Health Information and Online Publics
From individual exposure to ecosystem dynamics
The papers gathered here trace a progression in how we think about problematic health content online: from the individual encounter, through coordinated actors, to ecosystem-level imbalances. At the individual end, Lyons2026-ca uses web-tracking to show that exposure to low-credibility health content in the US is rare in the aggregate but heavily concentrated among older adults, predisposed conspiracists, and consumers of dubious political news — with referrals coming overwhelmingly from other low-credibility sites rather than from search or social platforms. Adam2026-tz complements this with German/Swiss panel-plus-tracking data, showing that conspiracy beliefs about COVID-19 arose from a “marriage” of media exposure and predispositions: populists selectively engaged with alternative media and counter-argued against mainstream debunking. Together, these studies push back against panic narratives of mass exposure while specifying who is vulnerable and through which pathways.
Coordinated infrastructures and asymmetric polarisation
A second cluster moves attention from audiences to coordinated supply-side actors. Giglietto2022-0e951ac5 develops the Coordinated Link Sharing Behaviour (CLSB) framework as a content-agnostic detection signal, mapping Italian covid-skeptic networks that grew out of existing political clusters and adapted tactically — first-comment links, link laundering, image-macros — to evade enforcement. Marino2023-9137f448 extends this lineage into Alternative Influence Networks, showing how an “Intellectual Dark Web” of fringe figures was amplified largely through remediation of mainstream TV and newspaper appearances, with right-leaning legacy outlets and journalists as central nodes. Song2025-yh generalises CLSB comparatively, finding that coordination is ideologically agnostic: pro-vaccine networks in the US were larger and institutionally grounded (CDC, NHS), while UK anti-vaccine networks were denser and ideologically driven. Efstratiou2026-ij sharpens the picture for science communication on Twitter, identifying a coordinated retweet network that is 96% contrarian and disproportionately amplifies a small set of credentialed anti-consensus experts — coordination not powered by bots but by sustained human cooperation. A clear throughline emerges: coordination is structural, often legal, frequently parasitic on mainstream gatekeepers, and asymmetrically tilted toward contrarian voices.
Mainstream media as conduit, not firewall
Several papers complicate the conventional split between “good” mainstream and “bad” alternative media. Adam2026-tz finds that even quality mainstream outlets circulated substantial conspiracy-supporting content. Ghezzi2023-8bebc91f documents partisan sorting in British newspapers’ citation of pro- and anti-precautionary COVID scientists, showing that politicisation of science occurs through legitimate sourcing rather than fringe smuggling. Marino2023-9137f448 and Giglietto2022-0e951ac5 highlight how covid-skeptic actors weaponise mainstream broadcasts and reputable URLs to evade fact-checking. Efstratiou2026-ij empirically traces that news coverage of preprints tends to follow superspreader activity on Twitter, with low-trust outlets aligning with contrarian amplifiers. The collective implication is that mainstream and alternative ecosystems are entangled, with the former often providing the raw material that the latter recirculates.
Protest publics, platform migration, and the political ecology of vaccines
Rothut2026-or zooms out to the meso level, theorising “protest-facilitated mainstreaming”: the German Querdenken movement on Telegram bridged moderate pandemic dissent to far-right and QAnon-adjacent communities, with shared anti-elitism rather than explicit ideology as the connective tissue. Song2025-yh shows analogous national-political specificity: UK anti-vaccine discourse focuses on safety, US anti-vaccine discourse on freedom — vaccine narratives are inseparable from each country’s political grammar. These papers reframe health misinformation as embedded in broader struggles over political legitimacy and protest opportunity structures, not as a discrete domain.
Quantifying consequences: from exposure to outcomes
A final strand pushes toward causal and structural quantification. Bollenbacher2026-vz combines a SIRVA compartmental model with geolocated antivaccine tweets to estimate that Twitter antivaccine exposure caused roughly 14,000 vaccine refusals, ~545 cases, and ~8 deaths in the US between February and August 2021 — providing a rare causal bridge from online speech to epidemic outcomes. Scalco2026-bd proposes “information voids” as measurable supply-demand imbalances, showing that voids around vaccine events persist longer than overabundances and correlate with reduced credible content and elevated misinformation. Read together, these papers move the field past correlational and descriptive work toward mechanistic, outcome-anchored accounts.
Cross-cutting tensions and open questions
Several productive tensions run across the collection. First, exposure is rare (Lyons2026-ca) yet consequential (Bollenbacher2026-vz) — concentration on vulnerable subpopulations may matter more than aggregate prevalence. Second, coordination is widely documented (Giglietto2022-0e951ac5, Song2025-yh, Efstratiou2026-ij) but its harm depends on actor alignment, raising normative questions about content-agnostic detection. Third, the mainstream/alternative boundary is increasingly porous: legitimate scientific disagreement (Ghezzi2023-8bebc91f), remediated experts (Marino2023-9137f448), and protest-mediated bridging (Rothut2026-or) all blur the line that fact-checking regimes presume. Finally, the topic description gestures toward India and Nigeria, but the filed papers are overwhelmingly European and North American — a notable gap, given that coordinated health misinformation in the Global South likely operates through distinct platform ecologies (WhatsApp, regional broadcast media) that none of these studies directly addresses.