It's a book by Robert Axelrod, who set up some groundbreaking game theory experiment/contests back in the day. He set up a computer program that would run other computer programs. Specifically, it ran an iterated prisoner's dilemma game between these programs. He encouraged folks to send him game-playing programs. The iterated prisoner's dilemma (and its variations) gets into interesting territory: trust, reputation, the value of cooperation. Folks might remember ten years back I tried some genetic-algorithm-ish tinkering with the iterated Prisoner's Dilemma which resulted in an incoherent essay. When you see that the title of this book is The Evolution of Cooperation, you might think it's about genetic algorithms, but it isn't really. Axelrod wrote about not-necessarily-genetic evolution based on natural* selection and how it might lead to cooperation. He mostly writes about his programming contests, but there is also an aside about shirkers in World War I: soldiers who wouldn't shoot at their "enemies".
Anyhow, if you read just one book about the iterated Prisoner's Dilemma... then read Poundstone's Prisoner's Dilemma, the popular-science book which covers more material. And if you want to read more details after that, go find some papers by Lindgren who builds upon Axelrod's work and thus gets further. But after those, if you're still curious enough to want to get closer to some of Poundstone's sources, then by all means read The Evolution of Cooperation. I enjoyed it, and you might be mired in the right set of obsessions to enjoy it too.
*Well, computer-simulated-natural, which is the next-best thing, right?