IP: F.A.A. Gives Computers Bigger Role in Airport Security

[Dave Farber <farber () cis upenn edu>]

> Date: Sun, 13 Jun 1999 09:50:31 +0200 (CEST)
> From: Anonymous <nobody () replay com>
> Comments: This message did not originate from the Sender address above.
>         It was remailed automatically by anonymizing remailer software.
>         Please report problems or inappropriate use to the
>         remailer administrator at <abuse () replay com>.
> Subject: F.A.A. Gives Computers Bigger Role in Airport Security
> To: cypherpunks () toad com
> Sender: owner-cypherpunks () toad com
>
> F.A.A. Gives Computers Bigger Role in Airport Security
>
> By MATTHEW L. WALD
>
>
>
>
> WASHINGTON -- It may not be obvious to passengers, but the Federal Aviation Administration is trying to close three 
> weak spots in airline security: relying on human judgment to spot suspicious passengers, allowing checked bags to 
> catch a connecting flight when the suspicious passenger doesn't, and permitting baggage screeners to grow complacent. 
>
>
> But the remedies may still be at least two years away. On April 19, the F.A.A. published a proposed new security rule, 
> which is open to public comment for 60 days, and the airlines' response was to ask to extend the comment period 
> another 60 days. It would require that for all flights on planes with more than 60 seats, airlines use computers with 
> profiling software to identify suspicious passengers instead of relying on airline personnel. The software was 
> developed with a Federal subsidy, and most major airlines have been using it voluntarily since January 1998; the 
> F.A.A. will not reveal most of the criteria used by the software, but one of them is whether the passenger paid cash 
> for a ticket at the last minute. 
>
>
> When this computer profiling began, civil-liberties groups and Arab American groups complained that it would be based 
> on ethnicity or religion and thus violate some passengers' rights. But the Justice Department, privy to the criteria, 
> ruled that the criteria were not discriminatory. Now the F.A.A. wants to make the automated profiling mandatory, it 
> says, partly because it eliminates "the potential perception of personal biases." Also, the F.A.A. says, the computer 
> can juggle more factors than a ticket agent can, and if the criteria are computerized, they will be easier to keep 
> secret. 
>
>
> The airlines have generally accepted profiling, which is integrated into their reservations computers. On domestic 
> flights, the bags of passengers who meet the profile are either scanned for explosives or are "matched," meaning that 
> they will not be carried on a flight unless the passenger boards as well. (On international flights, no bags are 
> supposed to be carried unless the owner is on board.) 
>
>
> There are loopholes, however. If the bag is checked through to a final domestic destination and the passenger is 
> connecting on a different airline, the second airline may not continue with this "bag match" procedure. Even if the 
> passenger is connecting on the same airline, the match may not occur on the second flight. The fear is that a 
> terrorist might fly to a hub with a checked bag, "miss" a connecting flight, then catch another and be out of the 
> country by the time the bag explodes. The new rule seeks to eliminate both possibilities, but the airlines are fearful 
> that matching bags on the connecting flights would cause delays. 
>
>
> One way to reduce the need for bag matching is to increase bag screening, and this is gradually happening, with the 
> installation of additional sophisticated scanning machines that look for bombs in checked bags. But the airlines have 
> sometimes been slow to start using these machines. 
>
>
> Congress provided $157 million for such machines for the last fiscal year and the current one, and the F.A.A. is 
> asking for another $100 million for the fiscal year that begins on Oct. 1. The initial impetus for the program, the 
> belief that the explosion on TWA Flight 800 in July 1996 was a terrorist act, has been disproved, but the security 
> program continues. 
>
>
> A far bigger cost than equipment is the cost to the airlines from flight delays that bag screening or bag matching may 
> cause. The F.A.A.'s proposed rule includes a lengthy cost-benefit analysis of why these expenses are still worthwhile, 
> based on a hypothetical example of preventing the destruction of one Boeing 737 with 73 passengers, each person with a 
> value of $2.7 million, plus $16.5 million for the plane itself, $12.5 million in property damage, $28.6 million for 
> the investigation and $3.6 million for legal fees, for a total of $271.8 million. Also included in the analysis is the 
> amount of business that the airlines would lose in subsequent weeks as fearful travelers stayed home. (The whole 
> calculation, in the Federal Register notice of April 19, is available at www.access.gpo.gov/sudocs/ 
> aces/aces140.html.) 
>
>
> A simpler change already being made is a new method for testing the alertness and training of the security screeners 
> who look at the monitors of X-ray machines or other kinds of scanners for hours at a time, watching thousands of bags 
> go by. 
>
>
> The traditional test method is for F.A.A. undercover agents to put a weapon or a bomb into a carry-on or checked bag 
> and see if the screeners notice. But this has a high manpower cost; the carry-on baggage screeners quickly learn to 
> recognize the testers, so the agents can run only a few tests at each airport. 
>
>
> But the newest computers on airport concourses use "threat image projection," a technology that inserts the image of a 
> weapon into a bag. "Every 50 to 100 bags, there could be a gun, a knife or an improvised explosive device," said Tony 
> Fainberg, director of policy and planning for civil aviation security at the F.A.A. The computers are attached to 
> X-ray machines that scan carry-on bags. The more frequent tests would measure not only the alertness of screeners but 
> also the effectiveness of their training, as well as measuring how long a shift can be before screeners start to lose 
> effectiveness, and how long the interval between training sessions can be.