Posted on Mon, Mar. 29, 2010
last updated: March 30, 2010 09:43:05 PM
WASHINGTON — Three recent events — the foiled Christmas Day bombing of a Detroit-bound airliner, the Dec. 30 assassination of seven CIA officers and contractors by a Jordanian double agent in Afghanistan and the difficulties that U.S. Marines in Marjah, Afghanistan, have encountered — all have something in common: inadequate intelligence.
To lower the odds of similar troubles in the future, the government has launched a swarm of spooky, out-of-the-box research projects known collectively as the Intelligence Advanced Research Projects Activity.
"The intelligence community needs to place bets on high-risk, high-payoff research that might not work, (but if it did) would give us an overwhelming intelligence advantage over future adversaries," IARPA director Lisa Porter said in an interview at her sparkling new headquarters just outside Washington in College Park, Md. "We need to fundamentally change the way we do business."
Porter's boss, Director of National Intelligence Dennis Blair, said that IARPA's task was to be "an intellectual ferment or primordial stew out of which great things will come." He wants Porter's researchers to "generate revolutionary capabilities that will surprise our adversaries and help us avoid being surprised."
IARPA is modeled after the Defense Advanced Research Projects Agency, which has conducted far-out research for the Defense Department since 1958. DARPA's many innovations include the Internet, GPS and robotic vehicles.
Founded two years ago, IARPA has contracted with about 75 university research laboratories and 50 technology companies, large and small, to work on innovative solutions to future intelligence needs. More contracts are coming soon, Porter said.
Some IARPA projects have a distinct science-fiction feel.
One program, Reynard, for example, has signed contracts with five research teams, mostly from major universities, to develop systems to observe "avatars" — animated computer images — that take part in popular "virtual world" games such as Second Life and World of Warcraft.
Such games have more than half a billion players around the globe, according to Reynard program manager Rita Bush. Players include many young Muslim men.
The idea is to study how these avatars — like those in the hit movie "Avatar" — behave and communicate with one another for insights into how real-life people in hostile cultures think and act.
IARPA officials think that analyzing avatars' behavior in a "virtual world" can produce useful insights into the nationalities, genders, approximate ages, occupations, education levels, even the ideologies of their creators in the "real world." Players also use avatars to communicate with one another.
"One of the goals of this program will be to understand how terror groups might use such virtual worlds to communicate," said V.S. Subrahmanian, the director of the Institute for Advanced Computer Studies at the University of Maryland, who isn't connected with IARPA.
"This is a laudable goal. However, it is also a major challenge," Subrahmanian said in an e-mail message. "To identify how terrorists communicate in a VW (virtual world) requires the ability to first identify which conversations are in fact legitimate or normal and which ones are suspicious. This is hard to do."
"If it weren't hard, we wouldn't be doing it," Porter said. "Failure is OK. We can learn from failure."
Another IARPA project, named ICARUS, will attempt to model the way human brains make sense of a bewildering mass of data. The ALADDIN project is meant to pick out key items in the tsunami of video images that spy agencies collect. A program called TRUST will try to help intelligence officers determine who can be trusted and who can't.
Although IARPA resembles DARPA, there are important differences. DARPA research is aimed at pressing military needs, with a timeline of a year or so. IARPA is designed to help the intelligence community solve long-range problems.
It probably will take five to seven years before the CIA, the FBI, the National Security Agency or other intelligence agencies benefit from IARPA's projects, Porter said.
The ALADDIN project is intended to help intelligence analysts cope with the thousands of video images that pour into their offices each day from unmanned aerial vehicles, on-the-ground surveillance and other sources in danger zones.
"We get way too much video," Porter said. "We have time to look at only a small portion of it. ... We want an automatic tool that looks at 100 percent of the videos and identifies things of interest."
An ALADDIN system could "automate lower-level tasks, such as detecting tiny changes in images that a human might miss or take a lot of time to detect," she said. "Machines are good at that."
The TRUST program differs radically from traditional lie detectors, or polygraphs, which measure people's heart rates and perspiration to see whether they're lying. Instead, a TRUST goal is to measure subconscious biological signals in one's own body.
"We generate signals in ourselves when we first meet people," Porter said. "There's been a lot of research on this."
Porter said a TRUST program might have helped save the CIA officers whom a Jordanian double agent betrayed and killed in Afghanistan last year.
Still another program, called Knowledge Discovery and Dissemination, might have helped detect Umar Farouk Abdulmutallab, the Nigerian bombing suspect who's alleged to have nearly caused a tragedy on Christmas Day in spite of a raft of clues, which weren't put together in time.
IARPA claims that KDD projects could improve massive databases that don't mesh well with one another, allowing key connections to go undetected.
In the Christmas bombing case, "the dots simply were not connected," Russell Travers, a deputy director at the National Counterterrorism Center, told the Senate Judiciary Committee last week at a hearing on the incident. "The U.S. government needs to improve its overall ability to piece together partial, fragmentary information from multiple collectors."
IARPA's ICARUS program will exploit the latest research by neuroscientists on how the human brain operates.
"Recent advances in our understanding of brain function ... have laid the groundwork for an ambitious new effort to understand human sense-making," according to IARPA's description of ICARUS.
For example, Juyang Weng, a Michigan State University expert on how robots learn from experience, attended an ICARUS information session in January and intends to submit a proposal to IARPA. He told the group that he's already working to develop machines that demonstrate "brain-like sense-making and reasoning."
"The subject of ICARUS is very challenging, but doable based on the latest breakthroughs," Weng said in an e-mail message. "The machine 'brain' must be autonomously developed so that it can accumulate experience from rich real-world experience."
Similarly, computer giant IBM's "Blue Brain" project aims eventually to use supercomputers to "replicate an entire brain," project director Henry Markham told a computer technology conference last year in Long Beach, Calif.
Computer scientists Robert Sloan and Gyorgy Turan, of the University of Illinois at Chicago, won a $500,000 grant from the National Science Foundation to develop methods to build "common-sense knowledge bases" that can evolve as they take in new information.
ON THE WEB
MORE FROM MCCLATCHY
Follow the latest politics news at McClatchy's Planet Washington
McClatchy Newspapers 2010