The Signal and the Noise

Would definitely recommend this to anyone interested in stats and forecasting, especially as it relates to how we make policy. Silver's voice tracks his writing in 538 closely: thoughtful, smart, engaging, funny, etc...

I think it's an exceptionally clear and well-written book, especially given the dense topics involved. But this will probably still feel like it's too deep in the weeds for many readers who just aren't interested. On the other hand, in some chapters, I felt like Silver had room to come down with more math and less “here's a vague approximation of what the stats are doing.”

In general, the book's earlier topics (baseball, meteorology, financial crisis) felt a little tighter than the later wandering into climate change. In any case, I think this book is good evidence of why Silver is good at what he does - he thinks about complexity, draws from his background as a well-trained statistician, tries to analyze why his models fail, is reasonably skeptical of overly confident predictions, etc. Portions of the book are lengthy asides into seemingly minute trivia that end up playing a large role in a given prediction's variation – and that kind of attention to bad predictions is exactly why Silver does a better job than most.