Agent-based modeling for political culture theory application

[ECPR2020 abstract] Section 63 of the ECPR 2020 conference aims at considering polity (state structure, -operation, and -dynamics) from the perspectives of political culture theory (focusing on attitudes, beliefs and sentiments that give order and meaning to polities) and associated research methodologies.

Culture exists in networks (of top-down, peer-to-peer and bottom-up connections) in which individuals/citizens and institutions/states participate. When we want to explain emergent phenomena like periods of political stability and instability and the tipping points in-between we need not only understand how citizens’ political cognitions, values, beliefs and attitudes contribute to the operation of the state, we also need to know what the state is aiming for and how it deploys cult-carrying stories to realize them. Political cultures are volatile and multi-faceted, networked things, full of feedback loops. They are complex adaptive social systems.

The social sciences have (roughly speaking) evolved by acquiring knowledge via qualitative methods that, considered non-rigorous, have been appended with statistical methods that, considering their limited support for formal modeling have been appended with mathematical modeling that, considering its bias towards using computation-friendly but unrealistic assumptions has been appended with the algorithmic methods of agent-based simulation.

The number of available methods has encouraged specialization. Repositories of social-scientific research results have become kaleidoscopic collections of knowledge fragments that required serious investments in time and money to achieve. Moreover, these results are often situated and have unlike the natural sciences poor predictive value. (We do know the chemistry of how to generate CO2 but do not know the moments when and the places where what coalitions of polities will eventually realize to domesticate the production of CO2 — if at all.)

We expect that at least half of the knowledge required for deliberate political decision making can be harvested from academic natural- and social-science repositories while the other half has to be acquired ad hoc because situations require local knowledge that often is contingent. For research projects and individuals like voting citizens and policymaking representatives this raises a serious question:

Are there cost-effective quality methods and tools to combine diverse research results in order to quickly estimate the influence of a situated political decision on the stability of the polity involved?

This question has been bothering us in the last decade when approaching our own diverse research projects on policy decisions considering, e.g., parole in the Netherlands, file sharing, EU privacy in the PRC, failing ICT in the Dutch judiciary. Our chosen approach tallies with mainstream complexity theory that supports partly stochastic and partly rule-based simulations of networked individual and collective behaviors.

Based on our experiences we created a framework for situated political research via agent based models that combine rules and stochastic distribution-formation mechanisms. We call it Epiframer. We (aim to) present it with its pros and cons in panel #5, illustrating its operation with how it helps to come to grips with the insurgency timeline that records the USA-PRC trade-war dynamics from April 2017 until April 2020.