Aleksa's Book Review: The Evolution of Cooperation

Other than being one of my favourite fictional TV hedge fund managers out there, Robert Axelrod also wrote a book. This book is a uniquely simple introduction to the ideas of advanced game theory and behaviour optimisation. Dr. Axelrod opens the book with an explanation of a "tournament" he held during the early days of the computer era, in which the standard prisoners dilemma problem was opened to competition: contestants would write programs that would decide to "cooperate" or "betray" under various circumstances.

During this iterated game of prisoners' dilemma, various programs were brought about in varying stages of complexity. The strange thing is that the most successful program by far ended up being the TFT or "Tit for tat" program, which would just cooperate by default at first and then do whatever the other player did afterwards, always tallying up a score that's at least equal to that of the other party because of the opportunity to learn on part of the "adversary".

Axelrod then goes into how songbirds, nation-states, local businesses and many others choose to cooperate (or not) depending on their likelihood of repeat interaction, with plenty of mathematics from the tournament to back up his claims. It's nowhere near water-tight, but the arguments that he makes for the long-term viability of a cooperative-tit-for-tat model in anything from foreign policy to negotiation are very compelling.

At the final stage of the book, he explains how native and outsider models (memes) function within a network, and how even though you may have a more complex and foresightful model as either the native or the outsider, the model that fosters long-term cooperation is the one that will win out simply because of the ability of learners to understand that it's in their best interest. Whether or not this is a justification of a whiggish view of history is up to the reader. I'm highly impressed by the book.
9/10