Debate over usefulness, pitfalls of biometrics intensifies

How accurate — and effective — is biometric technology? A recent issue of

Nature (sub. req.) offers the views of two biometrics skeptics. David Moss of the London-based Business Consultancy Services writes (sub. req.) that biometrics supporter

Anil K. Jain (see below) gives a misleading picture of the effectiveness of biometric technologies. Jain writes that, since the inception of the US Visitor and Immigrant Status Indicator Technology (US-VISIT) in January 2004, more than 75 million visitors have been processed through it and about 1,000 have been denied entry. According to the U.S. Department of Justice’s analysis of US-VISIT after its first year of operation, on average 118,000 people pass through each day. Of these, 22,350 are subject to secondary inspection and 1,811 of these are refused entry — “considerably more than Jain states,” writes Moss. Of those who are subject to secondary inspection, 92 percent are subsequently admitted to the United States. “These figures raise doubts about the reliability of the biometric systems used for primary inspections,” says Moss.

Applications such as US-VISIT need to reject virtually all impostors, and keeping the false accept rate (FAR) close to zero pushes the false reject rate (FRR) up. Jain writes that current fingerprint-recognition systems can provide a FRR of up to 0.01 percent at a FAR of 0.1 percent. “However, in a 2005 U.K. Passport Service trial, the FRR for fingerprints was 19 percent,” writes Moss, and concludes: “One objective of the proposed U.K. National Identity Scheme is to make it easier for people to prove their identity. At current performance, biometrics based on finger-prints could instead make it harder for nearly a fifth of the population to prove that they have the right to work in the United Kingdom and enjoy social entitlements.”

Andrew Watson writes that Jain’s arguments for bometric adoption have other flaws. “We all leave a biometric trail in our daily lives,” writes Watson. “Our fingerprints on a drinking glass, our voice on a telephone answering machine, our iris patterns on a photograph. We have little ability to change such characteristics and little control over this trail, which makes biometrics useful to forensic science.” This trail also creates problems, becasue it is exactly these properties that make biometrics a poor replacement for passwords and ID cards, since it is easy for an intruder to collect someone’s fingerprint or iris scan without their knowledge, and then inject it into a biometric identification system. “Even if the victim becomes aware of the problem, it’s impossible to revoke the biometric,” says Watson. If your credit card is stolen, the card company can send you a new one with a different number, but one cannot get a new set of fingerprints. “It is precisely because biometric information is irrevocable and unwittingly provided in our daily lives that it is so useful to organizations that regulate the individual (for example, the US Immigration Service), but of little use where the individual controls identification and authorization,” Watson concludes.

-read more in Anil K. Jain, “Technology: Biometric Recognition,” Nature 449 (6 September 2007) | (doi:10.1038/449038a) 38-40 (sub. req.)