Trial by Note: 65 Submissions, 3 Published

A Washington Post columnist volunteered for Meta’s new Community Notes system on Facebook, Instagram, and Threads. Over about four months, he wrote 65 notes debunking everything from a fake image of Donald Trump at the papal funeral to a supposed ICE and DoorDash partnership. He cited official livestream footage, Snopes, time stamps, everything. But only three notes were ever published, all related to floods in Texas. That’s a publishing rate under 5 percent despite targeting widely circulated falsehoods.

How Community Notes Is Supposed to Work

The system is open to any U.S. Meta user over 18 who applies, but acceptance is random. Once in, you can flag misleading content with a note and link to credible sources. Those notes must then be rated "helpful" by enough contributors from different ideological backgrounds before they appear publicly. That mechanism, called a bridging algorithm, demands consensus among users who have previously disagreed to reduce bias.

Why It Barely Works

Meta removed professional fact-checkers in January to make way for user-driven moderation, and it shows. First, real efforts are rejected because the algorithm is overly cautious. Even fact-based, non-polarizing notes fail if they don’t gather agreement across divergent raters.

Second, the contributor pool is shallow. Some posts can’t even receive notes because they come from accounts outside the U.S. Technical issues and inconsistent participation only add to the mess. Volunteers rarely stick around when nothing they write makes an impact.

Meta Throws Up Its Hands

Meta says the program is too new to judge. Spokespeople call it a test-and-learn phase, meant to scale slowly and build a contributor base. They refused to share how many people participate, how many notes get published, or whether it’s making a difference. Transparency has been promised repeatedly, but not delivered.

Outside Experts See Cracks Too

Alexios Mantzarlis, who leads the Security, Trust and Safety Initiative at Cornell Tech, ran an independent evaluation in June. He concluded the program is “not yet ready for prime time.” Only a few notes offered meaningful context, and some had errors. Meta’s so-called corrective system is still just a skeleton. It hasn’t rolled out critical updates or shared performance data that would inspire trust.

That Bridging Algorithm Is Too Smart for Its Own Good

Kolina Koltai, who helped build Community Notes at X (formerly Twitter), says Meta’s algorithm is “very, very conservative.” It avoids false positives but at a huge cost, it buries most accurate notes unless they survive a gauntlet of disagreement among raters. On X, Koltai says her publish rate is about 30 percent. On Meta, she suspects it’s far lower.

A Growing Trend and Growing Risk

This isn’t just Meta’s problem anymore. YouTube is testing a similar system, and TikTok is piloting something called Footnotes. If the community-note model becomes the default way to fight misinformation, it risks failing on a massive scale if the foundation is broken.

How to Fix It

If Meta cares about truth, it needs to:

• Allow international users to submit notes
  
  
• Prioritize posts that are likely to spread harm or go viral
  
  
• Reward active contributors (badges, incentives, or even pay)
  
  
• Let professional fact-checkers earn verified roles with credentials

Truth doesn’t have to come from experts or crowds alone, it needs both if accuracy is the goal.

Why This Matters

About 54 percent of American adults say they get news from social media. That means Meta’s current Community Notes system isn’t a safeguard. It’s a gap. Meta already cut its in-house fact-checkers. Without fixing what replaces them, the platform risks becoming a breeding ground for unchecked misinformation.