Not impressed with a political post on X? Elon Musk’s social media platform may show you more of those, says new study

Algorithmic Misalignment: A Study on X's Feed Algorithm

A recent study published in the Proceedings of the National Academy of Sciences has shed light on the inner workings of X's algorithm, revealing a concerning pattern of 'hate management'. The research, titled 'Value misalignment of X's feed algorithm', found that the algorithm prioritises engagement above almost everything else, constantly shaping what users see on their personalised feeds.

The study involved 715 active American users of X, who participated in the research during September and October 2024. The researchers matched participants based on ethnicity, gender, and political affiliation, and installed a browser extension to collect their feeds, including both their 'For You' and Following pages.

The study found a clear misalignment between stated values and content shown. Users followed accounts that generally reflected their personal values, but amplified content often contradicted these same stated values. This misalignment became particularly evident through user replies, which represented just 6.8% of interactions overall, but carried disproportionate influence over algorithmic behaviour.

The researchers noted that the algorithm learns from user replies, particularly those that provoke anger, and continues to serve more content in that direction. This creates a feedback loop that consistently encourages outrage, and is particularly true of Democratic users.

The study also found that the pattern appeared across the political spectrum, though unevenly distributed. Democrats specifically confronted misaligned content through replies more frequently, exploiting the algorithm's engagement-maximising design.

The researchers expressed concern about the potential risks of value-based feeds creating isolated content bubbles, and encouraged users to reflect deliberately on desired content beforehand. They also suggested that further research is needed to explain the partisan difference in the study's findings.

Implications for Society

The study's findings have significant implications for society, particularly in the context of social media's influence on daily life. The researchers noted that these algorithms hold immense power over information consumption, with very little public transparency.

The study's results highlight the need for a deeper understanding of social media design and its impact on user values. It also underscores the importance of intentional content consumption practices, such as pausing before automatically engaging with provocative posts.

A Call for Further Research

The study's lead author, Ziv Epstein, expressed a desire for further research to explain the partisan difference in the study's findings. He suggested various possible explanations, but ultimately noted that more research is needed to understand the complex dynamics at play.

The study's findings serve as a warning about the potential risks of social media algorithms creating echo chambers and amplifying provocative content. It highlights the need for a more nuanced understanding of social media design and its impact on user values, and encourages users to be more intentional in their content consumption practices.