Slow, and worth it.

A student named Bethan told me in week four that she'd stopped reading somewhere around the discussion of regression to the mean, because the book kept catching her out and she was tired of being caught out. I have thought about that admission more than I have thought about most of the arguments in here. Kahneman is doing something the eighteenth-century moralists attempted and mostly botched: demonstrating, with evidence rather than wit, that the mind is not a reliable witness to itself. Mandeville would have adored him. Mandeville would also have found him rather slow.

Because it is slow. Parts I and II earn the reputation; Part III on overconfidence is the finest sustained stretch of exposition I have read in a popular science book, and I say that as someone who reads very few of them. Part IV, the prospect-theory material, is where the general reader quietly begins to skim, and I can hardly blame them. The final quarter is a man refusing to abandon a result he loves.

The prose is plain to the point of dryness, which suits the material. No metaphor is asked to carry weight it cannot bear. He tells you where the evidence is thin, which is more than most in the genre bother with, though a book published in 2011 could not know how much thinner some of it was about to get.

I still assign chapter 19. Every year. It does more for undergraduate essay writing than any style guide I own.