U.S. SCIENCE POLICY:
Congress Splits Over Plan to Consolidate Intelligence Research
Yudhijit Bhattacharjee

U.S. intelligence agencies need new surveillance tools to fight global terrorism. But it's not clear how they should carry out the necessary research

In October 1962, U.S. reconnaissance airplanes provided evidence that the Soviet Union was building up an arsenal of warheads on Cuba, only 150 km off the Florida coast. Those pictures led to high-level talks between the two superpowers that averted what many believe could have been a nuclear war.
In hindsight, intelligence experts say that finding missile sites was a piece of cake compared to the surveillance challenges in the post-9/11 world. Unlike during the Cold War era, they say, intelligence agencies today must track not only government military installations but also terrorist networks and individuals. To meet that challenge, the Director of National Intelligence (DNI), Michael McConnell, has proposed cobbling together existing U.S. research and development (R&D) programs at 14 agencies into a new organization. Modeled on the Defense Advanced Research Projects Agency (DARPA), the proposed Intelligence Advanced Research Projects Activity (IARPA) would be built mainly from merging the Intelligence Technology Innovation Center at the Central Intelligence Agency (CIA), the Advanced Research and Development Activity (ARDA) at the National Security Agency (NSA), and the National Technology Alliance at the National Geospatial-Intelligence Agency.

McConnell says the new arrangement will stimulate long-range research on the gathering and analysis of intelligence that now falls outside the mission of a particular agency. "We are in a rut right now, turning the crank on the same technologies," says IARPA's acting director Steven Nixon. "What we need to do is swing for the fences."

But Congress is divided over the plan. The House version of the 2008 Intelligence Authorization Act passed last month contains language that forbids DNI from merging any existing research programs under IARPA. "They are creating just another agency looking for a piece of the pie that will push their own pet rocks," says Representative Heather Wilson (R-NM), who thinks that McConnell's office should instead continue to coordinate programs run by each intelligence agency. Wilson and others worry that the new organization might end up subsuming the entire science and technology operations of individual agencies, leaving them with no science portfolio of their own.

Two weeks later, however, the Senate endorsed the idea when it passed its version of the overall bill. "We think IARPA can fill in gaps between the needs of single agencies," says a Senate aide, who expects the plan to survive when the two bills go to conference this fall. "It's an invalid concern that IARPA is suddenly going to become the program manager for all the science that's done by the intelligence community."

DNI officials say the new agency, with its 40 program managers headquartered in leased space at the University of Maryland in College Park, is needed to keep pace with the rapidly evolving threat of global terrorism. IARPA would sponsor basic and applied academic research on intelligence-related topics such as machine translation of foreign languages, pattern recognition, and quantum encryption with grants to academia, national labs, and industry.

Although President George W. Bush's fiscal year 2008 budget request asks Congress for only a modest increase in 2008 over the science budgets of IARPA's constituent programs, DNI officials hope for a 5-year doubling of current budgets. The size of the budget is classified, but outside experts speculate that the intelligence agencies are now spending between $250 million and $350 million on the programs IARPA would consolidate.

Researchers seem enthused by the plan. "It seems likely to me that an integrated approach to tackling 'grand challenge'-type R&D could yield a greater return on investment," says Ruth David, an electrical engineer who once led CIA's science and technology directorate and now runs Analytic Services in Arlington, Virginia.

Current programs funded by CIA or NSA strive to develop products that can "immediately plug into existing intelligence analysis systems," says Mark Steyvers, a computer scientist at the University of California, Irvine, who has done unclassified research for the CIA on extracting meaningful text from huge data sets such as e-mail chatter on the Internet. But he says those products lack broader applicability. Steyvers predicts that IARPA would spark "exciting new collaborations" between disciplines such as computer science, statistics, and the social sciences.

IARPA plans to seek exactly those kinds of interdisciplinary proposals, says Nixon, with an eye toward decoding chat-room conversations between terrorists and monitoring weapons and people in otherwise inaccessible regions. "During the Cold War," says Nixon, "we had big monolithic targets, and so we needed technologies to answer questions like, 'Where's the airfield, and how many bombers do they have?'Now, the targets have shrunk, and they are all over the place."