Can a crowd, as a whole, guess the weight of a cow better than any individual within that crowd? Yes, and Surowiecki tells us that it is because of the wisdom of crowds (TWOC), a situation where a diverse set of individuals, each with their own private knowledge acting independently, arrives at a solution to a problem. This assumes that there is some way to combine the solutions of the individuals; in this example, taking an average of the guesses suffices. Surowiecki begins his book with this retelling of the cow weight-guessing experiment by Francis Galton in 1906 that highlights the necessary criteria of diversity and independence, either of which, when not present, can cause TWOC to be not so wise. A common example where the crowd is unwise is herding, when dependency creeps in and individuals start piggybacking and following others.

The book covers a lot of ground, drawing on examples from politics, economics, psychology, general science, sociology, and computer science. This has become a popular business book, yet there is plenty in the book to keep the computer scientist smiling. Why does Google work so well? Because a large number of diverse and independent Internet users continually “vote” on Web sites when they visit them, or link to them from other Web sites. Surowiecki provides a similar explanation for why Linux and the open source movement provide useful and robust software that is better than that of many corporations. Many of the examples discussed in the book are amenable to computer simulations, such as why birds flock, and why traffic can either flow smoothly or break down into a jam for no apparent reason.

Much of the book concentrates on providing examples of why businesses succeed or fail, and why financial bubbles and crashes occur. Surowiecki provides numerous examples of where TWOC succeeds or fails, along with explanations relying on the conditions of independence and diversity. While the examples provide convincing evidence of TWOC, I was still left with a gnawing feeling that this is all a little magical. What are the fundamental underlying scientific or mathematical principles involved? The book does not provide a rigorous explanation of this. Detailed notes at the end of the book provide ample references to the literature.

This is a well-written and intense read. For computer scientists, TWOC is reminiscent of cellular automata, where independent agents interact with neighboring agents, often performing something useful, or with some emergent behavior. This book is useful as a rich source of computer science applications, and may even provide a framework in which to think about potential solutions to problems (think diversity and independence). Buried deep within the book are some popular accounts of classic computer science examples, including single-server and multiple-server queueing simulations, graph problems, and social networks.