“Crime and terror would be better addressed with 10,000 more police and a national border force, rather than wasting £3 billion on ID cards that didn’t protect people in the US or Spain and which would curtail British rights and liberties.”
Lib Dem chairman Matthew Taylor, on the Queen’s speech debate, 23 November 2004

Bertrand Russell once told the story about the Turkey who, noting that he had always been fed at 9 am, concluded that “I am always fed at 9 am” held as a general rule. On Christmas morning, it therefore came as a bit of shock when, instead of getting feed down his neck, he got his neck wrung.

The moral of the story is that although we have no choice but to base our expectations of the future on the experience of the past, as they say in adverts for investment products, past performance is no guarantee of future returns.

Just how we do reliably infer facts about the present and future from the past is the so-called problem of induction, and it has bothered philosophers for centuries. But although the deep philosophical issues remain unresolved, we have some idea what, for practical purposes, counts as drawing an unwarranted conclusion from the past.

Matthew Taylor’s objection to ID cards counts as one such unjustified inference. He points out that ID cards failed to stop terrorist attacks in New York and Madrid. He doesn’t explicitly draw any conclusions from this, but it only makes sense for him to mention these facts if he thinks they are relevant to judging the merits of Britain’s proposed scheme. This he does, as he says the money the scheme requires would serve security and crime-fighting better if spent on “more police and a national border force”.

As an argument this is no better than the old “my grandmother smoked 20 cigarettes a day and lived to 101” chestnut. The right response is, so what? No one says that if you smoke you will definitely die because of your habit. The claim is only that it massively increases your chances of an early death, and no individual chain-smoking centenarian can make that claim untrue.

In the same way, there are lots of ways you can increase you chances of living longer, and the case of someone who didn’t smoke, drank little, exercised everyday and wasn’t overweight, but who dropped dead of a heart attack aged 32, doesn’t change these general facts.

On 9/11, of course New Yorkers weren’t protected by compulsory ID cards, because the US doesn’t have them. In Madrid, they didn’t stop a terrorist attack. But not even their most passionate advocates claim that ID cards are a foolproof prophylactic against terrorism. Like the healthy man who drops dead, the ID-carrying Spaniards who were killed on 9 March 2004 prove nothing about the contribution identification might make to security.

The error of thinking they do is compounded when you consider that we are not even comparing like with like. Britain is not considering adopting the Spanish system, so even if their system did fail to protect its citizens, the British version might well do so. When learning from the past, we have to take into account how present facts and circumstances are different from the past ones we are comparing them with.

The general problem is a too-quick move from “it didn’t work there and then” to “it won’t work here and now”. Several things can contribute to a gaping chasm between the premise based on past experience and the conclusion about what to do now. The first is if the “it” is not the same – or at least sufficiently similar in all the relevant respects – in both cases. The second is if the circumstances of here and now differ in relevant respects from those of there and then. And the third is if the sense of “work” shifts. A measure which is designed to increase or decrease the likelihood of something happening can “work” even if on a particular occasion it didn’t stop or cause that thing to happen. Bans on drunk driving work to reduce road deaths, but they don’t stop all of them. Taylor’s invocation of Madrid in the ID cards debate is weak partly because it falls victim to the first error, but mainly because it is an example par excellence of the third.

I must conclude, however, by stressing that none of this shows that ID cards are a good idea. If you think I’ve argued that, then you need to read an earlier Bad Move.