A new study published in the Proceedings of the National Academy of Sciences finds that X’s feed algorithm prioritises engagement above all else when generating a user’s For You Page. The paper, titled Value misalignment of X’s feed algorithm is a reflection of value tensions in engagement, also shows that the platform serves more ragebait to people who identify as Democrats, though the exact cause remains unclear.
How the system works
Ziv Epstein, a postdoctoral researcher at Stanford University and co-author of the study, told 404 Media that while the headline might not be surprising in 2026, the underlying mechanism warrants closer inspection. “X’s feed algorithm… is optimized for engagement,” Epstein said. “But it turns out that not all types of engagement are considered equally.”
The research team recruited a nationally representative sample of 715 Americans who were active users of X in September and October 2024. They matched the group by ethnicity, gender, and partisanship before installing a browser extension to collect data from their For You Page and Following feeds.
The study was observational, designed to answer questions about how users want their feeds to function and who controls that design. “There are these social media algorithms that have enormous amounts of power in our lives,” Epstein said. “They shape the information that we consume, and we have very little transparency into how they operate and what their implications are.”
Researchers used the Schwartz Theory of Basic Values to gather a values inventory from volunteers. This tool presents a wheel with 19 points covering traits like tolerance, dominance, and openness to change. Participants also reported their political alignments.
The data showed that while posts from accounts users followed reflected their stated values, there was a negative correlation between those values and the content the algorithm amplified. In short, users saw content that conflicted with what they said they cared about.
Consider a user seeing a press release from a politician in a party they dislike. If they reply to the post, X registers that action as engagement. The algorithm then sends more infuriating posts to that user’s feed. It does not matter who they follow or what their stated preferences are.
“When we look at commenting… that’s where we actually see some kind of meaningful misalignment between people’s values and the values of the content they’re replying to,” Epstein explained.
Most participants liked and reposted content, receiving more of the same. Replying was rare, accounting for just 6.8% of interactions, but it had an outsized impact on the system. “Replying is only a fraction of engagement, but there does seem to be some evidence that these algorithms are prioritizing and learning more from this kind of rarer form of engagement,” Epstein said.
This creates a feedback loop. The algorithm learns that a user gets outraged, then continues to serve more content in that direction. This dynamic appeared particularly true for Democratic users in the study.
While this occurred across the political spectrum, ragebaiting happened more frequently for users identifying as Democrats. “Democrat users confront the abundant value-misaligned content by replying to it, which the algorithm in turn preferentially learns from and continues to feed them,” the study said.
The authors highlight a core tension: engagement-maximising algorithms exploit frictions between users’ stated preferences and their reactive confrontations to create runaway feedback loops of increasing value misalignment.
Epstein noted he needed further research to determine why X serves ragebait to Democrats more often than Republicans. Possibilities included a higher volume of rightwing content on the platform overall or a tendency for Democrats to engage with disagreeable posts more often. “There might be some kind of differential effects on information diets there, or it might be something more psychological about how different you know partisan identities are triggering different kinds of actions and reactions,” he said. “Ultimately I don’t want to speculate too much.”
Epstein acknowledged concerns that prioritising values could lead to siloed user bases and echo chambers. “Where people just… have the certain values that they have, and all the content they see is just aligned with those values. I think that is a very scary and dark reality,” he said.
He also suggested that thinking about values before opening a feed could be positive. “You’re pumping the brakes, you’re taking a breath, and you’re thinking about what you actually really care about,” Epstein said. He speculated that many people want to see diverse content and engage meaningfully across lines, even if that is not a specific point on his 19-dimensional wheel. “This is just a starting point of thinking about our intentions and thinking very deliberately about the kind of information we want to be exposed to, versus the knee-jerk reaction that we’re kind of learning in this very kind of short, shallow attention span, emotion and negative affect-driven model of these very myopic forms of engagement.”
X did not return 404 Media’s request for comment.
