If Biometrics Aren't Foolproof, What Is Point of Using It?

The government's ID card team this week made an important admission. They admitted that, much of the time, the new systems won't work. It de-fangs a common argument against the scheme - that it was conceived in an IT consultant's fantasy world, where citizens are obedient and technology always functions. (The opponents' next logical leap, from "It'll never work" to "It'll create a totalitarian state" is beyond the intellectual scope of this column.)

Katherine Courtney, the programme's director, and Ian Watmore, the government's chief information officer, were updating journalists about how the card will be introduced, parliament willing. They presented the process as a near-seamless transmission from current efforts to improve passport security and to tidy up government systems. There will be no big bang, Watmore stressed.

That may be so, but many minor explosions will be necessary to keep things on track. The plan is to begin enrolling on to a new identity register in 2008. The passport agency, shortly to be renamed, will deal with "a few hundred thousand" early adopters in that year. The most likely enthusiasts are frequent travellers, as the card will function as an EU travel document.

The register will hold our names, dates of birth, a new national identity number, immigration status and address. The last is likely to cause some grief - unless they are expecting a cheque in the post, Britons aren't conscientious about telling government where they live. Even when they do, there is no guarantee that an address will match one held on one of the state's three competing address databases.

Finally and most controversially, the register will hold biometrics. The plan is to collect three - a face image, both irises and all 10 fingerprints. The idea of taking three different sets is that for any given individual, at least one of them is pretty certain to work.

However, Courtney and Watmore acknowledged that biometric matching won't be infallible. They claim the technology is proven in one-to-one matches, to show that an individual is the one whose characteristics are on file. The fun and games begin with one-to-many matching, needed to prevent people registering two or more identities. No one has proven the technology on anything like the scale required by the card. "We'll be pushing the boundaries," said Watmore, a term he often uses about government IT. "We need to do further testing on one-to-many," said Courtney.

The pair also dismissed claims that to function, the scheme will need a massive network of biometric readers. In practice, biometric matching will be the exception, used for example when applying for a job. In day-to-day transactions, such as renting a car, we will identify ourselves with a pin - the card will be compatible with retail chip and pin readers.

Even with these concessions to the real world, the programme will be a big undertaking by a sector whose record is, well, mixed. Watmore hopes to avoid disaster by drawing on best practice accumulated from a decade of painful experience. The first stage, negotiations with the IT industry about what it is capable of doing, has already begun.

Why not further minimise the risks to the programme by doing away with intrusive biometrics and concentrate on creating a single national register of names, addresses and numbers? Courtney said that although biometrics is "no golden bullet", it will reduce fraud. Criminals will have to work hard to get two identities and work very hard to get three.

Opponents will say that conceding the fallibility of biometrics in day-to-day life removes the card's big selling point, so the whole programme should be scrapped. The question to throw back at them is, if the card could be shown to work perfectly, would they accept it?

(c) 2005 The Guardian.