Donovan, J. (2025). EXPRESS: A short history of misinformation-at-scale and efforts to mitigate it. Journal of Public Policy & Marketing. https://doi.org/10.1177/07439156251384249
Summary
Joan Donovan offers a historically grounded account of how “misinformation-at-scale” was socially constructed as a public problem between 2016 and 2021, and how the major social media platforms responded — first with neglect, then with an expanding repertoire of moderation tools, and finally with a politically driven retreat. Her central argument is that misinformation-at-scale is not an aberration but a structural feature of engagement-based platform business models: falsehoods become consequential only when amplified by densely networked actors (politicians, journalists, celebrities), and platforms’ advertising logics reward exactly that amplification. Rather than tackle veracity directly, platforms developed behavioral workarounds (most notably “coordinated inauthentic behavior”) that preserved revenue while gesturing at governance. Donovan argues that the resulting privatization of speech governance leaves citizens without what she calls TALK — timely, accurate, local knowledge — and that addressing this gap requires a whole-of-society response.
Key Contributions
- Defines and operationalizes misinformation-at-scale as a sociotechnical event distinct from individual rumor, contingent on networked amplification by high-reach actors.
- Introduces TALK (timely, accurate, local knowledge) as the public good platforms systematically fail to provide.
- Presents a taxonomy of fourteen content moderation methods, mapped to the actors (individuals, groups, platforms, governments) empowered to use them.
- Provides a periodization of platform moderation from non-intervention (pre-2016) through pandemic- and election-era expansion to post-2021 retrenchment.
- Argues for whole-of-society mitigation: advertiser pressure, civil society coalitions, journalistic strategic silence, and pre-bunking.
Methods
Comparative historical analysis of platform policies (2016–2024), combined with ethnographic case studies of key misinformation events (Charlottesville, COVID-19, the 2020 election, January 6th, the Hunter Biden laptop story). Donovan also draws on participant-observation with civil society groups (Disinformation Defense League, Election Integrity Partnership), synthesis of whistleblower disclosures and congressional hearings, and conceptual/typological work culminating in the moderation-methods taxonomy.
Findings
- Nine of fourteen catalogued moderation methods target amplification rather than removal, revealing platforms’ preference for throttling scale over adjudicating truth.
- Facebook’s CIB framework repurposed spam/fraud tooling so it could act on behavior without ruling on veracity.
- Labeling Trump’s tweets reduced on-platform engagement but increased cross-platform spread — what Donovan calls “platform filtering.”
- Platforms moderated medical misinformation more aggressively than political misinformation, using information displacement (e.g., COVID information centers) to surface authoritative sources.
- After 2021, trust and safety capacity collapsed: Meta laid off 11,000 workers, Musk gutted X’s civic integrity and ethics teams, Google cut 6% of staff in 2023.
- Bad actors adapt with coded language (“va33ine,” “pizza” for Pfizer), backup accounts, and migration to less moderated platforms.
- The January 2021 deplatforming of a sitting U.S. president marked an inversion of political power between platform firms and the state.
Connections
This paper provides a historical and conceptual scaffolding that complements empirical work on coordinated amplification and platform interventions, including Giglietto2019-e9be81c1, Giglietto2022-b30e8b4e, and Starbird2025-jj on the Election Integrity Partnership context Donovan herself worked within. Its diagnosis of engagement-driven amplification and the retreat of trust-and-safety regimes resonates with Bak-Coleman2026-mk, Budak2024-ef, and Freelon2024-sc, while its emphasis on pre-bunking and civil-society responses links to van-der-Linden2026-jt and Lewandowsky2026-ob. The argument that platform governance has become a privatized form of speech regulation also connects to Rieder2026-pp and Marwick2025-ov.