CO16: Digital Tools of Repression and Contestation – Part 1

Early research on the internet in authoritarian regimes highlighted its potential for protest coordination. More recently, scholars have turned to how governments use digital tools for repression—through surveillance, censorship, and connectivity disruptions. This paper examines the political economy of digital repression in the Islamic Republic of Iran. After multiple protest waves, the regime has grown increasingly reactive to online dissent. Following the death of Mahsa Amini in police custody in September 2023, mass protests erupted across all 31 provinces. In response, the government deployed internet slowdowns and shutdowns to suppress unrest. However, such measures carry economic costs and risk public backlash, especially as internet access has become vital to everyday life and commerce. We argue that the regime targets severe internet disruptions on days and in locations with high protest intensity, but avoids full shutdowns in economically important areas. In high e-commerce regions, disruptions tend to be partial and occur outside business hours. We test this argument using daily, province-level data on internet outages in Iran between September 2023 and March 2024.

Governments around the world often restrict access to information, particularly during times of political contention. While political scientists have extensively researched the effects of increased access to internet services, especially social media, on protest behavior and have recently turned their attention to the effects of online censorship, less is known about how circumventing temporary disruptions in online access affects such behavior. We systematically examine the effects of bypassing online censorship on offline protest behavior, leveraging novel data on Virtual Private Network (VPN) usage across all nationwide elections and coups d’état of African autocracies between 2017 and 2021. Using two-way fixed effects (TWFE) regressions, we analyze changes in protest behavior following censorship and its circumvention. Our findings reveal that censorship alone does not significantly affect protest behavior, but when individuals circumvent these restrictions, the likelihood of protests increases significantly. This paper highlights the critical role of circumvention of online censorship in shaping offline protests, providing a comprehensive evaluation of its effects across various model specifications and robustness checks, including instrumental variable techniques.

Research on digitally mediated protests shows that social media can catalyze collective action. A parallel literature on digital repression documents how states now use the same platforms to blunt dissent, though the micro-level mechanisms through which such repression operates remain understudied. Among the strategies states employ, an increasingly common one is the deployment of “cyber troopers”—paid or coordinated online agents who manipulate public opinion and promote disinformation while posing as ordinary citizens. Drawing on social psychology and social movement theory, this study examines how cyber trooper activity on social media fosters interpersonal distrust among online activists, undermining the trust-based mechanisms that sustain collective mobilization and thereby contributing to protest demobilization. The analysis relies on 14 million tweets from Iran’s “Woman, Life, Freedom” movement. Using a large language model, the study identifies accusatory tweets—users labeling each other as cyber troopers—as markers of social distrust on social media. The findings show that the distrust generated by cyber trooper activity is associated with the movement’s demobilization, supporting the theoretical view that digital repression operates through the erosion of social trust.

This article employs matched wake analysis to examine disparate police responses to protestor partisanship in France, Germany, and the United Kingdom. Using the Armed Conflict Location and Event Dataset (ACLED), a comprehensive observational protest database with event descriptions and political actor identification, I develop a novel multinational dataset of direct action group policy positions. Utilising a supervised machine learning technique, I classify protesters as "left", "centre", "right", and "unknown", enabling a nuanced analysis of ideological variations. By algorithmically identifying ideologically similar protests within close temporal and geographic proximity, I create matched protest pairs to isolate how police responses evolve over time. The analysis spans 2018-2025, examining both immediate aftermath effects (1-7 days) and long-term patterns up to 30 days following significant events. The research assesses whether there is an increasing police presence and/or escalatory police response associated with protestor partisanship across these three countries. Drawing on theoretical insights about institutional path dependencies, the study explores how initial instances of over-policing might create systematic intensifications or diminutions of police response to specific political movements over time.

Since early 2025, calls for Ukraine to cede territory to Russia have gained renewed international traction, particularly following Donald Trump’s victory in the U.S. presidential elections. At the same time, recent surveys suggest a modest shift in Ukrainian public opinion towards greater willingness to accept territorial concessions, which coincides with an escalation in Russia’s campaign of airstrikes on civilian infrastructure, including the widespread use of ballistic missiles and Iranian-made Shahed drones against civilian targets. Yet, a systematic examination of the effectiveness of Russia’s strategy of civilian victimization has been absent in scholarship on the war. To test for this effect, I combine fourteen rounds of nationally representative survey data on wartime attitudes with geo-referenced locations of reported Russian airstrikes on civilian infrastructure. Leveraging a quasi-experimental design that compares respondents exposed to an airstrike shortly before versus after their interview date, I find a tentative negative effect of civilian victimization on support for territorial concessions.