It Takes a Village to Moderate a Platform: An Exploration of How Networked Platform Governance in the Digital Services Act affects Freedom of Expression

Abstract

This dissertation examines how the Digital Services Act (DSA) creates a system of networked platform governance in content moderation, and how that system affects freedom of expression online. The DSA is positioned as a regulatory instrument that fills the gaps left by the e-Commerce Directive. It further harmonises platform regulation in the EU and is intended to combat manifesting social harms. The central question is "who determines what we can say online, and how is that determined", with a focus on the DSA's procedural design rather than on the substantive limits of permissible speech. The dissertation uses legal-doctrinal, empirical, and computational methods to assess that question. The second chapter examines the DSA transparency database, a novel transparency tool that aggregates content moderation decision. The empirically examines what the content moderation landscape looks like, and how social media platforms use the transparency database. Findings indicate that platforms prefer moderating on the basis of their terms and conditions. Platforms have vary in the degree of transparency their notices create. The third chapter assesses the hosting liability exemption under article 6 DSA. It argues that, despite largely unchanged statutory text from the ECD, the DSA's due diligence expectations support a narrower interpretation that excludes platforms acting in bad faith or without diligence from the liability exemption. The fourth chapter assesses notice-and-action mechanisms employed by AliExpress, TikTok, Twitch, and X. It concludes that implementation varies widely. A central paradox is that user-friendly notice designs conflict with the legal precision required for a notice to trigger the earlier examined liability exemption. The fifth chapter analyses trusted flaggers under article 22 DSA. Trusted flaggers are DSC-certified entities whose notices platforms must prioritise, improving enforcement against illegal content but raising over-removal, fundamental rights, and rule of law concerns. Despite transparency gains, users have limited means of holding trusted flaggers accountable. The sixth chapter evaluates the Digital Services Coordinators as national enforcers of the DSA. It finds that decentralised enforcement risks divergent speech protection across Member States, as DSC powers and practices shape the content moderation environment in each jurisdiction. The seventh chapter evaluates "jawboning", a term coined to describe informal pressure by state actors leading to policy change in private actors. This can result in unconstitutional limitations of speech, especially when no means of accountability are available to address those informal pressures. The eighth chapter offers a synthesis. The thesis builds a critique of the DSA's multi-actor design, and its capacity to protect freedom of expression at scale.