The shopbot model portrays a commodities market in which all buyers seek to minimize the price they pay for the commodity. In many markets, however, products are a good deal more complex and configurable, and consumers may be concerned less with price than they are with other attributes of the product. Information goods and services will have a number of configurable parameters. Seller agents will both price and configure information goods and services, and buyer agents will need to apply complex multi-attribute utility functions in order to evaluate and select them.
In an effort to understand how agents might deal with complex multi-attribute goods and services, we have explored a variety of models that emphasize different aspects of product differentiation. This section provides an overview of three such models. The first deals with a vertically differentiated market in which products are distinguished by a simple ``quality'' parameter, and consumers have different tradeoffs between price and quality. Here we observe that the market can be prone to price wars that are more complex in form than those observed in the shopbot model. The second and third models explore two related scenarios involving horizontal differentation: information filtering and information bundling. In both models, different consumers value various categories of information differently. Several important issues surface in these studies, including the emergence of cyclical behavior in price/product space and its detrimental effect on sellers and buyers alike, economic incentives for specialization, and the interplay among learning, optimization, and dynamics in multi-agent systems.