On one level, this paper is frustratingly similar to my work. On the other hand, it allows for easier justification of my methods. Good work.
The Model:
Nowak-style PD game. n X n lattice (n = 100) with each cell a player. Each time step agents play all their neighbors. Here's the catch: wealth from payoffs is accumulated. In addition, the investment is determined by a coefficient alpha (equal to 1 in the baseline model), and multiplied by the poorer agent's resources. The investment is then multiplied by the baseline payoffs (T > R > P > S). This creates a "rich-get-richer" effect. All players then adopt the strategy of the neighbor (von Neumann neighborhood) that had the highest payoff that round if higher than its own.
They start with a random population of 50% cooperators, R = 1, T = 2, P = S = 0.
Results
Initially, defectors dominate by exploiting cooperators. However, clusters of cooperators can form, and grow richer by cooperating with one another. This does not happen without wealth accumulation. In fact, without wealth accumulation, the proportion of cooperators goes to zero (since T is large).
They show that wealth accumulation also allows cooperator to do better for a wider array of payoff matrices, extending into other games such as snowdrift and stag hunt.
The larger the multiplier alpha, the better cooperators do. When it is equal to or greater than 1, cooperators almost always dominate. The final proportion of cooperators decreases with alpha.
The results are robust with using a Moore neighborhood. Although you need a larger population to ensure that initial clusters form, those clusters are better able to resist invasion.
No comments:
Post a Comment