Sunday, May 29, 2005

ID Cards - As Bad as you Thought, But Worse!

The Government's crusade to breathe life into its dead ID Cards scheme ran deeper into trouble this week. First, along with the old-new Bill, the results of the Home Office's trial of biometric identifiers were out. The HO tested its gizmos on 10,000 guinea pigs/citizens, and came back with a best result of a failure rate of 4%. Now, 4% does not seem very much. It seems even less if you call it a success rate of 96%. But, as Karl Marx said, quantity has a quality all of its own. 4% of 10,000 means that 400 people were misread, misidentified, or not identified at all.

As the ID card and its monster database threaten us with a fine of up to £2,000 for anyone whose card is "damaged", even if they don't knows it (that is to say, anyone whose card does not swipe properly), and much worse for anyone who has a false card or someone else's card (that is to say, anyone whose card the machine misreads), this would have been 400 possible miscarriages of justice just from the trial alone. If "miscarriage of justice" is too much, then it was at least 400 cases of inconvenience, embarrassment and administrative cost. That was for the sci-fi super soaraway iris scan. All the other options were worse by an order of magnitude at least.

Now, there are 60 million or thereabouts people in Britain. 44 million are slated for tagging. Even as a one-off, that makes 1,760,000 cockups from the iris scan alone, not to mention the hopeless facial scan, which failed on 30% of cases.

Better yet, the system didn't just not work, it discriminated. Older persons and - guess who? - black people were far more likely to be misidentified. That is to say, one of the biggest target-groups for police ID checks is also the most likely to be victimised by the machinery. Advice: if the ID Cards Bill comes in, and you're black, get a good lawyer now.

Today's Observer pours on the vitriol. According to a study carried out by the London School of Economics, it seems, the cost of ID cards and a monster database of absolutely everybody will be north of £12 billion and nearer £18 billion, or £300 a card. Not only that, they argue that the cards will need renewing twice as often as the government says. And changes of address will require something between 300 million and 1.2 billion changes to the Big Database over 10 years. We are looking at an IT equivalent to the Tower of Babel here. To get that error rate down to non-terrible levels, the whole thing will have to be reliability-engineered to the standards of the public-switched telephone network. Which costs money. Lots of money, especially when it's not just a database but also a biometrics system. 99.999% reliability on 99.999% uptime would still leave us with 44 wrongful arrests, though.

Eighteen billion quid. For what? To save us whatever fraction of the highly dubious "£1.3 billion" actually is attributable to frauds the cards might stop? Given that a majority (let's say, for argument's sake, half) of that figure is credit-card not present scammery, that's £650 million a year. On the lowest costing, we would be paying £1.2 billion a year to save £650 million. This is madness.

No comments:

kostenloser Counter