Meta’s Community Notes program is promising, but needs to prioritize transparency

Meta, the parent company of Facebook, Instagram and WhatsApp, has more than 3.35 billion combined monthly active users. Recently, Meta has changed its approach to fact-checking in response to criticisms of its role in circulating fake news and disinformation. The company frames its Community Notes program as a way to uphold free expression.

Although Meta has not officially announced a launch date for Community Notes in Canada, interested users can join the waitlist via Meta’s Community Notes page.

The initiative was first launched in the United States, and will be expanding globally. Meanwhile, X (formerly Twitter) has already experimented with a similar program, with mixed results. The experience of X’s Community Notes (previously known as Birdwatch) underscores that both platforms and regulators must take an active role in refining these programs.

Meta has the opportunity to learn from four years of Community Notes evolution at X and improve upon its shortcomings. This involves adjusting features, addressing algorithmic biases and ensuring that they function as effective tools rather than mere symbolic gestures.

Community-driven moderation

X launched Birdwatch in January 2021 when it was known as Twitter. Marketed as a way to broaden the range of voices beyond platform-based and centralized fact-checking efforts, the program aimed to curb the spread of misinformation through community-driven moderation.

Over time, the team refined many of its features based on feedback from pilot participants and internal research. When Elon Musk acquired Twitter in 2022, he implemented major changes, including a rebranding of the program’s name.

Community Notes operates on the principles of crowdsourcing, a method proven effective in various domains. Research has shown that groups of users can collectively identify low-quality news sources and misleading content.

On X, users participating in the Community Notes program contribute additional context to posts in the form of notes. They can also rate others’ contributions. Notes that receive supportive ratings from a diverse group of users become publicly visible. Once approved, they appear directly beneath the original post, providing added context for the broader audience.

However, even if a post is widely deemed misleading by Community Notes contributors, the platform does not take action against the post itself or the individuals who spread misinformation. Instead, the program relies solely on surfacing user-generated context rather than the company moderating content.

Positive impact?

Preliminary research suggests that the Community Notes program has had a positive impact on curbing the spread of misinformation on the platform. Recent work shows that when a note is attached to a post, authors often voluntarily retract their posts by deleting them.

On the content creation aspect, participation in the program appears to influence user behavior: contributors tend to adopt a more measured tone, reducing extreme sentiment in their writing after engaging with the system.

One of the most notable strengths of X’s Community Notes is its transparency. Since the program’s inception, X has provided public access to both the data and the algorithms that determine which fact-checks are displayed.

This open-source approach has allowed researchers—both within and outside the company—to study the program and propose improvements. This stands in contrast to the recent trend of social media platforms rolling back data-sharing partnerships.

Prior to Musk’s acquisition, X also had a dedicated team researching the impact of the program. Early changes to the program were shaped by feedback from participants and internal research.

For example, in November 2021, X introduced anonymity for fact-checkers to prevent trolling and harassment. This decoupling of roles between content creators and fact-checkers has had a positive effect, reducing the risk of retaliation and fostering a more positive content creation by the participants of the program.

Challenges and limitations

Despite its potential, X’s Community Notes program faces several significant challenges, including its low popularity among users. Meta now has an opportunity to address these shortcomings from the outset.

One of the biggest concerns is manipulation by co-ordination. Given the presence of organized troll networks on social media, there is a high risk that coordinated groups could misuse the program to flag legitimate content as misinformation.

To counteract this, X implemented a consensus-based approach, where a note is only made visible if users with diverse viewpoints agree on its accuracy.

While this system appears sound in theory, in practice it has led to a severe lack of approved notes as less than 9% of submitted notes reach the general audience. Many contributions never gain visibility, often due to insufficient ratings from diverse users.

Another limitation of the consensus approach is that the algorithm must first recognize diverse viewpoints, which are not always straightforward. Social media platforms operate across hundreds of countries, where political, cultural and social divisions can be complex and nuanced. In such cases, enforcing consensus among a diverse audience may be highly unreliable and require reassessment.

Shifting responsibility

There is also the risk that Community Notes serve as a smokescreen, allowing platforms to shift responsibility away from active misinformation management. Since taking over X, Musk has laid off more than 80% of the company’s Trust and Safety team.

This included members of the Community Notes team, leaving critical gaps in oversight and research. Meta’s recent move to distance itself from third-party fact-checkers suggests a similar retreat from proactive intervention.

Legal frameworks across different countries add further complications. Although Community Notes contributors remain anonymous to the public, it is unclear how platforms will respond if governments demand access to contributor identities.

The Wikipedia legal case in India serves as a cautionary example of how platforms may be pressured into compliance. In September 2024, the Delhi High Court issued a contempt-of-court notice to Wikipedia over the site’s delay in providing identifying information about edits.

No real consequences

Finally, these programs are further weakened by the platforms’ explicit assurance that they will not take enforcement action based on Community Notes outcomes. Without real consequences for those spreading misinformation, the program risks being a symbolic effort rather than an effective tool for content moderation.

Overall, there is hope that Meta’s Community Notes program can be effective, but its success will depend on continuous experimentation and improvement. The company must prioritize transparency to rebuild public trust and ensure the program does not become another performative gesture.

Regulators also have a crucial role in holding platforms accountable, ensuring that data from these programs remains accessible to independent researchers and that the algorithms determining which notes are displayed are fair and unbiased.

Without these safeguards, Community Notes risks becoming yet another tool that platforms use to shift responsibility rather than a meaningful solution to misinformation.

This article is republished from The Conversation under a Creative Commons license. Read the original article.