Does the person in the next seat intend to blow up the plane? The Transportation Security Administration (TSA) believes it can answer this question via a proposed second-generation system known as Computer-Assisted Passenger Pre-Screening, or CAPPS II. In terms of the variety of information, the system goes much further than the 1.0 version--and may be too complex to pull off without huge additional funding.

The original system, which began in 1998, requires U.S.-based airlines to pass reservation data through a secret, government- supplied algorithm intended to identify fliers who pose a risk to safety. It was implemented at a time when airline safety focused on bombs in checked suitcases. (It was also meant to be temporary, to be replaced by a system that matches passengers to their checked luggage, the standard outside the U.S.) After September 11, 2001, officials extended CAPPS to include all passengers and required airlines to refuse to board anyone with a matching or similar name to those on the government¿s "no-fly" list without permission from law-enforcement officials.

The proposed CAPPS II will be an attempt to build a "threat-assessment tool" that would be the world¿s first fully automated system to check passenger backgrounds. The most recent proposals would compare name, date of birth, home address and home telephone number with private-sector databases, potentially including credit and criminal records.

But as Edward Hasbrouck, author of The Practical Nomad and an expert on travel industry infrastructure, points out, this information is not typically listed in passenger name records, which are the data that the Transportation Security Administration planned to use. To work, CAPPS II would require "the most profound change that has ever been proposed in the basic concepts of how passenger information is exchanged," Hasbrouck says. Right now airlines outsource their computerized work to external reservations systems such as Sabre. Passenger data are collected by tens of thousands of travel agencies; the agencies in turn use a variety of third-party software to run their businesses and interface with the reservations systems. As a result, data formats are not standardized across the industry, which has protocols that predate the Internet. Moreover, passenger name records and passengers do not necessarily match up one to one: a group traveling together may have one record with only travel agency information in it.

Altering current practices to suit CAPPS II will be costly. Hasbrouk thinks that $1 billion is a "conservative lower-end estimate" and that the TSA has grossly underestimated the complexity of the necessary changes. (The agency has requested $35 million for 2004 for developing CAPPS II, part of $1.7 billion overall for passenger screening.)

Still, such a system could possibly succeed: "Technically, there is almost nothing that can¿t be done given enough time and resources," comments retired FBI profiler Bill Tafoya. But with limited understanding of other cultures and the fact that data mining is only as successful as the mind-set that produces the search criteria allows it to be, he favors a risk-based assessment system. An example is the one proposed by the Reason Public Policy Institute, a Los Angeles-based think tank. Its system would identify high-, average- and low-risk passengers and focus security attention accordingly. That approach isn¿t perfect, either: Terry Gudaitis, a former terrorist profiler for the CIA who now works for Psynapse Technologies, a security firm in Washington, D.C., notes that someone with a clean record and registration as a trusted traveler would be a target for identity theft. And terrorists would have a substantial incentive to try to get themselves accepted as low-risk.

The fundamental problem, Gudaitis observes, is the "developmental nature of human beings." For example, the same terrorists who carried out the 9/11 attacks flew to their starting points. "During that flight they were not a threat," she notes. "So what was the pattern of profile, the behavioral change that occurred in an hour¿s time span? They disembarked and got on another plane."

Comments