Effect of media on elections in an Ising-like voter mode

Abstract

"In this project we develop a bottom-up model to study the influence of media on election dynamics in a two-party system (say, the US). We start from the classic Potts model which is a generalization of the Ising model, usually used to describe interacting spins on the lattice. In our case, the nodes represent voters which can have 3 states: Democratic voter, Republican voter, or non-voter. Then, voters can interact to influence each other's opinion, depending on transition probabilities and individual affiliations to parties. Some nodes represent media sources. These are highly-connected and influential nodes, which are randomly located in the network and have the role of spreading external influence (e.g. information on the state of the economy) throughout the population. By varying the number of these media in the network, we observed a phase transition between the configuration in which the population is completely uninformed and ordered and one continuously influenced by the many sources, thus extremely disordered. Finally, several experiments were done by adding internal feedbacks between the media and people, to assess under what conditions deep partians divisions, hence polarisation, can be observed. Therefore, we witnessed how important it is to maintain a certain variety in the media's opinion to prevent the formation of the so-called "echo chambers" and therefore of polarisation as an emergent behavior."