What this book does well is to show how valuable numerical information can be if you know how to analyse it properly, and how unreliable human judgment can be compared to what the data can tell us. The examples are well chosen and telling. Where it goes too far is in expressing almost unqualified belief that the number stuff trumps everything else, all the time. Contrary to the title, not everything can be predicted. Specifically, the sudden turns of events, in markets or social behaviour (which are often what we are most interested in) are not predictable on the basis of past patterns. By defintion, surprises are not predictable. And there always has to be a human doing the modelling, working out the dynamics of any set of causal relationships. Sometimes, these people get it wrong. Then it's the old case of "rubbish in, rubbish out." But though he has pushed the argument too far, he is interesting, imaginative and persuasive that the balance could be tipped a little further than it now is, towards making better use of statistical data.