1 Agent-based Model of a Simple Prehistoric Exchange System Computer Simulation in Social Sciences, Spring 2006 Joshua Watts and Yu-Ru Lin {Joshua.Watts, Yu-Ru.Lin}@asu.edu Abstract We describe an agent-based model inspired by the prehistoric Hohokam pottery exchange system from central Arizona, circa A.D. 500 – 1400. A conceptual model informed by archaeological research in the region and influenced by a small body of theory specifically focused on the economics of small and middle-range societies is outlined. Details of our formalization and implementation of that conceptual model using the Netlogo platform are described. Finally, the results of our sensitivity and model dynamics experiments are presented. The results show that simulations of our agent-based model generated ceramic distribution patterns quite similar to those predicted by theorists, and may prove to be a useful starting point for future modeling efforts involving more concrete archaeological datasets. 1. Introduction This report summarizes our work on an agent-based model of an abstracted pottery exchange system. Much of the model’s implementation and successive simulation was influenced by the prehistoric Hohokam economy as understood by archaeologists who work in central Arizona. Our model was designed with the goal of eventual comparison to archaeological datasets. To date, our work on this model has focused on the effort to understand how a small number of key parameters affect resulting distribution patterns of discarded pottery on a landscape. Do agents using different logic for finding trading partners cause different patterns of pottery distribution? Specifically, do agents participating in market-like exchange networks generate different distribution patterns than those agents only participating in down-the-line trades with nearby agents? Lastly, are there situations in which similar distribution patterns may occur, but very different exchange logic was employed by the agents? Archaeological theorists have addressed these questions on a conceptual level, but rarely provide a more rigorous evaluation of their assumptions. This modeling and simulation project introduces an effort to methodically explore and evaluate some aspects of a prehistoric, pre-marketplace trading system. 2. Background 2.1 Archaeological and theoretical setting Archaeologists studying prehistoric small and mid-range societies rarely, if ever, have the luxury of direct evidence for the kinds of economic activities engaged in by their subjects. Scholars of complex- societies research may recognize actual archaeological remains of marketplaces, or may even have textual records of the kinds of goods circulated and the ways those good moved around a community (Zaccagnini 1979, Hirth 1998). Prehistoric peoples of North America left no such evidence, and archaeologists instead are forced to distill relevant data from faint traces of past behavior. It has long been recognized that prehistoric peoples, especially in the Southwest USA, traded craft goods over enormous distances. Typically those items were luxury or ritually significant items such as shell jewelry, turquoise, copper baubles, or exotic birds; this exchange was not critical for the survival of the parties involved. These kinds of systems often result in distribution patterns where items are found less and less frequently at greater distance from the source, according to Renfrew’s (1975; see Hirth’s spatial evidence [1998]) law of monotonic decrement. Down-the-line trading of utilitarian items, such as ceramic cooking vessels also typically causes a predictable spatial fall-off of goods away from their