I've mentioned Nate Silver's The Signal and the Noise before - Michael Mann took a fairly hefty pop at Silver for mentioning an unapproved (by Mann, at least) scientist.
I've now got hold of a review copy and I must say I'm very taken with it. In fact I would go so far as to suggest that it's a must-read for climate scientists.
Apart from the chapter on climate science that is.
Silver says that William Happer agrees that the world will tend to get warmer with increased CO2. Which is fine. But he then seems to spend much of the chapter trying to convince us that CO2 is a greenhouse gas. He tells us that the hypothesis deserves credit because it can be shown to have a cause. But this is to miss the point. The demand for a policy response is not predicated on CO2 being a greenhouse gas. It is predicated on it causing catastrophic warming. Sceptics mostly accept that CO2 is a greenhouse gas, we are arguing about how much warming and what the effects are and whether the whole thing is being hyped.
I was therefore interested to read about how critical climate scientists are of climate models, but this did make me return again to my observations about climate sensitivity. If climate scientists have so little confidence in computer models, how come the IPCC's estimates of climate sensitivity are based on models rather than the less alarming empirical measurements? I think we need to know.
So the climate chapter does not really help very much. But, as I said, the rest is really good - a fascinating tour through the science of prediction, looking at everything from baseball to weather forecasting.