Titolo della tesi: Repression Without Borders: How Digital Technologies Reshape Political Violence at Home and Abroad
This dissertation investigates the transformation of political repression in the digital age, focusing on how states deploy information and communication technologies (ICTs) to monitor, control, and suppress dissent both domestically and transnationally. We integrate three theoretical strands to conceptualise contemporary repression. Drawing on practice theory, we move beyond static regime typologies to examine repression as embedded in everyday governance across democratic, hybrid, and authoritarian contexts. To account for the transnational extension of repression, we incorporate diaspora mobilisation theory, which explains the political agency of transnational actors and the conditions under which they become targets of state coercion. Finally, rationalist models of repression inform our understanding of state behaviour as shaped by cost-benefit calculations within the dissent–repression nexus. Empirically, the dissertation addresses two interrelated research puzzles. First, we analyse the domestic diffusion of digital repression using cross-national, time-series data from the Digital Society Project (DSP), the Digital Repression Index (DRI), and the Digital Repression Capacity Index (DRCI). Results demonstrate that while digital repression is more prevalent in authoritarian settings, it is increasingly adopted in democratic and hybrid regimes, often in tandem with analog forms of coercion. Second, we develop ReprLM (i.e., Repression Language Model), a Large Language Model-based classification pipeline that detects and categorises political repression events in unstructured textual data. By combining a modular Chain-of-Thought (CoT) prompting architecture with embedding-based exemplar retrieval, ReprLM enables the construction of the first fine-grained, scalable dataset capturing the actors, targets, tactics, and spatial geographies of transnational repression (TR) and digital transnational repression (DTR). This dissertation makes three primary contributions: (i) it advances theory by reconceptualising repression as a routinised, cross-border practice and by offering a taxonomy spanning domain (i.e., physical and informational), mode (i.e., coercion and channelling), and medium (i.e., analog and digital); (ii) it introduces a novel LLM-based framework that enables the large-scale identification, attribution, and hierarchical coding of repression events; resulting in (iii) the creation of a global, fine-grained dataset of 2,414 cross-border repression events, capturing the actors, targets, tactics, and geographies of TR and DTR. Together, these contributions advance comparative politics, security studies, and computational social science by examining how states reconfigure political violence through digital technologies in both domestic and transnational contexts.