Explaining the survival of guerrilla organizations: The pros and cons of virtual experiments

Experiments are a most reliable way of identifying the causal regularities governing certain social settings. Unfortunately there are many social situations and configurations that are very unpleasant for the actors involved and therefore do not permit experimentation in the traditional sense. Civil wars, for example, are recurring phenomena whose internal laws are of great interest to the social scientist; yet at the same time they cannot and should not be converted into experimental settings. In order to overcome this limitation of the experimental approach and to extend its applicability I want to elucidate the option of conducting virtual experiments.

My presentation starts with a brief account of the typical social configuration of the guerrilla war. I describe the parties involved – the armed forces, the guerrilla organization, and the civilian population – as well as their mutual relationships. Subsequently I raise two questions concerning certain causal regularities of this social setting: To what extent is the survival of a guerrilla organization dependent on the support of the insurrection by the civilian population? Which degree of support is needed as a necessary condition for the survival of a guerrilla organization?

I seek answers to these questions by employing a multiagent simulation model of the guerrilla war as a virtual experimental setting. A systematic modification of the degree of support granted by the virtual civilian population enables me to measure the effect of this variable on the probability of survival of the virtual guerrilla units. The results thus obtained need to be discussed afterwards, particularly because of the high degree of abstraction of the simulation model. I conclude my presentation with some remarks about the advantages and disadvantages of virtual experiments in general.