Jeffrey O. Kephart and Amy R. Greenwald
IBM Institute for Advanced Commerce
IBM Thomas J. Watson Research Center
P.O. Box 704, Yorktown Heights, New York 10598, USA
kephart@watson.ibm.com, amygreen@cs.brown.edu
Shopbots are Internet agents that search for information that pertains to the price and quality of goods and services. With the advent of shopbots, a dramatic reduction in search costs seems imminent, which promises (or threatens) to radically alter market behavior. This research includes the proposal and theoretical analysis of a simple economic model that is intended to capture some of the essence of shopbots, and attempts to shed light on their potential impact on market behavior. In addition, experimental simulations of an economy of software agents are described, which are designed to further study the dynamics of interaction among electronic buyers, sellers, and shopbots. The desire to explore the economic impact of shopbots as sources of price and product information leads to a model that is similar in spirit to those that have been investigated by economists interested in understanding the phenomenon of price dispersion. The underlying assumptions and methodology of this research, however, are directed toward the eventual goal of designing economic software agents, and thus differ substantially from previous economic studies. A series of shopbot experiments are discussed in which (i) search costs are nonlinear, (ii) some portion of the buyer population does not have access to search mechanisms, and (iii) shopbots are economically motivated agents that strategically price their information services in an effort to maximize their own profits. One of the main contributions of this paper is that it explores some of the ways in which economics of the Internet -- particularly, that which is conducted via software agents -- vastly differs from traditional economics.