Social-Behavioral Modeling for Complex Systems

28 Lessons from a Project on Agent‐Based Modeling

Mirsad Hadzikadic¹ and Joseph Whitmeyer²

¹ Department of Software and Information Systems, Data Science Initiative, University of North Carolina, Charlotte, NC, 28223, USA

² Department of Sociology, University of North Carolina, Charlotte, NC, 28223, USA

Introduction

Hadzikadic, O'Brien, and Khouja published an edited volume Managing Complexity: Practical Considerations in the Development and Application of ABMs to Contemporary Policy Challenges in 2013 (Hadzikadic et al. 2013), almost five years after the completion of a DARPA‐sponsored project called Actionable Capability for Social and Economic Systems (ACSES), which focused on the feasibility of creating a computational social science laboratory. The effort was, at the time, one of the most comprehensive efforts to use agent‐based modeling for computation‐based quantitative research in social sciences. The contextual framework was the battle for winning the hearts and minds of the population at the height of the ongoing war in Afghanistan. The book outlined the computational platform that was developed during the project and the way that end users were supposed to use it to analyze problems of interest.

However, the book did not address fully some of the interesting issues noticed throughout the project – issues that were simply not within the project's scope – including the complexity of verification and validation of agent‐based models, the elusiveness of self‐organization ...