John Daugman asserts that his iris recognition system generates no errors (13 December, p 34). This claim is akin to a goalkeeper arguing that, based solely on the number of saved goals, he has a perfect record. A more accurate indication of the performance of iris recognition can be derived by taking into account the unruly "misses" that Daugman ignores.

It may well be true that those irises recognised by the system can be matched with extreme accuracy against other irises recognised by the system. The key question is how many irises the system fails to accurately detect and therefore fails to match. The project in the United Arab Emirates cited by Daugman provides no insight. Even if I take at face value the UAE Minister of Interior's assertion that none of the 3684 "positive hits" was contested, I still have no clue about how many prohibited identities slipped through the system, or how many innocent people were wrongly identified. And neither, I suspect, does Daugman.

A report issued last year by the US General Accounting Office reported that the largest iris scanning system currently in use had only 30,000 records. The GAO warned that it was "unknown" how a system with many millions of records would perform. Generally speaking, whenever the population of a biometric system is increased, the threshold (sensitivity) of the system has to be decreased. The authors of the GAO report and other studies are right to be concerned that a system involving tens of millions of identities would be obliged to compromise security and "positive match" accuracy to avoid an unacceptable level of false hits.

Iris recognition has a role to play in security and population management. But just as with any other biometric system, that role will not be found in an over-ambitious national or global identity system. Such a proposal would involve risks and burdens that are wholly unacceptable to any free society or indeed any marketplace.

Daugman holds the patent for iris recognition. I can understand his sensitivity to criticism. But if he continues to focus only on the most reliable element of his system he does a disservice to those who believe his technology can solve key economic and societal problems.