Predicting the Evolution of Conflicts using Fuzzy Recurrent Games

Game Theory is premised on a modeling a conflictual interaction between two known parties where tactics are well-defined and payoffs can be easily measured, like a game of chess. This paper offers an extension of the Game Theory model incorporating ‘fuzzy variables’ representing possible situations and results, and also the discrete logistic curve to effectively capture the chaotic aspects of population dynamics impacting the decision- making in the composite societies (players). These are the typical actors in multiplayer political-economic games. This would reflect more closely the real-world behaviors and provide greater predictive accuracy. A trained AI-enhanced model, using regression data collected from previous, resolved games as well as previous ‘moves’ in the current game could prove helpful in predicting likely next ‘moves’ in the continuing game. This will allow us to model the evolution of strategies over the course of time in real conflicts and help mitigate exacerbation of extreme violence by simulating ways to reduce the driving force – the degree of deviationbetween what is acceptable and unacceptable to the players involved.