Modeling Bot Activity in the Online Political Communication

This paper presents a new agent-based model of political communication with the participation of bots - automated social media accounts. It aims to recreate politically significant informational interactions between Twitter users in a controlled environment. Our model incorporates a variety of options and mechanisms including limited attention, heterogeneity in account activity and performance characteristics, limited lifespan of tweets, and preferential attachment network structure. The features of political communication are reflected in the model through the ideological stances of human users and bots, varying tolerance levels, and leadership. We make a detailed description of the algorithms and test them using a computational experiment. The latter expectedly shows that the bots' reach increases in their proportion. More interestingly, it demonstrates substantial variation in the dynamics and levels of the reach of bots depending on the level of polarization and bot warfare.