Peter Godfrey-Smith is among the very best of a new breed of philosopher of biology whose contributions are very strong both in the biological sciences and the philosophy thereof. This book reviews some major controversies in evolutionary biology over the past half-century, with the author's own attempt to adjudicate among the contestants.
It is not clear to me to whom this book is targeted. Godfrey-Smith describes the controversies at too high a level for a novice reader, but the descriptions are too detailed and labored for a reader who has followed the debates. His own contributions are very powerful in some cases, while in others he more or less follows the lead of others, supplying little that is new at all.
His most important point is that the replicator/vehicle approach to Darwinian evolution was motivated by paradigmatic controversies unleashed by Dawkins' The Selfish Gene, and is really a step backward. He shows rather definitively that the standard definition of Darwinian evolution by Lewontin and others is both adequate and much broader than that proposed by the replicator/vehicle proponents. Indeed, he gives examples of Darwinian evolution where there are no replicators at all. Godfrey-Smith also claims that the replicator/vehicle approach often embraces and ``agential'' view of genes, endowing them anthropomorphically with "plans" and "objectives." He does not make clear why this is a problem. I do not believe it is. There is nothing wrong with game-theoretic behavioral models in which different alleles at a locus are associated with different phenotypic behaviors. This "phenotypic gambit" has allowed evolutionary biology to develop very powerful models of strategic interactions within species.
Godfrey-Smith also presents a novel and interesting critique of the "gene's-eye view" approach to Darwinian evolution (which is of course closely related to the replicator/vehicle view). While recognizing that sometimes the gene's-eye view is the correct way to view the problem, it is not correct in general because the notion of a "gene" is simply not sufficiently well-defined to serve as a "Darwinian population." While this critique is novel and interesting, I do not think it is more powerful than the traditional critique that genes evolve only as part of gene complexes, and the highly non-linear interaction among genes renders the idea that we should always model evolution at the level of the gene implausible