The European Social Simulation Association (ESSA) honored SFI External Professor Joshua Epstein (New York University) with its most prestigious award — The Rosaria Conte Outstanding Contribution to Social Simulation Award. A pioneer and world leader in agent-based modeling, Epstein was among the first scientists to use bottom-up simulation to replicate the statistical macrostructures seen in complex social systems.
Agent-based modeling is one of the trademark tools of complexity science. In contrast to top-down approaches, where phenomena such as cooperation or competition are reduced to mathematical equations, agent-based models are ‘bottom-up.’ They create artificial worlds inside computers by programming individual software “agents" with a core set of rules for interacting with each other and their digital environments. In the canonical Sugar Scape model developed by Epstein and SFI External Professor Robert Axtell (George Mason University) in 1996, agents competed for “sugar” on a digital landscape. They could interact with each other by trading, borrowing, or even inheriting the sugar they consumed and metabolized.
“The idea was how simple can we keep things and still get some core behavior groups, competition within groups, competition between groups and interaction with an environment, all pursuant to Einstein's other great quote, 'as simple as possible, but no simpler,'” Epstein told the ESSA. “So apropos of that, we began to add dimensions to the model — sexual reproduction, tribe, formation, combat, inheritance, trade — we even had a unified epidemiology.”
To the creators’ great surprise, the distribution of wealth in Sugar Scape, which started out random, converged on a Pareto distribution where a small number of the individual agents held the majority of the wealth. They had recreated a pattern first observed in real-world income distributions in 1906 by economist and sociologist Vilfredo Pareto.
Since its inceptionin the 1990s, agent-based modeling has found applications across the social sciences, from archaeology to economics to epidemiology. Epstein’s landmark 2009 Nature paper, “Modeling to contain pandemics,” was published as the world was responding to the H1N1 swine flu pandemic. It described how agent models could giving decision-makers quick feedback on how interventions might play out, and were uniquely suited to capture complex, real-world phenomena, such as irrational behavior.
Epstein is now working at NYU’s Agent-Based Modeling Lab, where he is Professor of Epidemiology. His books include Nonlinear Dynamics, Mathematical Biology, and Social Science (Wiley, 1997), Generative Social Science: Studies in Agent-Based Computational Modeling (Princeton, 2006), Agent_Zero: Toward Neurocognitive Foundations for Generative Social Science (Princeton, 2013), and with Axtell, Growing Artificial Societies: Social Science from the Bottom Up (MIT, 1996).
Watch Epstein’s presentation at the ESSA award ceremony (Sept. 24, 2020)
Read Epstein’s recent op-ed on the novel coronavirus and fear, in Politico (March 31, 2020)