Meta-backed research finds exposure to ‘untrustworthy’ social media is rare. The fine print is less reassuring

A new study on social media misinformation sounds reassuring at first: the average Facebook user sees barely one percent untrustworthy content, and Instagram users see even less. But the research, from a Meta and academic collaboration on the twenty twenty election, comes with significant fine print. When researchers removed posts from repeat offenders from half the study's sixteen thousand users for three months, misinformation exposure dropped seventy percent. Yet people's beliefs didn't change. Behavioral data scientist Marian-Andrei Rizoiu says the limitations are crucial. The study only counts misinformation that Meta's fact-checkers actually caught, missing a huge grey zone. One example: undetected COVID vaccine misinformation had forty-six times the impact of flagged content. The baseline itself was already scrubbed: Meta had deployed sixty-three emergency election measures cutting untrustworthy content by a quarter before the experiment started. The removal dose was modest, just three of two hundred fifty daily posts, tested on volunteers. And since twenty twenty, Meta has ended third-party fact-checking and replaced it with Community Notes, which even Meta's own Oversight Board called inadequate, per The Conversation.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton