First of all, if you are looking for a book about how the author went about successfully predicting the results of the last two US presidential elections, you will be disappointed. That is not raised at all. It is about predictions in a number of different fields.
It deals with a number of subjects including the difficulties of predicting the stock market, sports (note for non-US readers of the book: this gets pretty boring if you are not a fan of baseball) and betting (a good way of explaining Bayes’s theorem). I wasn’t so keen on the poker section. All this talk of gambling is necessary, I suppose, when you consider that statistics grew out of a desire to predict coin tosses and dice throws.
The chapter that deals with predictions about climate change goes into the most depth when looking at the accuracy of past predictions. It is slightly long-winded, however it is a good exercise in how to judge previous efforts, especially when you are dealing with something as challenging to model as the environment. An interesting idea is that it might be better sometimes to have a simpler model – the results can be more accurate in some cases than something more complex. This is one of the highlights of the book. The look at weather prediction and how it has improved is also very good.
The final chapter on unknowns, “What you don’t know can hurt you”, is the only one that I didn’t finish – its focus on terrorism as an example gets repetitive. It is a valid point to raise though.
This is one of the few books on which I have actually used a highlighter to indicate passages I might want to return to. This goes against my firmly held belief that you should look after books, but it goes to show that there are observations that are of some importance. Even so, it is something that declined the further I went into the book. Does that mean anything? Maybe…
I was hoping for a brilliant book. It’s definitely worth reading but it is not perfect by any means. I think I was expecting too much.