Analysis of 1K+ posts from 6 partisan Facebook Pages shows that least-accurate are shared widely; 38% of right-wing, 19% of left-wing posts are false/misleading
Craig Silverman / BuzzFeed :
Context & Ripple Effects
In the final weeks of the 2016 campaign, Craig Silverman's BuzzFeed analysis put numbers on what editors had been seeing anecdotally: across six partisan Facebook Pages, the least-accurate posts traveled furthest, and falsehood was not evenly distributed — 38% of right-wing posts versus 19% of left-wing ones rated false or misleading. The asymmetry matters because it predicts which publishers the platform's engagement mechanics will reward.
The later corpus confirms the pattern held and hardened. Right Wing News outperformed mainstream pages like Breitbart and Fox News from 2013–2016 before being swept out as inauthentic, and a 2021 study found far-right publishers earn a 65% engagement premium when they publish fake news, while liberal and centrist outlets are penalized for it. The 2016 dataset is the baseline measurement for that incentive structure.
First-order effects
- Facebook faces immediate pressure over a ranking system whose top performers on partisan Pages are its least accurate content, while columnists and commentators simultaneously warn the company against responding with broad censorship of false speech.
- Page owners on both flanks now have a measured benchmark — and for right-leaning operators especially, a documented payoff for posting unverified claims.
Second-order effects
- Fact-checking is structurally outgunned at the distribution layer: the related data shows top fake stories generating tens of millions of engagements against roughly 128K for fact-checks, so publishers and advertisers can price reach around falsehood rather than correction.
- Competing platforms inherit the same audit question — later studies of Twitter, YouTube, and Facebook during COVID-19 found most rated-false content stayed up unlabeled — making cross-platform enforcement consistency a competitive and reputational issue.
Third-order effects
- If the engagement premium for false political content persists, Facebook's ad-driven model systematically subsidizes the most inaccurate publishers, pushing the industry toward engagement-based pay structures that reward fabrication regardless of ideology.
- The recurring gap between detection studies and enforcement outcomes points toward regulation or external auditing of recommendation systems as the likely resolution path, since voluntary labeling has repeatedly covered only a minority of flagged content.
The trend: Political misinformation on social platforms is shifting from an anecdotal complaint to a quantified, recurring finding — each election cycle producing new measurements of how much more engagement false content earns than corrections.