Replicating the Classic Sugarscape in MASON

Replication is essential in science and agent-based computational social science is no exception. We present results from a replication of the Sugarscape model (Eptstein and Axtell, 1996) as initially presented in Growing Artificial Societies (GAS ). Sugarscape is a classic agent-based model and contemporary simulation toolkits usually only have a partial replication consisting of a few core model rules without documenting simulation outcomes; code supplied with Repast, Swarm, and NetLogo, for example implement very few of the rules in Sugarscape. By contrast, we demonstrate a detailed replication of Sugarscape in the MASON (Multi-Agent Simulator of Neighborhoods and Networks) environment, and describe various outcomes such migration waves. For major outcomes documenteed in GAS, we describe the degree to which we replicated those results and conclude by outlining major challenges in replication activities and the young field of agent-based modeling in general.