Book review: ‘How to Predict Everything’ by William Poundstone

During one of the more tense phases of the Cold War in 1969, unknown astrophysicist and Harvard graduate J Richard Gott III visited the Berlin Wall. Standing before this bleak monument to totalitarianism, the young Gott wondered idly how long it would be before it was demolished. On the spot, he devised a simple equation, the outcome of which was his prediction that the Wall would stand for at least another 2.66 years, but would no longer be with us in 24 years. In 1987, President Ronald Reagan told the Russians to “tear down this wall” and by 1992 it was demolished - 23 years after Gott’s prediction. Gott was right.

Gott had devised the ‘Delta T argument’, a simple trick of mental maths that can be applied to predicting virtually anything, from how long your marriage will last to how long the Simpsons will continue to run. It inspires and infuriates in equal measure because, while ‘Delta T’ is almost certainly pseudo-maths, it’s also very accurate - at least, when it works.

In 1999, The New Yorker ran a profile on Gott, who was by now a minor celebrity, under the headline ‘How to Predict Everything’. This in turn provides William Poundstone with the title for his new book (OneWorld, £12.99, ISBN 9781786075710), in which he examines the maths behind telling the future, including the median birth-clock Copernican prediction which, if applied to current population estimates, states that our civilisation has only another 760 years before we become extinct. He assesses how to predict the notoriously unpredictable stock markets, citing John Burr Williams, the Harvard economist who devised the formula that future dividend payments are equal to current dividend-paying stock. The book is full of powerful statements and anecdotes such as these.

Where it gets really interesting is when Poundstone turns his attention to the mathematics behind an arcane paper written by an unknown clergyman of Tunbridge Wells by the name of Thomas Bayes – a paper that remained unpublished in his lifetime. Even after the 18th-century minister’s ideas were rescued from obscurity and tidied up by more orderly mathematicians they gained little traction, due to the fact that the iterative process of making inverse probability adjustments required too much work. That was until computers came along, after which Bayes’ discoveries became the bedrock of the modern maths that goes into predicting which of our emails are spam and much more besides.

Given that Poundstone is a renowned sceptic, it should come as no surprise to find chapters such as ‘Twelve Reasons why the Doomsday Argument is Wrong’ and readers will discover that the author’s ability to argue against the probability theories he investigates is one of his key strengths. Above all, ‘How To Predict Everything’ is thoroughly entertaining reading and it’s not hard to foresee a future in which readers everywhere will find it impossible to put down.