A great example comes from the world of medicine. Imagine a disease called Weird Syndrome. Weird Syndrome is pretty rare - but still affects 1 in 1000 people. Medical scientists have created a test that predicts whether or not someone has the disease with 95% accuracy.
Now you're worried that you have Weird Syndrome and decide to get yourself checked out. At the same time, 999 other people take the test. Two weeks pass and the results arrive in a plain brown envelope. Nervously, you open the envelope... and the result is positive: according to the test, you DO have Weird Syndrome. Distraught, you settle down to write your will and organise your affairs as best as possible.
But hold on: do you know how likely it is that you do actually have Weird Syndrome?
You're 95% sure, right?
In fact, it is 98% likely that you DON'T have Weird Syndrome at all. Here's why - and you can play with the figures to your heart's content.This matters enormously when it comes to matters of mass surveillance and identity registers.
For example, when the Labour government was seriously looking into the matter of biometric ID cards, the best performing system - iris recognition - achieved only 96% accuracy. Again, this appears relatively successful on the face of things, but if had been rolled out for use in millions of situations every day it would have led to colossal numbers of people being wrongly identified. It doesn't take much imagination to see how quickly this would have seen criminals going free and the innocent convicted.