Proofiness

I'm not really sure for whom Seife wrote this book. The majority of people who like math and/or statistics will already be very aware of most of the statistical concepts that Seife introduces in his book: significant digits, the importance of looking closely at how axes are labelled, appropriate population sampling and correlation vs. causation. And the people who don't like math won't voluntarily read a book on math. So that leaves...I don't know: people who like math but are bad at it? Middle-schoolers? And unfortunately, this book won't work great for those people either, because rather than using the actual names for the mathematical concepts, like I did, Seife makes up terms so that if this is your first exposure to the concepts, you won't actually be able to communicate about them or google more about them. I think my turning point with Seife was in an appendix about the difference between sensitivity and positive predictive value, where I was originally annoyed that he didn't name-check Bayes and then realized that he also didn't mention sensitivity or positive predictive value in the entire appendix even once! This appendix was literally about how just knowing the sensitivity of a test without knowing the prevalence of disease results in not being able to predict the positive predictive value and he didn't use the names for a single one of those concepts.

I found the latter half of the book more interesting: Seife largely moves away from mathematical concepts and investigates political hijinks, such as the Franken election, Bush v. Gore and gerrymandering. It doesn't really add to numeracy, nor have that many striking examples of “proofiness,” (except that humans can't count numbers to 6 digits worth of significant figures, which hopefully most people intuitively know) but it is interesting.

Overall, it's not a bad book. I might give it to a child who was interested in math, but I don't think most adults will enjoy it very much.