Key findings
Why people share false information:
-
Most sharing is not intentional deception. In experimental studies, when people share misinformation, only ~33% of instances are due to genuine confusion (believing false content is true). The majority (~51%) occurs because sharers simply don't consider accuracy—they're focused on engagement, personal image, and social dynamics. Only a minority deliberately share known falsehoods.
-
Accuracy nudges work. A simple direct message reminding people about accuracy on Twitter shifted attention toward truth evaluation and improved the quality of content people shared. The effect is modest but scalable.
-
Thinking style predicts susceptibility. People low in reflective thinking, high in overconfidence, or dogmatic in reasoning show greater susceptibility to misinformation. Active open-mindedness—questioning your beliefs—is protective. This varies across the political spectrum.
How platforms amplify misinformation:
-
Algorithms reward engagement, not accuracy. Social media feeds are optimized for clicks and dwell-time, not truth. This environment doesn't make brains "melt" into irrationality—it simply distracts people from considering accuracy before sharing.
-
Affordances enable organized campaigns. Platforms provide tools for rapid amplification (retweets, shares, follows). Both legitimate information-seekers and coordinated disinformation actors use these affordances. The infrastructure is neutral; the difference lies in intent and scale.
-
Information cascades cross communities. DiResta's study of the Plandemic video showed how misinformation seeded in anti-vaccine echo chambers traversed into holistic health, then Q-Anon communities, then mainstream pro-Trump groups, and finally into apolitical communities (e.g., dog-lover groups) where it was discussed through a local lens.
Group dynamics and belief formation:
-
Identity and belonging drive sharing. People who share misinformation often see themselves as "warriors for truth," motivated by altruism and desire to inform their communities. The salience of group membership, priming (repeated messaging within communities), and collective sense-making override individual fact-checking.
-
Consensus happens in public now. Unlike expert consensus-building in private institutions, online communities build consensus transparently. When institutions (health authorities, media) communicate uncertainly or incorrectly, it creates voids for misinformation and undermines trust when corrections come.
Current limitations of interventions
-
Fact-checking doesn't scale. You cannot label all misinformation faster than it spreads. Once labels become widespread, absence of a label signals truth, backfiring.
-
Fact-checks can be dismissed. When fact-checkers are seen as partisan (e.g., "mainstream media," "fact-checkers are funded by big pharma"), their verdicts are rejected, especially by those already skeptical.
-
Backfire effects are rare but possible. Repeating misinformation while debunking does not reliably cause belief in the false claim. However, defensive reasoning and identity-protective cognition mean corrections are often ineffective for ideologically committed audiences.
Solutions require systemic coordination
The panelists propose a Swiss cheese approach:
-
Education: Long-term media literacy and critical thinking taught to children, though this faces politicization ("indoctrination" accusations) and generational timescales.
-
Accuracy incentives: Reminders and nudges that shift focus to truth before sharing. Crowdsourced fact-checking (asking crowds to rate accuracy) can be both scalable and effective.
-
Platform design: Changes to algorithms, ranking, and recommendations to not solely optimize engagement. Alternative approaches (Wikipedia-style version control transparency, media publishing with living updates instead of static corrections).
-
Institutional communication: Experts and institutions should communicate transparently about uncertainty and evolving understanding, rather than projecting false certainty that later gets debunked.
-
Policy and regulation: Government oversight of platform affordances and algorithm behavior; rules tailored to specific harms (election misinformation, health misinformation) rather than generic content moderation.
Notes
The dialogue between DiResta (systems/networks perspective) and Pennycook (individual psychology perspective) is instructive. Pennycook emphasizes that most people don't maliciously spread falsehoods—they're inattentive and distracted. DiResta's addition is crucial: even if individual intent is benign, the platforms themselves provide scalable infrastructure for organized actors who do intend to manipulate. Neither explanation alone is sufficient; both psychology and systems design matter.
The panelists' emphasis on multi-layered solutions and rejection of "silver bullets" reflects the scientific consensus on misinformation interventions: no single approach (not fact-checking alone, not media literacy alone, not platform design alone) scales to the problem. The question then becomes how to coordinate these approaches—a challenge as much institutional and political as technical.