There is a counterintuitive gap in the club theory of religion. While it elegantly accounts for the success of strict sectarian religious groups in recruiting members and maintaining commitment, it is less satisfactory when attempting to account for groups requiring neither extreme nor zero sacrifice. Moderate groups are always a suboptimal choice for rational, utility maximizing agents within the original representative agent model. The corner solutions of zero and absolute sacrifice, however, are rarely observed empirically compared to the moderate intermediate. In this paper, we extend the original model to operate within an agent-based computational context, with a distribution of heterogeneous agents occupying coordinates in a two dimensional lattice, making repeated decisions over time. Our model offers the possibility of successful moderate groups, including outcomes wherein the population is dominated by moderate groups. The viability of moderate groups is dependent on extending the model to accommodate agent heterogeneity, not just within the population of agents drawn from, but heterogeneity within groups. Moderate sacrifice rates mitigate member free riding and serve as a weak screening device that permits a range of agent types into the group. Within-group heterogeneity allows agents to benefit from the differing comparative advantages of their fellow members.
Also via Kevin Lewis, here is an interesting Dan Ariely paper on who benefits from religion. And here is a rational choice model of papal infallibility.