The ID card project is still on track - more or less. Jacqui Smith is just the latest in a long line of Home Office ministers to sell us the benefits of ID cards, while casually informing us of the latest rise in costs or slippage in its implementation schedule. Ms Smith is also yet another Home Secretary who subscribes to the “pixie dust” school of technology: computation is a magic substance to be sprinkled over problems, that, hey presto, then vanish. Little wonder that Britain has an appalling record in government IT projects.

The ID project is one of the biggest computer systems envisaged - far more complex than the failing NHS system. And it's another disaster waiting to happen. Still the politicians naively claim there will be no problems: it will be totally secure because of biometrics. Apparently iris scans, fingerprints, face-recognition software will all work perfectly, be amazingly cheap to implement - and all foolproof. It must be true, as they've been told this by those selling the technology. Baroness Anelay of St Johns, with a group of parliamentarians, was once given a demonstration of a facial recognition system. It failed; indeed the system subsequently crashed, twice. The reason? The baroness was told her face was “too bland”.

(Article continues below)

The only property that all systems have in common is that they fail. And the bigger the system - 60 million entries on a compulsory ID card database - the greater the opportunity of failure. Systems are much like any life form: they degrade over time, they entropy. In the case of databases, the pick up errors and then build data error upon error. The DVLA in Swansea in 2006, for instance, admitted that a third of entries contained at least one error, and that the proportion was getting worse.

We've all had encounters with computer systems that get it wrong. Barclays once refused one of my transactions because they said I was accessing an account owned by a teenage girl named Ian Angell, who lived at my address and was a professor at LSE. I still had to take a morning off work to explain that a 14-year-old couldn't own an account that, according to their own records, had been open for 35 years.

And however scrupulous the managers might be, errors leak and take on a life of their own. They are sampled by other databases, known as “farming”: errors, even when corrected in the original database, live on elsewhere.

But the ID project will be different, we are told. According to the rhetoric, an ID card, one central point of reference, will be so much more efficient and beneficial than you having to prove your identity daily, by producing driving licences, gas bills and so on. Its proponents fail to see that if any of these documents is erroneous, then we don't use the one with, say, a mistake in the address to prove our identity. With the ID card, we won't have the choice. Even if the card is not compulsory, all financial systems will converge on it, and anyone without a card faces great cost and inconvenience. Just like Oyster cards on the London Underground, you're not forced, but it's so much more expensive and tiresome without one.

However, the ID card itself isn't the real problem: it's the ID register. There, each entry will eventually take on a legal status. In time, all other proofs of identity will refer back to the one entry. If the register is wrong - and remember fallible human hands will at some stage have to handle your personal information - then all other databases will be wrong too. Given the propensity of officialdom to trust the details on their computer screen, rather than the person in front of them, you will have to conform to your entry in the register - or become a non-person.

Full article here.