Virtualstan: An Agent-Based Modeling Strategy for the Comparative Dynamics of
Authoritarian Regimes: Bureaucratic Authoritarianism, Bureaucratic Feudalism,
and Neopatrimonialism
Ian S. Lustick
University of Pennsylvania
## email not listed ##
Britt Cartrite
Alma College
## email not listed ##
Abstract
In a significant theoretical and empirical advance in the comparative study of
authoritarianism Barbara Geddes used game theory and aggregate data analysis on the
performance of authoritarian regimes to argue that three different forms of authoritarianism
give rise to different patterns of strain, success, and breakdown. Here we identify certain
problems with Geddes’s approach, in part linked to the limits of her research strategy and
suggest that an agent-based modeling approach to the same problem can substantially improve
the generality and rigor of Geddes’s approach while sustaining her main line of argument and
analysis. Although we are not ready to report the results of our research, we report here in
detail on the development of the Virtualstan model, protocols for its use, and experimental
conditions we have devised. Our dual focus is on systematic differences in the shapes of the
distributions of outcomes produced by batches of computer simulations across three types of
authoritarianism, Bureaucratic Authoritarianism, Bureaucratic Feudalism, and
Neopatrimonialism. Included in our work is a special focus on the impact of succession crises
on outcomes associated with each of type.
Page 1 of 28 
Convention