Using Markov Decision theory to manipulate simple graph-based models

We investigate the creation and destruction of opinion bubbles in small ($N=16$) social networks. The model is constructed through a graph interpretation of the Ising model. In this model, parallels are assumed between the interaction between social individuals and information distribution and the interaction between neighbouring magnetic particles and magnetic fields. By formulating the model as a Markov Decision Process, policies for creating and destroying bubbles through field manipulation can be constructed and compared to a policy representing targeted information exposure. It is concluded that with this model, there is no significant amount of bubble creation when the individuals' opinions are perpetuated by the outside field.