The Crowd is Watching: How Community Notes Redefines Online Truth

(Center and Knispel)

Introduction

In the fight against online misinformation, social media platforms have long relied on centralized fact-checking teams to flag false content. But that model is shifting. X (formerly Twitter) and now Meta are turning to users themselves through programs like Community Notes, crowdsourced systems that let the crowd decide what needs context.

At first glance, this looks like platforms handing responsibility to the people. But research suggests something more interesting is happening. Community Notes doesn't just flag facts. It changes how people behave, both the ones writing notes and the ones posting content.

Studies show that users who participate in Community Notes start posting higher-quality, less inflammatory content (Borwankar, Zheng, and Kannan 2022). When notes appear on posts, authors often delete their own tweets voluntarily. And when anonymity was introduced to protect fact-checkers from retaliation, they engaged more with controversial topics while keeping a neutral tone (Borwankar, Zheng, and Kannan 2024).

This article explores how Community Notes works, why the "bridging algorithm" matters, what the research shows about its effects, and what limitations still need fixing. Because understanding this shift isn't just about fact-checking, it's about how platforms shape behavior and who really holds the power.

How Community Notes Works

Community Notes lets users add context to misleading posts. But a note only publishes if it gets positive ratings from people who typically disagree with each other. This is the "bridging algorithm."

If a note about election integrity is rated helpful by both progressive and conservative users, the algorithm recognizes common ground. Notes that only appeal to one side never see the light of day. This forces contributors to write notes useful to everyone, not just their own echo chamber.

The Consensus Engine

The bridging algorithm sounds elegant, but less than nine percent of submitted notes actually get published. The consensus bar is high.

Defining "diverse viewpoints" is also hard. In the US, the algorithm looks for agreement across party lines. But in a country like India with six major parties, measuring diversity is much harder. There's also risk of coordinated manipulation by organized groups.

The Behavioral Shift

When researchers studied Birdwatch's first phase, they found the program changed how people wrote.

Borwankar, Zheng, and Kannan (2022) analyzed 1.1 million tweets from 1,100 users. After the program launched:

Cognition in writing increased—users posted more analytical content

Sentiment extremity decreased—less emotional volatility

Content read less like misinformation

But quantity dropped: new tweets fell 20%, length dropped 21%, retweets fell 39%

These changes came from awareness of being monitored, not time spent fact-checking. Separate research confirms that when notes appear, authors often delete their posts voluntarily.

The Anonymity Effect

In November 2021, Twitter anonymized fact-checkers with random aliases. The results were striking (Borwankar, Zheng, and Kannan 2024):

    Content frequency increased

    Neutrality improved (less negative emotion)

     Analytical thinking and authenticity rose

     Users talked more about politics; they felt safer engaging with controversy

Fact-checking activity didn't decrease. Anonymity didn't create a "displacement effect", it just made people post more while fact-checking the same.

Limitations

Despite the promise, problems remain. Less than nine percent of notes publish. Defining "diverse viewpoints" across different countries is messy. Platforms don't remove content even when users flag it,no real consequences. X also laid off over 80% of its Trust and Safety team, gutting oversight. (Borwankar, The Conversation, 2025)

Transparency

X made Community Notes data public from the start, a rare move. Regulators must ensure this continues, or the program risks becoming symbolic.

Conclusion

Community Notes changes behavior. Participants write better with less extreme emotion. Anonymity helps them engage with controversy. But there are costs: content drops, most notes never publish, and platforms keep ultimate power while gutting oversight. The crowd is watching. The question is whether platforms are listening.

Bibliography

Borwankar. “Meta’s Community Notes Program Is Promising, but Needs to Prioritize Transparency.” The Conversation, 2025.

Borwankar, Zheng, and Kannan. “Unveiling the Impact of Privacy-Preserving Policies in Crowd-Based Misinformation Monitoring Program.” 2024.

Borwankar, Zheng, and Kannan. “Democratization of Misinformation Monitoring: The Impact of Twitter’s Birdwatch Program.” 2022.

Center, News, and Sandra Knispel. “The Most Effective Online Fact-Checkers? Your Peers.” News Center, 17 Nov. 2025, www.rochester.edu/newscenter/crowdsourcing-fact-checking-community-notes-social-media-676142/.