An Introduction to Information Theory: Symbols, Signals and Noise

by John R. Pierce

Symbols, Signals and Noise lies somewhere in the aether between textbooks and popular science. It is certainly more easy to read than many examples of the former, with careful explanations in plain language of the phenomena being discussed, and intuitive explanations of any important equations or discoveries, yet it is also certainly a deeper treatment than a swathe of modern books which present only gaudy tidbits of information and some patronising examples. Pierce stresses that his book would be empty without maths, and aims to fill in the gaps in laymen's understanding rather than skim over them -- even if he does omit proofs in many cases.

Most notable about the book is its readable style. Rarely does treatment of a scientific topic have this narrative pull which keeps you reading, or such a sense of an author's voice. It is a pleasure to learn from this character, whose intuitive explanations, scientific modesty and occasional spot of mild humour all work to educate you about a deep and broad topic, both in its core components as established by Shannon and in its applications to fields as diverse as linguistics, psychology and physics.

The book is not without flaws, but these are for the most part textual. The terms 'information theory' and 'communication theory' are both used, seemingly interchangeably throughout the book, with no real explanation of why this is so. There are a number of typographical errors in unfortunately critical areas -- numbers which are mistyped, symbols which are exchanged in the middle of important explanations. I entertained for a while that the book itself was demonstrating a form of error correction -- the redundancy in the text being sufficient to allow the detection and correction of errors. In his chapter on cybernetics, Pierce discusses for a while computing and predictive programs, and his treatment here has dated quite amusingly, not that it is any fault of his that the history of computation has changed since the book's publication, or that problems he saw unsolved have been tackled with great energy.

Certainly a great read for anyone with an interest in the topic, and I daresay a valuable guide to understanding for an undergraduate in communications or computing, or a professional moving from another field.