Spite, the shady relative of altruism, involves paying a fitness cost to inflict a cost on some recipient. William Hamilton famously argued that spite is unlikely to ever evolve on the grounds that such harming behavior would destabilize a population, leading to extinction. Here I will present some collaborative work on evolutionary game that demonstrates the stability of spite in a range of conditions. I then use the models to refine Hamilton's argument, and raise some theoretical puzzles for the evolution and stability of prosocial behavior. In particular, I will discuss an apparent paradox: as the fitness benefit from prosocial behavior increases, that behavior becomes less likely to evolve. I will diagnose the evolutionary processes behind the result, and discuss the consequences for our understanding of biological and cultural evolution.