> Nate Silver, baseball statistician turned political analyst, gained a > lot of attention during the 2012 United States elections when he > successfully predicted the outcome of the presidential vote in all 50 > states. The reason for his success was a statistical method called > Bayesian inference, a powerful technique that builds on prior > knowledge to estimate the probability of a given event happening. > > Bayesian inference grew out of Bayes' theorem, a mathematical result > from English clergyman Thomas Bayes, published two years after his > death in 1761. In honor of the 250th anniversary of this publication, > Bradley Efron examined the question of why Bayes' theorem is not more > widely used?and why its use remains controversial among many > scientists and statisticians. As he pointed out, the problem lies > with blind use of the theorem, in cases where prior knowledge is > unavailable or unreliable.
> As Efron wrote, Bayes' theorem is "controversial," but not because > the equation itself is in doubt. (It's a mathematical theorem, after > all; it even has an interpretation in frequentist thinking, albeit > one that doesn't make reference to prior knowledge.) Rather, its use > is sometimes controversial, especially in light of unsophisticated > application of poorly chosen priors. Nevertheless, the real lesson is > that of the sharp kitchen knife that can cut you as well as the > vegetables you're chopping: use the blade of Bayes poorly, and you'll > regret it. Use it wisely, and it will serve you better than the dull > but reliable knife.