How Polarization May Help Combat Misinformation: Community Notes and Crowd-Sourced Fact-Checking

Speaker: David Rand
Affiliation: MIT (Erwin H. Schell Professor of Management Science and Brain & Cognitive Sciences; Director of Applied Cooperation Team)
Year: 2026
Venue: UC Berkeley I School Distinguished Speaker Series
Video ID: d7U1ZGC25Dk

TL;DR

David Rand presents a counter-intuitive argument: while polarization is widely viewed as exacerbating misinformation, it may actually improve crowd-sourced fact-checking systems like Twitter's Birdwatch/Community Notes. Rather than a traditional technology approach (machine learning classifiers), platforms can scale fact-checking by harnessing crowd wisdom. Rand introduces a theoretical model showing that people are motivated to flag content by two competing drives: an accuracy motive (flag falsehoods) and a partisan motive (flag content that opposes their politics). In experiments, he finds that more polarized individuals flag false information more aggressively—including false claims aligned with their own politics—without substantially increasing false flagging of true information. This suggests that polarization, properly channeled, could be a feature rather than a bug in crowd-sourced content moderation.

Key mechanisms

The scalability problem: Platforms face a multi-layered challenge in combating misinformation. Machine learning classifiers struggle because: (1) claims exist on a spectrum of truth/falsehood, not binary; (2) misinformation evolves constantly, making training data non-stationary; and (3) the sheer volume of content—billions of posts daily—exceeds human fact-checking capacity. Professional fact-checkers like PolitiFact and Snopes do rigorous work but cannot scale. Thus the question: how do platforms identify false claims at scale and speed?

Wisdom of crowds approach: Rather than relying solely on expert fact-checkers or automated systems, platforms can harness the judgments of ordinary users. Twitter's Birdwatch (now Community Notes) uses this model: users flag misleading posts; the system aggregates these flags and uses them to inform moderation decisions or display context notes. This leverages billions of users as a decentralized fact-checking workforce, achieving scale that experts alone cannot.

Two competing motivations for flagging: Rand's central theoretical contribution is modeling what drives users to flag content: - Accuracy motive: People care about truth. They flag false information regardless of whether it aligns with their politics. - Partisan motive: People care about politics. They flag information that opposes their side's interests, regardless of whether it is true.

For any piece of content (true/false × aligned/opposed with one's politics), these two motives interact. A person with a strong accuracy motive flags all false claims. A person with a strong partisan motive flags all opposing claims. Someone with both motives may strategically avoid flagging false claims that support their side.

The polarization paradox: Conventional wisdom says more polarized people should be worse at crowd-sourced fact-checking because they prioritize partisan goals. Rand's experiments suggest the opposite. In a study where participants reviewed a mix of true and false headlines (aligned and opposed to their politics), more polarized individuals: - Flagged more false headlines overall, including false claims aligned with their politics. - Did not substantially increase false flagging of true information opposed to their politics.

This pattern suggests that polarized individuals have stronger accuracy motives—or at least that their partisan and accuracy motives both push toward flagging falsehoods. The result is that crowd-sourced systems benefit from including polarized participants.

Mechanisms underlying the effect: Rand discusses several possible mechanisms: 1. Polarized individuals are more politically engaged and thus more motivated to participate in fact-checking. 2. Polarized individuals have higher political knowledge, making them better calibrated in assessing factual accuracy. 3. Political engagement correlates with cognitive effort: polarized individuals may scrutinize claims more carefully. 4. The experimental setting (explicitly being asked to rate content for accuracy) may activate the accuracy motive over the partisan motive.

Challenges in open flagging systems: When anyone can flag any content (rather than using hired contractors), two problems emerge: - Coordinated attacks: Bad actors can launch systematic flag campaigns against disliked content, manipulating the system. - Public goods problem: Users donating time to flag content for a profit-seeking platform may under-provide, lacking incentive to contribute.

Rand's work explores how to structure crowd-sourced systems to mitigate these issues while leveraging the benefits polarization appears to bring.

Interventions discussed

Scaling beyond expert fact-checkers: Professional fact-checking is high-quality but slow and non-scalable. Crowd-sourced approaches, if well-designed, can achieve speed and coverage.

Leveraging political engagement: Rather than viewing political polarization as purely destructive, the research suggests that polarized, politically-engaged users may be valuable contributors to content moderation.

Wisdom of crowds: Aggregating many imperfect judgments produces better decisions than individual or expert judgments alone, provided diversity is preserved and systems guard against coordinated manipulation.

Exploring motivation structures: Future work should investigate how to incentivize accuracy-motivated flagging over partisan-motivated flagging, and whether heterogeneous groups (diverse polarization levels) perform better than homogeneous groups.

Connections

• Misinformation — defining and identifying false information at scale
• Fact-checking — professional and crowd-sourced approaches
• Community notes — crowd-sourced content context and labeling (Twitter/X)
• Content moderation — platform responses to harmful content
• Polarization — ideological divergence and affective divides
• Wisdom of crowds — aggregating collective judgment
• Social media and misinformation — moderation strategies and interventions
• Crowdsourcing — collaborative human-powered systems

Notes

This talk reframes polarization as a potential asset in fighting misinformation, rather than exclusively a liability. While polarization clearly has serious societal costs (increased out-party hostility, reduced cross-cutting exposure, gridlock), Rand's evidence suggests that political engagement and partisan motivation, when combined with accuracy concerns, may enhance crowd-sourced fact-checking.

A key insight: the framing and context matter enormously. When users are explicitly asked to rate accuracy, the accuracy motive becomes salient. In other contexts—e.g., social media feeds where political affect dominates—partisan motivation may dominate instead. The implication is that platform design (what questions are asked, what signals are highlighted) shapes whether polarization helps or hurts.

The work also highlights tensions between scalability and quality. Scaling beyond expert fact-checkers requires opening the system to non-experts, which introduces variance and risk of manipulation—but expertise alone cannot keep pace with content volume. Crowd-sourced systems accept lower per-judgment quality in exchange for scale; the question is whether aggregation recovers accuracy at the system level.

Finally, the public goods framing suggests that sustainable crowd-sourced fact-checking may require economic incentives or status/recognition rewards, not just intrinsic motivation—a design question platforms are actively exploring through systems like Birdwatch/Community Notes.