Computer simulations have long been used as an effective tool in engineering, economics, psychology, and a number of other social sciences. Engineers typically use simulations to predict performance of a system that has known dynamic characteristics. These characteristics are typically obtained from theory and are then articulated in the simulation as difference or differential equations. The goal of engineering simulation is then to assess the dynamic performance of a system based on a priori knowledge of the dynamic relationships among the various elements of the system. Forrester (1961, 1973) was one of the earliest and most influential advocates of simulation modeling of dynamic social systems. Forrester advocated this approach as a way to model and assess the dynamics of industrial and world phenomena. Sterman (2000) provides a recent review of research on dynamics simulation from this tradition. While this approach has produced a considerable number of studies, it too is based on the assumption that the researcher has a priori knowledge of the dynamic relationships among elements of the system. Indeed, many of the results of these models have been criticized for specifying relationships that were at best untested and at worst flawed. In response to these criticisms, more recent interest has focused on redefining the utility of simulations in the social sciences. Rather than using simulations to test the long-term dynamics of systems with known interrelationships, theorists (Carley & Prietula, 1994; Contractor, 1994; Hanneman, 1988) have suggested that social scientists should use simulations to help construct theory, to identify the heretofore-unknown interrelationships. This section describes the traditional use of computer simulations as well as the adaptation of this approach toward theory construction and testing in the social sciences. Later sections will apply these general approaches to the computational modeling of networks in particular. Carley and Prietula (1994) suggest that the emergence of the new field of computational organizational theory (COT) signals the growing interest in the construction of computational models to augment and assist theory building. Most social science theories are richly evocative but highly abbreviated (Poole, 1997), that is, they offer explanations that suggest complex interrelationships but do not provide precise, falsifiable mathematical formalizations of the theory.