Algorithms do widen the divide: Social media feeds shape political polarization (english.elpais.com)

U.S. researchers have demonstrated that the sequence of political messages displayed on social media significantly influences users' polarization, a pressing concern in the realm of digital political engagement. Published in Science, the study utilized a browser extension that rearranged the feeds of over 1,200 participants on the platform X, scoring content for "antidemocratic attitudes and partisan animosity" using a large language model. This innovative approach provided insights into how algorithmic transparency, or the lack thereof, affects political views, especially in the context of the upcoming 2024 U.S. presidential election. The findings reveal that altering the order of visible posts can substantially reduce affective polarization, regardless of users' political affiliations. The study emphasizes the potential for algorithmic interventions to promote social trust and mitigate partisan hostility on social media platforms. Notably, this research methodology bypasses the need for direct platform collaboration, marking a significant step forward in assessing algorithmic impact amidst increasing restrictions on data access by social media companies. The implications extend beyond X, indicating possible applications in other networks to foster healthier political discourse.