Generative Microdynamics:
A Novel, Unrealizable Epistemology
The second half of the twentieth century witnessed an
epistemological shift, the full impact of which has yet to be fully realized.
Grounded in the work of Goffman and Garfinkel, its focus is on dynamic microinteraction.
However, while the focus is situational and resolutely empirical, the analysis
is scale-free (Hilbert 1992). That is, the most differentiated and/or
hierarchical social structures ultimately emerge from, and are realized within,
interaction situations (Collins 1981).
Collins writes that repeated microbehavior is what produces
social structure, but without specifying the nature or type of the recurrence.
To the contrary, Garfinkel (1967; 2002) and his ethnomethodological associates
document the reflexive and continuously emergent nature of social order (Rawls
1987; 1989). The empirical nature of ethnomethodology focuses upon
the generative nature of human interaction at various scales, in diverse
settings.
The combination of reflexive interaction and generative
dynamics may be helpful in to define the goals of a second generation of
agent simulation (cf., Cederman 2002). First generation social agent
simulation has demonstrated that simple rules are capable of generating diverse
aggregate effects (Schelling 1978; Epstein & Axtell 1996; Axelrod 1997).
The insights generated thereby have yielded important insights. However,
models of discrete agents controlled by exogenous rules, imply, rather than
capture, the reflexive, generative dynamics that characterize living social
processes. Ethnomethodological studies define a standard relative to
which generative models can be assessed.
No formal or computational model can ever be completely
ethnomethodologically satisfying because it can never realize the richness
of ‘ordinary, immortal society’ Garfinkel 2002). Yet, criteria other
than ethnomethodological should be considered as well. By tautly defining
interaction foci, and by clearly specifying the grain of agent perception,
communication and behavior, generative models provide the basis for an innovative
epistemology, distinct from both ethnomethodology and conventional formal
analysis, albeit partially inspired by both.
David L. Sallach
University of Chicago
Social Science Research Computing
www.src.uchicago.edu
sallach@uchicago.edu