X's Algorithm Feeds Off Ragebait and Impacts Democrats More, Study Finds

A new study published in the Proceedings of the National Academy of Sciences (PNAS) reveals how X's algorithm feeds users more content that aligns with their negative emotions and partisan biases. According to the paper, X's algorithm prioritizes engagement over all else, often delivering "ragebait" content to its users. The researchers found that Democrats on the platform are more likely to receive such content, although the exact reason remains unclear.

The study involved 715 active X users, who installed a browser extension to collect their For You Page and Following feeds. The researchers used a values inventory based on Schwartz Theory of Basic Values to assess users' self-reported values, which consist of 19 points such as tolerance, dominance, hedonism, and openness to change. They also asked users about their political affiliations.

The study found a negative correlation (misalignment) between users' stated values and the values of the content that the algorithm amplified. When users encountered posts that made them angry, such as political statements from parties they disliked, X interpreted the act of replying as engagement and served more of that infuriating content. Despite replying to posts being a rare behavior (just 6.8% of interactions), it had a significant impact on the algorithm's behavior.

The researchers concluded that this feedback loop of outrage baiting reinforces value misalignment, creating a runaway loop that further polarizes users' information diets. While this effect was observed across the political spectrum, the study found that it was more pronounced for users who identified as Democrats. Epstein, the study's co-author, noted that more research is needed to understand why X serves ragebait content to Democrats more often than Republicans.

The reasons could include a higher volume of right-wing content on the platform or Democrats engaging more frequently with posts they disagree with. Epstein also expressed concern about the potential for value echo chambers, where users are only exposed to content that aligns with their pre-existing beliefs. However, he also emphasized the importance of users consciously considering the content they want to see on social media platforms.