The Signal and the Noise, by Nate Silver

I loved this book for the same reason that I loved The Predictioneer's Game and Data, A Love Story. All of them lie at the intersection of math/statistics/data/modeling and psychology/sociology. While I still think Nate Silver is brilliant, after reading this book I have a better understanding of just how terrible at modeling and predicting so many other people are (& I'm talking about people who are paid for making predictions) & why it's so easy for him to look that much more brilliant by comparison.

What it's really about is the use & abuse of statistics & data--what sorts of things can be predicted (short and/or long term) and which kind of can't, the most common mistakes people make when they try to use data to make predictions, and how living in the age of "big data" actually puts us more at risk for bad predictions.

The great tragedy of this book is that the people who are most likely to read it are probably the people who already have a pretty decent understanding of data/statistics, and the people who have a less-good understanding of those things are probably the people who would most benefit from reading it.