Tuesday, April 28, 2015

April's Book: How to Not Be Wrong

How to Not Be Wrong by Jordan Ellenberg
Math (2014 - 322 pp.)

How to Not Be Wrong is a bluntly titled book that, to borrow a phrase from Cleveland-based paint giant Sherwin-Williams, does exactly what it says on the tin. It is a book on how math can be applied in everyday life, divided into five easy parts: Linearity, Inference, Expectation, Regression, and Existence. The opening section, "When Am I Going to Use This?", shows the kind of jocular spirit Jordan Ellenberg has in writing the book. It is very clear he does not intend a dry academic textbook. He has succeeded in making math accessible to anyone willing to read what he has to say.

Some of the book's most entertaining moments are when seemingly bizarre occurrences have logical, sensible mathematical explanations upon a little thought. Three of these occur in the Inference chapter, which focuses on probability and Bayes' theorem. The Baltimore stockbroker story illustrates the problem of information asymmetry applied to a surprisingly high number of observed events. "Dead Fish Don't Read Minds" shows the hilarious consequences of failing to account for a control group. "The International Journal of Haruspicy" is perhaps the starkest example of confirmation bias I have ever seen; a scientist hoping to find predictions of world events in sheep entrails can, well, find those predictions if he or she looks hard enough. Bonus points if you knew the word "haruspicy" in advance, which I did not.

Ellenberg's frequent use of charts and graphs makes the book more fun. He scrawls them just as professionally as is necessary to explain his point, with axes often pointing off in not-so-perfect-right-angle directions. In "Galton's Ellipse", in the Regression section, Ellenberg explains the trigonometric foundation of correlation in such clear words and pictures I felt sheepish* having either not known it before or having forgotten it. "Does Lung Cancer Make You Smoke Cigarettes?", unsurprisingly in the Regression section, speaks to the fallacy of confirming the antecedent (i.e. X --> Y, therefore Y --> X), and then shifts gears to show in very simple visual terms why unattractive ugly men go unnoticed. Perhaps best is when the charts and graphs combine with pop culture. Ellenberg's opening chapter, "Less Like Sweden", explains the Laffer curve in a simple graph with support from the classic movie Ferris Bueller's Day Off.

I would love to read an advanced version of How to Not Be Wrong, incorporating concepts my exposure to statistics and economics has given me a lifelong interest in. Unfortunately, I doubt this is forthcoming. The whole thing makes me want to get my PhD and write the How to Not Be Wrong/Freakonomics of game theory.

Even if you know all the math in here, How to Not Be Wrong is a fun read. I finished it two weeks ago, making this the longest between finishing and posting about a book for this blog. For any given person, there will probably be something new to discover. For me, it was the life of Francis Galton, who was simultaneously ahead of everyone else in statistical research yet stuck in his time regarding eugenics. For the record, yes, you will use all the math in this book, and no, that will not only be the case if you become an engineer, actuary, or whatever other career that chart in my old high school mentioned.

Ease of Reading: 6
Educational Content: 9

*Is this a haruspicy joke? I think we should say it is.

No comments:

Post a Comment