Cool take on things, but not really fundamentally new as an explanation of evolution of cooperative traits. Models involving reputation and punishment have been around for awhile. Basically, once you introduce reputation and memory, together with punishment, cooperation becomes more viable as a strategy. It's even led to models of higher-order cooperation, of enforcing the enforcers, etc.
To me it seems like the paper isn't really offering a fundamentally new explanation of how these traits arise, it's just providing a sophisticated analysis of the dynamics.