Customer Reviews

From Cosmos to Chaos: The Science of Unpredictability

5 star:		(1)
4 star:		(0)
3 star:		(1)
2 star:		(0)
1 star:		(0)

 

 

The astronomer author sets out "to cut a broad swathe through the physical sciences" emphasizing how they relate to probability. He succeeds in creating a well-written and interesting book, treating a range of topics which the author himself has clearly thought about carefully. It is more sophisticated than most popular science books, both in writing style and by including occasional equations, but it is still accessible to the motivated reader.

Two thirds might be described as "standard" topics, but the author often presents some slightly novel angle. For instance, the brief history of probability naturally focuses on astronomers and physicists, e.g. Gauss and least squares. For elementary mathematics of probability, a routine discussion of poker deal probabilities is followed by a more elaborate analysis of finesse-or-drop in bridge. A 23 page chapter outlines some less elementary mathematics of probability (Markov chains, computer simulation, Kolmogorov complexity, chaos, the first digit law, random point processes being more clustered than untutored intuition suggests, Brownian motion) in admirably concise yet informative style. Turning to physics, there are three remarkably lucid chapters which contain more solid content than most non-mathematical treatments. One on statistical mechanics and entropy, another on quantum theory, and the third on the standard theory surrounding the Big Bang. Probability enters into the latter because random spatial density fluctuations in the early universe are presumably responsible for the later emergence of structure.

The other third of the book is a thread starting from the classic frequentist-Bayesian debate, moving through the "maximum entropy" principle (well established in a statistical mechanics context) for choosing Bayes priors where little information is available, and ending (Chapters 9 and 10) with frankly speculative material on the theme of the observable Universe as one random realization of many possible (or one amongst many existing but unobservable) other Universes, thereby seeking to infer something from Bayes rule. The latter leads to technical topics such as a prior distribution on the matter density parameter Omega; and to more philosophical ones such as the Anthropic principle, via the Fermi paradox and the Doomsday argument. Most scientists (and I) are skeptical of such topics, regarding them as based on inventing priors on mathematical-philosophical rather than scientific grounds, but motivated readers will find it interesting to ponder the arguments for themselves.

The book is based on a grand idea of uniting probabilistic thinking from games of chance to the universe as a whole. Even though each of the chapters are OK, and some are really good, the bridge from bridge to Big Bang is not very strong. Coles is not afraid of mathematics: all the major equations are presented, and this will distract some of the fainthearted readers, but the mathematically inclined will love it because of the conciseness of explanations.