Thinking about organisms as if they were rational agents which could choose their own phenotypic traits according to their fitness values is a common heuristic in the field of evolutionary theory. In a 1998 paper, however, Elliott Sober has emphasized several alleged shortcomings of this kind of analogical reasoning when applied to the analysis of social behaviors. According to him, the main flaw of this heuristic is that it proves to be a misleading tool when it is used for predicting the evolution of cooperation. Here, I show that these charges raised against the heuristic use of this analogy are misguided. I argue, contra Sober, that such a heuristic turns out to be a perfect predictive tool in all relevant contexts where cooperation can at least evolve. Moreover, I argue that it constitutes a powerful and sufficient methodological framework for the analysis of social evolution.