Paper 1: "Varieties of Platform Governance"

In this paper, I conceptualize and classify different varieties of platform governance. I gather data on the institutions, policies, and norms of social media governance regimes and introduce the Varieties of Platform Governance (VPG) dataset. The first iteration of the VPG covers seven platforms from 2006 - 2020, and provides both disaggregated and aggregated measures of three different aspects of the internal politics of platform governmening bodies: content moderation, community autonomy, and discourse architecture.

Paper 2: "Community Sorting Across Platforms"

Do communities seek out favorable platform governance? In this paper I build an agent-based model that simulates political communities---both mainstream and extremist---sorting across different platform types. Based on a modification of the foundational Tiebout sorting model, I find two dynamics driving community radicalization: a combination of consolidation around a few massive platforms and the domination of algorithmic recommendation systems; and extremist communities compeiting to "raid" and recruit from mainstream platforms.

Paper 3: "Community Competition and Political Extremism"

This paper tests the theory that competition between political communities for public goods like attention, engagement, and audience access on social media platforms drives radicalization. To test this, I build a community detection pipeline combining a novel method of overlapping snowball sampling and an instruction-tuned large language model to map and classify closely related political communities. Focusing on the Qanon conspiracy and neo-sexist reactionary movements on Twitter, TikTok, Reddit, and YouTube, I find evidence that political movements with more overlapping communities in competition for online public goods harbor more extreme ideological commitments and political goals.