Is more terror in US inevitable?

The matrix sums up hints of terrorism plucked from a vast range of intelligence gathered by US and foreign governments: signals intelligence (such as intercepted phone, e-mail, and radio communications), satellite imagery, and reports from human sources (such as recruited agents and detained terrorist suspects). CIA counter-terrorism experts select what they consider credible threats. Next, they log each one under columns for describing the nature of the threat, the source of the threat and its credibility, and action taken to alert different government branches – from embassies and military commands to state authorities.

On any given day, the matrix could list from dozens to as many as 100 individual threats, US officials say. The threats are updated when new bits of information come in, or when new preventative actions are taken.

"We're privy to shared information that's coming from a broad coalition, in surprising range and depth. That ought to be a comfort," says Marine Lt. Gen. Gregory Newbold, director of operations for the Joint Chiefs of Staff. He described a dramatic "fusion" of intelligence on terrorist threats since Sept. 11, demonstrated by twice-daily secure teleconferences by representatives from all leading US civilian and military intelligence agencies as well as key users of the data.

In investigating terrorism, CIA analysts can use computers to manipulate vast databases.

For example, using pattern recognition, they can search for information that could tie known terrorist organizations to other, seemingly unrelated facts. They can also perform time-series analysis, clustering facts sequentially in order to track terrorist actions over time, says John Gannon, a former CIA deputy director for intelligence.

Moreover, the CIA, Pentagon, and National Security Agency are looking into new computer tools and mathematical techniques to enhance their ability to predict terrorist actions based on clues they gather, outside experts say.

One tool, named Bayesian inference after the 18th-century English clergyman Thomas Bayes, is being used to improve the best guesses of the US military about potential terrorist attacks on its installations, says Bryan Ware, CEO of Digital Sandbox, a Reston, Va., defense contracting firm.

Replacing a pencil-and-paper method of rating the vulnerability of military bases that dates back to the Vietnam War, the new technique uses math to improve human predictions by assigning reliability quotients to vast bodies of data. The software works much as does an online book retailer such as Amazon.com that uses past buying to suggest new purchases.

Researchers are also using complex math to try to decode messages hidden in digital photographs sent by e-mail - one of the communication methods the Sept. 11 terrorists are suspected of using, says Keith Devlin, a mathematician and expert on encoding information at Stanford University in Stanford, Calif.

Page: 1 | 2