Saturday, 12 May 2007

ID Cards: friend or foe?

There are many compelling reasons to be concerned about the U.K. Government's ID Card scheme, most of them presented (far more cogently than I could blog about them) on the No2ID campaign website. The spiralling costs, the unproven technology, the threat to civil liberties, the lack of any convincing argument to support the assertion that the cards will somehow protect us from terrorism and identity theft... the list goes on.

One objection that isn't often raised, at least in the media that I access, is what seems to me to be a fundamental flaw in the Government's understanding of the significance of biometric identification - not just in terms of its accuracy or feasibility, but in a failure to grasp what the point of biometrics is.

I worked for a while for a Belgian company that developed biometric products. One way they promoted these was in the form of a biometrically equipped VW Beetle, which neatly demonstrated the potential of biometrics - the convenience of unlocking your car at the touch of a finger on the scanner controlling the door locks, starting the engine by voice command and, most importantly, not having to carry keys around with you any more; keys that can be stolen, lost or copied. The same idea can be (and is already, in places) applied to ATMs: build in an iris-scanner and there's no need for bank cards and pin numbers any more. The point is that biometric ID removes the need for keys and cards, which not only clutter your pockets but are easily compromised. You are identitified by your biology. The best biometric data, if it was conveniently checkable, would be your DNA profile; meanwhile, fingerprints and iris-scans are secure enough.

The Government's intention is to maintain a register of all persons who are entitled to be in the country, and to have the means to check whether an individual is recorded on that register. It might like to do more (in terms of tax collection for example), but I'm using the Government's own arguments as to the necessity for an ID card system (i.e. to counter identity theft and terrorism). It will do this by recording the biometric data of all its citizens, and storing it on a central database. This is all that is required for their stated purposes. No need for a card. The relevant authorities are equipped with biometric readers, and when they need to identify someone, it's a simple(!) matter of checking their fingerprints and iris-scans against the database.

But the Government proposes that this centralised data will be duplicated and stored on our individual personal cards. Functionally, this adds nothing to the system except an extra layer of complexity and cost, and potential security problems. Imagine the scene:

PC: "Excuse me sir, can I see your expensive and redundant identity card?"
X: "Certainly, officer, unless I've left it at home or lost it, or it's been stolen... No, here it is."
PC: "Mm, looks a bit dodgy to me. Would you mind just putting your finger on this scanner, and I'll check you on the central database..."

The card system is equivalent to having biometric locks on your car doors but still needing a key to open them. I suspect that the Government doesn't see the redundancy of the card because seeing it requires what might be called systems thinking. Long ago, when I applied for my first job as a trainee programmer, I had to take an aptitude test to see whether I had the mental "right stuff" to make a programmer. I'd guess that members of the Government might not score very highly in such a test.

I'd also guess that they have been impressed by potential suppliers of the technology required for their system. All of the biometric technologies (face and voice recognition, fingerprints etc.) are prone to error (see this report from the Electronic Frontier Foundation). After September 11, worried American airports carried out many trials of face recognition security systems. Results were, to say the least, disappointing. The American Civil Liberties Union said in a report: "Anyone who claims that facial-recognition technology is an effective law-enforcement tool is probably working for one of the companies trying to sell it to the government".

Perhaps the one consolation is that, given the story so far of government-sponsored IT, there is probably little chance of a working IT card system being implemented. Meanwhile, though, the estimated costs have risen sharply to £5.31 billion. Not quite as much as the Trident submarine programme, but about as useless, and give it time...