We derive three system order parameters: dynamism, polarization, and clustering, to describe global states of attitude distribution and change for human social systems. Dynamism (f) captures the rate of change in a system, while polarization (Pt) refers to the increase or decrease of a minority position over time. Clustering (e) defines the spatial segregation of opinion based on system topography. These measures suggest a conception of human systems rooted in time and space that is distinct from other approaches. Their utility is illustrated through computer simulations showing that under a wide variety of circumstances, social influence models incorporating spatial distributions lead to unexpected outcomes of incomplete polarization and clustering, with alternative theories of how individuals encode information leading to quantitatively distinct results. A second set of simulations describes the intrusion of temperature, or unexplained randomness into these systems. Surprisingly, the self-organizational tendencies emerging from the iteration of simple laws of individual attitude change derived from Latané's (1981) metatheory of social impact appear to increase with moderate levels of randomness. We consider other approaches for measuring group level processes, among them network analysis-inspired indices and spatial autocorrelation, and suggest how our system order parameters could be used to predict political elections.