thanks for the tips - I've been meaning to check out 'The Selfish Gene'. I checked out the lesswrong.com article - I'm not sure that my argument is the same as what they are trying to disprove - it sounds like their models are based on individuals that collude or act altruistically in order to benefit collectively. I don't think that 'intention' is a good concept for understanding evolution in the first place. Also, from checking out the Wikipedia article on group selection, it doesn't sound like there is scientific agreement that this concept is disproved: http://en.wikipedia.org/wiki/Group_selection

I'll go and read The selfish Gene before I write more, but one more pesky comment about the experiment in the lesswrong.com article: I am suspicious of their conclusion that their case study disproves group selection. This does not seem to respect the vast scope of evolutionary time scale. Maybe group selection is like quantum tunneling - with large clusters of individuals, the chances of it happening are infinitesimal - you won't see it in the laboratory. But on an evolutionary time scale, infinitesimal chances do happen occasionally - and if they yield advantageous results, they will shape life's future. Maybe our love for music came about from freak genetic mutations that in fact didn't serve much purpose for individual survival, but this collective activity caused individuals to synchronize their behavior. Maybe this synchronization helps low-advantage/individual traits 'tunnel' probabilistically and exist long afterward because the same traits are high-advantage/collective traits. (?)

That math doesn't work out the same if the trait doesn't have an appreciable cost to the organism (the case he considered has one of the most extreme costs).