agency missions available to researchers has proven to be an incredibly productive way to enable innovation in the IT sector. NSF is very good at individual investigator initiated research, for example, and DARPA was very good at placing big bets on hard problems and hand-picking communities of researchers to focus on them. Between just these two diverse approaches an enormous number of innovations resulted.
There was also a recurring focus on cyber security in the member questions, in part spurred by the discussion about the ubiquitousness of computing devices and the increased access we now have to them. Winter pointed out that cyber security wasn’t always a concern for a company like Boeing, despite a widespread use of embedded computing devices in things like avionics systems. But now, these systems increasingly communicate with the world outside the airplane — exchanging data with other aircraft and other assets in the battlespace, enhancing the effectiveness of the systems, but also increasing their vulnerability to cyber attacks. There is much research to be done, the panelists agreed, on understanding how to secure these cyber-physical systems, and there were great concerns expressed that the current and projected workforce in the area is inadequate to the task ahead. Support for research in the area helps produce that workforce, the panelists noted.
Finally, there was also brief discussion about Reed’s recommendation, as someone who has served on both PCAST and the President’s Information Technology Advisory Committee (PITAC) before it was folded into PCAST, in support of reconstituting PITAC in order to really get adequate oversight of the NITRD program. Though there are some within the Administration who oppose the push to reconstitute PITAC, there was no objection from the committee members to the suggestion — in fact, Chairman Gordon pointed out that their reauthorization of HPCC in the America COMPETES Act actually called for the same thing. So perhaps we can look forward to the return of PITAC in the next Administration.
And that was about it. Despite a good turnout among Members of Congress for the hearing (I counted 11 present at various times), the committee managed to wrap up its review of the program in just 56 minutes — a record, in my experience, for a full-committee hearing of the House S&T committee. I take that as a good sign, however. The issues confronting the program are pretty clear, the steps required to address them aren’t terribly controversial, it just remains to do them. In the next few weeks/months, we hope to see the direction the committee plans to take regarding the PCAST recommendations.
As always, we’ll have all the details here….