Mark Zuckerberg’s Meta, the parent company of social media sites such as Facebook, Instagram and Threads, has started testing its new community notes fact-checking tool based on crowdsourcing, where accuracy of the post and its verification is left to the user’s wisdom. This is a borrowed concept from Elon Musk’s X.
The change replaces conventional third-party fact-checking on Meta’s platforms, which was shut down this January on the pretext that fact-checkers had become politically biased, hence the move is intended to encourage “free speech” and “counter bias and censorships”. CEO Zuckerberg, in a blog post, had said that it was “time to get back to our roots around free expression”. “We will allow more speech by lifting restrictions on some topics that are part of mainstream discourse and focusing our enforcement on illegal and high-severity violations,” he said.
He also said they have “destroyed more trust than they’ve created”, especially in the US. Reportedly, the move comes as Zuckerberg and other internet CEOs are working to improve relations with US President Donald Trump. Trump and his Republican allies have criticised Meta’s fact-checking policy, claiming it censors right-wing viewpoints.
This is a departure from Meta’s commitment to third-party fact-checkers like Reuters and AFP. Four years ago, after the January 6 Capitol riots in 2021, which were fuelled by claims of a rigged US presidential election, Zuckerberg had to give a testimony to the Congress where he boasted about Meta’s “industry-leading fact checking programme”, pointing out it was based on 80 “independent third-party fact checkers” to curb misinformation on Facebook and Instagram.
Zuckerberg’s opinion on those fact-checkers has changed. Inspiration has come from Musk who claims to be a proponent of “free speech” and his X. While people vetting information and essentially participating in curbing misinformation online sharing some responsibility sounds ideal, there is a debate on how effective they are, as reports, studies and case studies from X have something else to say.
X’s mixed record
Experts are sceptical. In a BBC report, Alexios Mantzarlies, director of the security, trust and safety initiative at Cornell Tech, has been quoted as saying that Zuckerberg is pandering to the incoming administration and to Musk.
Another issue is the mixed record of the fact-checking through community notes. X’s community notes, launched initially as Birdwatch in 2021, was structured to function like Wikipedia by depending on volunteer efforts. Reports suggest that the system generates “hundreds of fact checks per day”, significantly higher than professionally handled by fact-checkers, amounting to “less than 10% per day”.
Community notes have worked at times, as a matter of fact. For example, a 2024 study published in JAMA Network said a sample of community notes added to posts on X containing Covid-19 vaccination misinformation primarily addressed adverse events and conspiracy theories, were accurate, cited moderate and high credibility sources, and were attached to posts viewed hundreds of millions of times. The researchers looked at a sample of 205 community notes about Covid-19 vaccines and found that the user-generated information was accurate 96% of the time, and that the sources cited were of high quality 87% of the time. While only a small fraction of misleading posts were flagged, those that did get notes attached were among the most viral.
‘90% of notes never published’
As encouraging as it seems, there are still issues. In the BBC report, it is highlighted that its dependence on the need for building cross-ideological consensus before a note can be published holds up the process and eventually lets perpetuation of misinformation a go-bye. A Bloomberg analysis showed that fewer than 10% of the notes submitted through X’s system are shown in the app. This means 90% of suggested notes are never published.
In India, particularly, X’s community notes during the 2024 general elections had exposed some major drawbacks. It had failed to fight against disinformation about political figures because of the bipartisan consensus requirement, resulting in unaddressed and misleading posts. Examples include claims and speeches around prominent political personalities in whose reference opinions are always polarised.
Another aspect of X’s community notes is its use of an algorithm to decide which notes are shown, which is based on choosing notes that have been marked as useful by users who tend to disagree, which again opens a prospect of division on both or multiple sides of the issue. This also prevents many correct notes from ever being presented.
Same, same but different?
What is yet to be seen is the reception to the Meta’s community notes which it has signalled will do the same but few details are available. Meta said in its blog post that over time it plans to modify the algorithm to better serve platforms.
“As X’s algorithm and programme information is open source — meaning free and available for anyone to use— we can build on what X has done, learn from the researchers who have studied it, and improve the system for our own platforms,” Meta wrote. “As our own version develops, we may explore different or adjusted algorithms to support how community notes are ranked and rated,” it said. So far, about 200,000 people have signed up to become community notes contributors “and the waitlist remains open for those who wish to take part in the programme,” it further said.
