Connective action and digital repression during China’s COVID-19 protests: a computational analysis of multilingual coordinated activity on Twitter

Summary

This paper examines coordinated activity on Twitter during the November 2022 COVID-19 (“A4” / White Paper) protests in China, treating the platform as a dual-use infrastructure that simultaneously enabled grassroots protest mobilization and state-aligned digital repression. Using a coordination detection algorithm on multilingual tweets, the authors identify a large set of accounts engaged in coordinated communication and analyze the thematic, temporal, and linguistic signatures of competing narratives. The work bridges connective action theory with research on authoritarian digital repression, arguing that social media around contentious events in China are shaped by overlapping, oppositional coordinated campaigns rather than a single dominant logic.

Key Contributions

• Empirical mapping of the scale and structure of coordinated communication during a major protest event in an authoritarian context.
• Application of coordination detection methods to multilingual (Chinese and other languages) Twitter data, extending a largely English-centric literature.
• Theoretical bridging of connective action and digital repression by framing protest and state-aligned coordination as competing forms on the same platform.
• A case study showing that “coordinated inauthentic behavior” frameworks need to accommodate genuine protest coordination alongside repressive operations.

Methods

• Collection of Twitter data linked to the 2022 China COVID-19 protests.
• A coordination detection algorithm to identify accounts behaving in unusually synchronized ways.
• Multilingual content analysis of identified clusters, plus thematic and temporal pattern analysis to differentiate protest-mobilizing from repression-aligned activity.

Findings

• 13,557 Twitter accounts were classified as engaged in coordinated activity around the protests.
• Coordinated communication clustered into distinguishable thematic, temporal, and linguistic patterns, consistent with competing protest and state-aligned narratives.
• The volume and structure of coordination suggest social media served as contested infrastructure during the protest wave, not merely a mobilization channel.

Connections

This work sits alongside other recent efforts to detect and characterize coordinated behavior on platforms, particularly those treating coordination as a methodological problem solvable via temporal/behavioral signatures — see Luceri2025-tr, Minici2024-tf, and Mannocci2025-ig. Its focus on state-aligned and protest-aligned coordination in an authoritarian setting connects it to studies of geopolitically motivated influence operations such as Kuznetsova2025-nu and Zhao2025-ny, while its concern with multilingual and protest-context coordination resonates with Graham2025-gp and Pante2025-pq.