Marino, F. G. G. (2023). The power of Alternative Influence Networks (AIN) for spreading Covid-19 problematic information on Facebook during a year of pandemic. https://doi.org/10.1445/106772
Summary
This paper maps how Alternative Influence Networks (AIN) circulated problematic Covid-19 information on Italian Facebook during the pandemic’s first year (March 2020–March 2021). Working from a dataset of 175,228 posts by previously identified covid-skeptic accounts, the authors extract mentioned public figures, cluster them into five categories, and analyse engagement patterns and content. They argue that an “Intellectual Dark Web” (IDW) subset of influencers produces disproportionately shareable content, that legacy media outlets and journalists remain central conduits for these voices, and that remediation — the recycling of expert statements from TV and newspapers — is a key mechanism by which AIN actors amplify covid-skeptic narratives while evading fact-checking.
Key Contributions
- Empirical mapping of the Italian AIN during Covid-19, identifying an IDW subnetwork with a distinctive share-heavy engagement signature.
- Analytical use of a comments/shares ratio metric to distinguish hyperpartisan amplification from contested debate.
- Extension of remediation theory to alternative media, showing how AIN actors weaponise legacy-media-produced expert statements.
- Contribution to inter-media agenda-setting research by documenting co-production of visibility across AIN, alternative outlets, and legacy media.
- A replicable mixed-methods workflow combining coordinated-behaviour detection, named-entity extraction, clustering, engagement metrics, and stratified content analysis.
Methods
The authors collected 175,228 CrowdTangle posts from a pre-identified list of Italian covid-skeptic accounts (detected via coordinated link sharing). They extracted the top 2,000 capitalized proper nouns from post fields, manually classified 254 last names into five (non-mutually-exclusive) figure categories, filtered out those in apical public roles, and applied k-means clustering to the remaining 79 figures, producing five clusters. Engagement was analysed via mean interactions and a comments/shares ratio (−1 to 1). A stratified random sample of 150 posts drawn from the top 10% most-engaging content, stratified by cluster and ratio category, was subjected to deductive content analysis.
Findings
- Five clusters emerged: Politicians, Other Public Figures, Journalists, Doctors/Scientists/Experts, and IDW Influencers.
- IDW Influencers had the lowest comments/shares ratio (−0.28), indicating pure amplification; Doctors/Scientists/Experts also skewed to shares (−0.21).
- Journalist-related posts had the highest ratio (+0.16), indicating contested/commented framings; Politicians produced the most posts (9,879) with moderate engagement.
- Despite low volume, IDW Influencers (1,649 posts) and Other Public Figures (1,958 posts) achieved the highest mean interactions (1,030 and 983).
- Top amplifying pages were dominated by mainstream-affiliated journalists (Nicola Porro, Gianluigi Paragone) and right-wing newspapers (La Verità, Libero); covid-skeptic micro-celebrities (Montanari, Messora) appeared further down.
- Recurring themes: immigration and anti-government framings among politicians; judicial scandals among Other Public Figures; covid-skepticism citing figures like Zangrillo or Tarro in the science cluster; and conspiracy narratives (microchips, global tyranny) in the IDW cluster, often sourced from TV talk-show clips.
Connections
This paper extends the coordinated link sharing detection agenda developed by the same group in Giglietto2019-e9be81c1 and Giglietto2022-0e951ac5, applying it downstream to trace which figures those coordinated networks amplify. Its focus on the porous boundary between AIN micro-celebrities and legacy media resonates with work on alternative news ecosystems and hyperpartisan amplification such as Rossi2023-847d5a9f and Ghezzi2023-8bebc91f, and its Covid-19 infodemic framing connects to broader studies of platform-mediated health misinformation like DeVerna2025-dl and Bruns2026-yv.