Digital Feed Adjustments Rapidly Intensify Partisan Hostility, Study Confirms

New scientific evidence confirms that social media algorithms can dramatically accelerate political animosity between partisan groups. Researchers successfully demonstrated that barely noticeable changes to X users’ feeds produced levels of political polarization in one week that would typically take three years to emerge naturally, highlighting the platform’s massive influence on political attitudes.
The experimental design was both innovative and rigorous. Scientists created a system to manipulate X’s “for you” feed in real-time, using artificial intelligence to classify posts based on their potential to increase division. More than 1,000 users participated during the 2024 presidential election, with some receiving slightly more divisive content and others receiving less, while most remained unaware of the manipulation. The divisive content included posts supporting undemocratic practices, partisan violence, opposition to bipartisan consensus, and biased evaluations of politicized facts.
According to Martin Saveski, who helped lead the research, the algorithm’s power is demonstrated by the fact that barely perceptible changes to feeds resulted in significant differences in how users felt about political opponents. Tiziano Piccardi added that the shift observed in just one week corresponds to approximately three years of polarization based on established trends in American political culture.
The measurement approach involved asking participants to evaluate their feelings toward opposing political parties on a “feeling thermometer” scale ranging from 0 to 100 degrees. Users exposed to more divisive content showed increased hostility of more than two degrees—the same amount of increased polarization that occurred across American society during the four decades leading to 2020. Repeated exposure to posts expressing antidemocratic attitudes and partisan animosity significantly influenced polarization feelings while also boosting emotions like sadness and anger.
The study arrives at a critical moment for democratic societies. Polling indicates that majorities in multiple countries believe political divisions have become dangerously severe, with people unable to agree even on basic facts. While the research confirms concerns about platforms amplifying divisive content for engagement and revenue, it also reveals that relatively subtle algorithmic changes could significantly reduce political animosity. The researchers found that down-ranking divisive content resulted in only modest decreases in overall engagement, while users actually showed higher rates of meaningful interaction through likes and reposts, suggesting platforms could pursue social responsibility without completely undermining their business models.