An Introduction to Information Theory: Symbols, Signals and Noise
by John R. Pierce
John R. Pierce was uniquely positioned to write this book. An electrical engineer at Bell Labs for decades, he coined the word "transistor," played a central role in developing Telstar (the first commercial communications satellite), and worked alongside Claude Shannon — the founder of information theory — on early work in pulse-code modulation. First published in 1961 and revised in 1980, this book is his attempt to bring Shannon's foundational ideas to a semi-technical audience without stripping out the math entirely.
It covers the core territory of the field: information entropy, how messages can be encoded efficiently, the difference between noiseless and noisy communication channels, and the theoretical limits on how much information a channel can actually carry. Unlike a lot of popular-science treatments of information theory that stay purely conceptual, Pierce includes real mathematical rigor — this is closer to a technical primer than a breezy explainer.
A few things worth knowing before you pick it up: the prose is dense and somewhat old-fashioned by contemporary science-writing standards, and some of Pierce's predictions and framing haven't aged well, simply because the book is now over sixty years old. That said, several of his more speculative asides have turned out to be prescient, particularly around ideas that resemble modern machine learning. It's also inexpensive and widely available as a Dover paperback, which is part of why it's stayed in print this long.
I'd recommend it to readers who want a genuine, historically grounded introduction to how information theory actually works — including the math — rather than a purely conceptual overview, and to anyone interested in Bell Labs–era engineering history. If you want a gentler, more narrative introduction to the ideas first, this is better as a second book on the topic than your first.
View & Purchase on Amazon
