UK airport iris biometric system flunks its exams
Buried evaluation reveals awful truth
An evaluation of the Home Office scheme to operate border controls via iris recognition "pretty much fails" Project Iris, according to Tory MP Ben Wallace. Wallace has been doggedly pursuing the results of the evaluation since autumn 2005, and these were quietly placed in the House of Commons library in late December. They reveal, according to Wallace, that Project Iris "failed half its assessments."
The Register notes with some satisfaction that in November, shortly before the Government attempted to bury the data showing all was not well with Project Iris, that we doubted Project Iris was getting anywhere. Speaking to the Today programme yesterday, Wallace said "it wasn't available when it was needed at the right level; when the system crashed, it took over eight hours to fix."
Ministers have been rather less helpful on the matter over the past 18 months, as can be seen from their answers to Wallace's questions, here. By amusing coincidence the two involved, Tony McNulty and Joan Ryan, also happen to be the two currently being urged to resign over an entirely different tracking problem.
Why, though, has Project Iris been failing? From Wallace's claims it would seem that availability issues and overall system reliability have been major bugbears. Iris is theoretically the most accurate of the three (facial, iris and fingerprint) that the Home Office has been (increasingly theoretically) planning to use, and in principle it ought not to be beyond the wit of the system implementors to achieve a reasonably high enrollment acceptance rate (i.e., getting the initial record of the subject's iris onto the system) together with favourable false positive and false negative readings when the system's actually in use.
Badly set-up systems will of course produce dud results (oddly, earlier Home Office tests 'discovered' that local lighting conditions were important, a fascinating fact that practically everybody already knew), but the challenges here are widely known, and really should only present major obstacles to system implementors who haven't been paying attention. Setting up databases that work and keeping systems live and available also ought to be reasonably achievable for people who know what they're doing, but history suggests that UK Government skills in this area are notable largely for their absence.
Note that in this failure to answer the question from September 2005, McNulty prattles on about the recognition rates specified in the contract (with Sagem), and says that iris recognition "has been chosen as the biometric for the IRIS scheme because it outperforms all other biometrics in terms of security, speed and accuracy." This is fairly characteristic of the Home Office's attitude to biometrics - ministers are mesmerised by the theoretical recognition rates for the technology, and pay little or no attention to the importance of the underlying system. Faced subsequently with the need to build such a system (say, the National Identity Register), they might well funk the challenge and bodge something else together instead.
In Wallace's view the dud results from Project Iris raise further questions about the status of iris recognition within the UK ID scheme, indicating that the ID scheme is "running off the tracks", and could prove to be unworkable. And there have been suggestions that iris has already been dropped from the scheme.
One would however do well to bear two things in mind here. First, although the Home Office has expressed an intention to gather iris data as part of the ID scheme, iris was quite obviously never viable for general purpose biometric ID applications (and here we are saying so in 2004). Inevitably, it was always going to be the case that the Government might at some point get around to collect iris biometrics from all passport applicants, but it would probably not do this for a very long time, and actual use of iris recognition would be very limited, if it happened at all. And the current position is...?
That it might start collecting iris biometrics at some point in the future...
The other thing that one should bear in mind is that Project Iris is not the ID scheme. Conceptually it certainly has things in common with the Scheme, and if the Government's imaginary Big Database of Everybody ever exists (which it won't), Project Iris would operate in conjunction with it. But Project Iris was essentially conceived as a secure flight system, and could quite easily remain free-standing (as is the case Schiphol airport's iris recognition system). Schiphol's system works, and it's really quite difficult to see how even the UK Government could continue to fail to produce a working secure flight system using iris recognition. It could do so by trying to apply it to everybody rather than as a fast-tracking system for frequent travellers, (rendering the system unwieldy and inherently less secure), or by piggybacking it onto a population-wide ID scheme that is obviously doomed. So yes, come to think of it, it's really rather easy to see how the UK Government could continue to screw it up... ®