Homophily and Insularity Dynamics in an Echo Chambers: Computational Models for the Study of the Conspiracy Subculture in Facebook

The research focuses on developing a computational model (agent-based) to describe and analyse the structure and evolution of a conspiracy bubble within Facebook. The methodological framework relies on a hypothesis, previously validated in other digital contexts, asserting that user groups interested in conspiracy topics form not isolated cliques but rather constitute a subculture characterised by a system of “subjective understanding” (Tucker, W. T. (1965). Max Weber's verstehen. Sociol Q, 6(2), 157–165) and specific interpretative categories (Russo & Cecconi, Russo, V., Cecconi, F. (2023). Collettivi digitali e cultura della disinformazione. Analisi di una bolla cospirazionista durante la pandemia di Covid-19. Sociologia Italiana, (22).; Barkun, Barkun, A culture of conspiracy: Apocalyptic visions in contemporary America, University of California Press, 2013; Grusauskaite et al., Grusauskaite et al., New Media & Society, 2023). In accordance with this hypothesis, the research design incorporates four types of entities: the conspiracy articles generated in the digital space, the websites where they come from, the public pages and groups where they are shared, and the users engaging with this content. The methodological process unfolds in two phases: (1) descriptive, data mining and analysis of the content and structure of a conspiracy bubble, along with the construction of a relational dataset; (2) experimental, development of the agent-based model, creation of scenarios, model validation, simulation, and subsequent data analysis. The research findings reveal a bubble comprising a singular component characterised by hub nodes functioning as cultural catalysts and specific interpretative categories shared among users. Simulation results show that the network fragments in response to changes in two article properties acting as ‘glue’: political orientation and the incitement of fear and hate.