Most Facebook and Instagram users see very little content from sources that repeatedly post falsehoods, but a small slice of users receives a dramatically higher share, according to a new study published in Science Advances.
The research came from a collaboration between Meta and academic researchers studying the 2020 U.S. election. It was conducted directly on the live feeds of Facebook and Instagram, giving it access that outside researchers typically cannot obtain.
The team, led by Olivier Bergeron-Boutin from the University of California, Berkeley, measured how much content across 231 million adult U.S. Facebook accounts and 200 million Instagram accounts came from what they called untrustworthy sources. Those were defined as pages, groups, accounts, and websites repeatedly caught posting falsehoods by Meta's third-party fact-checkers.
On average, exposure was low. The median Facebook user saw untrustworthy content in 1.1% of their feed. For the median Instagram user, the figure was just 0.1%. But the distribution was far from even. About 53 million Facebook accounts, representing 23% of users, received nearly 80% of all content from untrustworthy sources. For the most exposed 2.5% of users, untrustworthy sources supplied roughly 60% of the political and social content they saw from pages and groups.
For three months around the 2020 election, posts from repeat offenders were removed from the feeds of half of nearly 16,000 consenting users. Exposure in those targeted feeds dropped by about 70%. Researchers then tracked 10 outcomes on each platform, including belief in false claims, the ability to distinguish true from false information, political polarization, and trust in media.
None of those measures moved. The result held even among the heaviest consumers of untrustworthy content, though the study notes that estimates for that specific group are less precise.
The findings echo results from similar experiments in which changing feed algorithms, removing reshared posts, or even paying people to log off entirely produced little movement in attitudes over the short term.
The study does carry limitations. Researchers writing about the findings noted that political attitudes are shaped by many forces, including television, talk radio, politicians, and personal relationships, making it difficult to isolate social media's specific contribution. People who seek out fringe content also tend to cluster together in tight networks, which can make them harder to capture in studies like this one.
The researchers also noted that the weakest parts of the study sit at the points between what was measured and what was not, leaving open questions about longer-term effects and populations not captured in the experiment.
