Probably Not: Future Prediction Using Probability and Statistical Inference

Reader Reviews
Amazon Verified Purchase(What's this?) Written in what I call the "textbook lite" style, intermediate between college texts teaching one to do math probability calculations in systematic detail, and the "popular science" style with words but no math. The contents might be put into 4 categories. (1) Initial chapters give clear and unhurried explanations (illustrated by graphs and tables) of the basic math surrounding probability distributions, averages and standard deviations, and how the Normal distribution arises in dice-throwing or random walk. (2) Scattered throughout the book are the classics of elementary math probability: the Binomial and Poisson distributions; conditional probability; Bayes rule illustrated by medical diagnosis; the birthday problem; Monte Hall; Benford's law and Simpson's paradox; gamblers ruin; analysis of games such as double-up-when-you-lose, and the Deal-or-no-Deal TV show. (3) "Popular science" topics such as social networks, stock markets, statistical mechanics, deterministic chaos. The preceding material is presented in a clear and readable way, though not markedly distinct from many previous books of varying styles, for instance (in the same "textbook lite" style) Probabilities: The Little Numbers That Rule Our Lives or Chance: The Life of Games & the Game of Life. The only part that struck me as somewhat distinctive was a fourth category, starting with a discussion of random number generators, and proceding to descriptions of and analysis of simple stochastic models for several phenomena, such as: driving through a series of traffic signals, waiting in a doctor's office, observing some birds flying in a forest and trying to estimate the total number of birds. Overall: well-written, worth reading if you haven't seen such material before, but no particular individual vision and so not a notable addition to existing literature. Like most existing books it is content to repeat theory without critical examination of real-world applicability or significance. The subtitle is misleading -- it's not specifically about future prediction. And at $64 for a 300-page paperback it is considerably more expensive than many alternative books I have reviewed, such as Chance: The Life of Games & the Game of Life or Chance Rules: An Informal Guide to Probability, Risk and Statistics.