Opinion: How best to forecast terror?

The UK terror threat has been raised from “substantial” to “severe"—but what on earth does this actually mean?The official explanation—that an attack is now “highly likely” rather than merely “a strong possibility"—does not make things any clearer.Given that the threat level had stood at substantial since last July until this weekend’s announcement, and there were no terrorist attacks during this period, we can infer that “a strong possibility” indicates an attack has a probability of less than 1% per day. But how much greater is the probability now that an attack is “highly likely”? Would it be 2% per day, or 5%?The obvious solution to this problem is to dispense with verbal labels entirely, and to express risk estimates in numerical terms. This is not a new idea; more than a century ago, William Ernest Cooke , government astronomer for Western Australia, argued that weather forecasters should attach numerical probabilities to their predictions.The idea is often rejected, however, on the grounds that it would be too complicated for most people to understand. This is rubbish. US National Weather Service forecasters have been expressing their forecasts of rain in numerical terms since 1965, and over the years they have got better and better at it. If weather forecasters can do it, why not the rest of us?