杏吧原创

In The Theory That Would Not Die, Sharon Bertsch McGrayne tells the surprisingly gripping story of the theorem that pervades modern life

FEW single ideas can boast such an impressive r茅sum茅. What else could have possibly cracked the Enigma code in the second world war, predicted John F. Kennedy鈥檚 close-run win in the 1960 US presidential election, and landed a winning blow against the nefarious antics of the tobacco industry?

The secret to these successes is a humble little theorem called Bayes鈥檚 rule that allows you to assess the probability of one event from observations of an associated event. On election night, for example, you might use an early outcome from one US state to help predict the winner in states that have not yet announced their result.

This simple rule has experienced a turbulent history since Thomas Bayes outlined his ideas in the mid-1700s. The sticking point has been the fact that you must first take a stab 鈥� sometimes not even a very well-informed one 鈥� at the relative probabilities of your hypotheses, before plugging in the relevant data to achieve a more accurate answer. Researchers trying to root out any form of subjectivity in their methods found this kind of guesswork outrageous, leading to the theorem being shunned over the next two centuries.

Sharon Bertsch McGrayne鈥檚 The Theory That Would Not Die is the first popular science book to document the rocky story of Bayes鈥檚 rule. At times, her tale has everything you would expect of a modern-day thriller. Espionage, nuclear warfare and cold war paranoia all feature as she tracks the theory鈥檚 crucial role in Alan Turing鈥檚 code-breaking during the second world war, and the US navy鈥檚 later use of the technique to track Soviet submarines.

Eventually, the theorem wormed its way into medical journals, insurance forecasting and business decision-making, before its eventual triumph in the age of the internet, as it became essential for spam filters, translation software and much more besides. In these less dramatic periods, a host of colourful characters and their bitter rivalries carry the tale. McGrayne can capture these mavericks in a single sentence: we are told that Robert Osher Schlaifer, who helped introduced the theorem to business, was 鈥渉ot as a pistol, sharp as a knife, clear as a bell, quick as a whip and as exhausting as a marathon runner鈥�.

鈥淏ayes鈥檚 theorem became essential for internet spam filters, translation software and much more鈥�

Despite these colourful descriptions, the book is not always an easy read. Even with a mathematics degree, I struggled to visualise just how Bayes鈥檚 rule solved many of these problems. And while McGrayne鈥檚 writing is luminous when describing the drama of the cold war, her attempts to portray earlier periods in a modern context sometimes fall flat. We are told that a theological pamphlet was 鈥渁 kind of blog鈥�, for example, and she feels the need to describe a philosophical thought experiment as a type of computer simulation.

These are minor gripes, however, and to have crafted a page-turner out of the history of statistics is an impressive feat. If only lectures at university had been this racy. Then again, perhaps we should expect nothing less from a theory once described by one of Google鈥檚 representatives as 鈥渢he crack cocaine of statistics鈥� seductive, addictive and ultimately destructive鈥�.