Introduction to Good (1952) Rational Decisions

Abstract

British statistics in 1951 was dominated by the ideas of Fisher and Neyman-Pearson. The decision-making ideas of Wald (1950), by then becoming popular in the United States, were just being noted by theoreticians but never became a significant force. Savage (1954) had yet to appear and even the name of de Finetti was virtually unknown. The work of Jeffreys (1939) lay largely unread. In September ofthat year, the Royal Statistical Society held a conference in Cambridge. An account of it is given in J. Roy. Statist. Soc., Ser. A, 115, 568. At the time, I was a demonstrator, a title suggesting, but not actually implying, practicality, at the university, attended the meetings, and even gave a paper. Regrettably, I have no memory of them, only of the Masonic Hall in which they were held. Into this environment came I.J. Good with this extraordinary paper that is totally outside the Fisherian mold and a precursor of subsequent, Bayesian developments, The Cambridge audience could have had a foretaste in Good’s (1950) book, but the paper is devoted to rational action, whereas the book was more concerned with rational thinking. I regrettably found the book’s heterodoxy unimpressive and was presumably equally unappreciative of the talk. How wrong that attitude was! Good, extremely succinctly, describes an all-embracing philosophy of statistics and mentions several applications. Important ideas are scattered throughout and often only receive brief mention despite their importance. Many of them are original and even those that are not are treated in a highly original way that shows them in a new light.