Elinor Ostrom’s paper makes it clear that the set of observed institutions is but a tiny fraction of the total possible permutations of values on the seven institutional dimensions she identifies. This implies that in order to account for the origins of institutions of a particular type, an argument must be made about the trajectories of institutional forms as they appear and move through this state space of possible institutions. The very small ratio between the types of irrigation regulation institutions observed and those that are possible also greatly complicates the use of natural experiments via structured focused comparisons as routes to powerful explanations. If investigators can only arrange controlled “natural experiments” that feature a small minority of the configurations of interest, it is unlikely that the effect of the most potent independent variables will be able to be identified. This is true regardless of whether interest is directed toward patterns of outcomes measured across institutional conditions under comparable environmental circumstances, or with respect to variation in efficacy of a particular set of institutional features under different environmental circumstances.
Elinor’s technique for mapping the state space of institutional forms in the domain of irrigation is of course of great importance for the design and evaluation of agricultural irrigation regimes in third world countries. More interesting from an analytic point of view is that it enables mobilization of powerful theories of evolution and potent computational techniques, including agent-based computer modeling, for purposes of understanding the origins of and variation in institutional forms. For these objectives, the crucial element is that each institutional form is translated into a string of values, each representing a particular feature on the seven theoretically specified dimensions necessary for describing any institution. By operationalizing such strings as distinctive patterns (institutionalized formations of hierarchy, communications, environmental instability, etc.) the performance of these institutions, other implications of variation in these forms, and hypotheses about the implications of such variation, can be explored and tested. Such simulation experiments can involve paired matching experiments or genetic algorithm procedures for exploring the space of the possible for those regions highlighted by different selection criteria.
From a substantive theoretical perspective, the intuition conveyed in the paper hearkens back to standard propositions of organization theory as it developed in the late 1960s and early 1970s. The general argument of this literature, arising from work by Lindblom, March, Simon, J.D. Thompson, LaPorte, and Landau, was that in the face of staggering levels of complexity, organizational designers would do well to opt-out of Weberian or Taylorian ideas about the “one-best-way” to establish and run an organizations in favor of forms that could be allowed to evolve, in a decentralized fashion, to the characteristics of the task environment, the locations and distributions of uncertainty, and variation in the closed vs. open nature of the technologies being deployed. Studying organizations, or institutions, from this perspective, as from the perspective articulated in this paper, entails stylizing the trajectory of the organizational form as the product of an evolutionary, or co-evolutionary, process involving competition among organizational tropes for survival, retention, and replication across time and space. However, although the organization theory literature to which I refer was implicitly evolutionary, it did not see itself in those terms or explicitly harness evolutionary concepts. Additionally, of course, these scholars lacked the computer capabilities necessary to even imagine that experiments could be done virtually to investigate theoretically justified expectations about the affinities that different institutional forms would have with different technological and political “habitats.”
Social scientists are now increasingly able to think in terms of evolutionary and complexity theories and have available to them computational modeling techniques that allow formal analysis of vastly complex domains without being shackled by the constraints of algebra that, in standard rational choice or game-theoretic analysis, prevent multi-dimensional and multi-player problems from being directly modeled. And yet these capabilities are not readily or plentifully available. Most social science practitioners of agent-based modeling are still heavily influenced by the conventional wisdom associated with Robert Axelrod and his students who adhere to the “Keep It Simple Stupid” (KISS) principle—that common sense, not detailed theoretical or empirical guidelines, should be used in the design of agent-based models; that, for the most part, the goal of such work should be restricted to highly abstract insights into general patterns of diffusion, segregation, heterogeneity, or mobilization. However, realizing the potential outlined in this paper will require mastering and employing techniques that use agent-based models, not to explore the non-intuitive implications of abstract and simple conditions, but to measure potent theoretical claims against one another in generic problem situations using ensembles of computational mechanisms as well as to probe the specific implications for particular institutions and institutional settings of variation along key parameters by using sophisticated operationalizations to virtualize situations using real world data.
1. For arguments in favor of leveraging evolutionary theory for social science purposes see Ian S. Lustick, “Taking Evolution Seriously: Evolutionary Theory and Historical Institutionalism,” Polity, (January 2011) pp. 1-31; and “Institutional Rigidity and Evolutionary Theory: Trapped on a Local Maximum,” Cliodynamics: The Journal of Theoretical and Mathematical History, Vol. 2, Is. 2, 2012.
2. For work along these lines see Ian S. Lustick and Dan Miodownik, “Abstractions, Ensembles, and Virtualizations: Simplicity and Complexity in Agent-Based Modeling.” Comparative Politics, January 2009. Vol. 41, no. 2 pp. 223-244; Ian S. Lustick, Brandon Alcorn, Miguel Garces, and Alicia Ruvinsky, “From Theory to Simulation: The Dynamic Political Hierarchy in Country Virtualization Models,” Prepared for presentation at the Annual Meeting of the American Political Science Association, Washington, D.C., September 2010.
Editor’s note: see also Ian Lustick’s Commentary on Wilson and Gowdy