To thwart terrorists, the Pentagon is developing a computer surveillance system that would give U.S. agents fingertip access to government and commercial records from around the world that could fill the Library of Congress more than 50 times.
The library's collection of more than 18 million books would be dwarfed by the size of the computerized files the government wants to mine for clues that terrorists are planning attacks.
The prospect of what the Pentagon calls the Total Information Awareness system has alarmed privacy advocates on both ends of the political spectrum.
In February, Congress barred use of the still-to-be-developed system against American citizens and ordered a full description of the plans developed by the Defense Advanced Research Projects Agency, or DARPA. That report was to be delivered to Capitol Hill on Tuesday.
The Pentagon has named two oversight boards to supervise the program.
Also Tuesday, the Center for Democracy and Technology, a group that advocates online privacy, was giving a House Judiciary subcommittee a report that concluded, "There are few legal constraints on government access to commercial databases."
Neither the Privacy Act nor the Constitution protect consumer data held by private companies, and other laws "are riddled with exceptions for law enforcement or intelligence uses."
The center's executive director, Jim Dempsey, said in prepared testimony, "Since 9/11, the FBI is authorized by the attorney general to go looking for information about individuals with no reason to believe they are engaged in, or planning, or connected to any wrongdoing."
In advice to would-be TIA contractors, DARPA disclosed that the project will require "gathering a much broader array of data than we do currently" and break down barriers that keep separate data already collected by a host of agencies.
"The amounts of data that will need to be stored and accessed will be unprecedented, measured in petabytes," the agency instructions said. A byte amounts to the electronic representation of one letter of the alphabet, and a petabyte is a quadrillion — 1,000,000,000,000,000 — bytes.
DARPA, which developed the Internet, is again trying to expand the boundaries of existing technology. Among the largest databases on the Internet is an archive of the last five years of Web pages; it consumes 100 terabytes, or one-tenth of a petabyte.
Conceived and managed by retired Adm. John Poindexter, the TIA surveillance system is based on his theory that "terrorists must engage in certain transactions to coordinate and conduct attacks against Americans, and these transactions form patterns that may be detectable."
The Pentagon describes the system as "a research program designed to catch terrorists before they strike."
DARPA said the goal is to predict terrorist actions by analyzing such transactions as passport applications, visas, work permits, driver's licenses, car rentals, airline ticket purchases, arrests or reports of suspicious activities. It aims to look for patterns in activities tat are associated with terrorist planning.
Other databases DARPA wants to make available to U.S. agents include financial, education, medical and housing records and biometric identification databases based on fingerprints, irises, facial shapes and gait.
Despite privacy fears, government documents reviewed by The Associated Press show that scores of major defense contractors and prominent universities applied last year for the first research contracts to design the surveillance and analysis system.
Poindexter's project would integrate some projects DARPA has been working on for several years, including an effort to develop a radar-based device that could identify people by the way they walk.