Thursday, January 3, 2013

Sharon Bertsch McGrayne, The Theory that Would Not Die ** 1/2

A disappointing history of Bayes' Rule, the insight at the heart of much modern statistical analysis and extremely relevant to Google. In her quest to appeal to general readers, McGrayne doesn't do a good job of explaining the rule or how it is applied. I got only a very vague idea of how Bayesians differ from frequentists.

The main message that I took from the book is that Bayesian inference has many practical applications but is (or was) anathema to statisticians who demand mathematical rigor. The secondary message is that Bayesian reasoning requires computer power, which is one of the reasons it took so long for it to gain acceptance.

The book is more interesting as an example of how personalities and academic in-fighting influence the course of scientific progress. The development of statistics as a discipline flows from a few strong individuals, and if they didn't like Bayes' Rule then it wasn't a proper object of study. The stories of non-statisticians using Bayesian reasoning -- to break the Enigma code in World War II, to set reasonable insurance rates, to locate objects lost at sea -- were entertaining in their own right, but didn't illuminate the central topic.

The long list of interesting applications in the final chapter just made me more frustrated that I didn't understand the theory!

No comments:

Post a Comment