The political effects of X’s feed algorithm

Why Your Social Media Feed Matters

Many of us now learn about politics from our phones, scrolling through endless posts chosen for us by hidden algorithms. This study asks a question that affects anyone who uses X (formerly Twitter): does the way the platform orders and recommends posts actually change what we think about real-world political issues—and if so, how?

Two Ways to See the Same Platform

The researchers focused on a feature X offered in 2023: users could choose between a simple chronological feed, which shows recent posts from accounts you follow, and an algorithmic feed, which mixes in posts from people you do not follow and reorders everything for maximum engagement. They recruited nearly 5,000 active X users in the United States and randomly assigned them to use either the algorithmic or the chronological feed for about seven weeks in the summer of 2023. Because some people already used one feed or the other, the experiment let the team study both what happens when the algorithm is switched on for someone used to a chronological feed and what happens when it is switched off for someone used to the algorithm.

What Changed in People’s Views

After the seven weeks, participants answered questions about how often they used X, which policy issues they thought were most important, and their views on hot-button topics such as the criminal investigations into Donald Trump and the war in Ukraine. The team combined these answers into broader measures of user engagement and political attitudes. For people who started with a chronological feed, moving to the algorithmic feed made them use X a bit more and, crucially, nudged their opinions in a more conservative direction. They were more likely to prioritize issues commonly stressed by Republicans, such as inflation, immigration and crime, more likely to say the investigations into Trump were unacceptable, and more likely to hold views sympathetic to the Kremlin’s position on the war in Ukraine. In contrast, people who started on the algorithmic feed and were switched to the chronological feed did not show meaningful shifts on any of these political measures.

What Did Not Change

The researchers also examined two broader traits often discussed in debates about social media: whether people feel emotionally hostile toward the other political side (affective polarization) and whether they identify as Democrat, Republican or something else (partisanship). On these, the algorithm seemed to have no detectable effect, regardless of whether it was turned on or off. This suggests that deeply rooted political identities and gut-level feelings about “us versus them” are harder to move over a short period than opinions about specific issues or current events. The study also did not find clear effects on participants’ self-reported wellbeing or life satisfaction.

How the Algorithm Steers Content

To understand why turning the algorithm on mattered but turning it off did not, the team looked at the actual posts people saw. A subset of participants installed a browser extension that captured the first 100 posts from each feed type. Using automated text analysis checked against human judgments, the researchers found that X’s algorithm strongly promoted content that already attracted lots of likes, reposts and comments. It also increased the share of political content overall and tilted that content toward conservative messages. Posts from traditional news outlets appeared less often in the algorithmic feed, while posts from political activists and entertainment accounts appeared more often. This pattern held for users on both sides of the political aisle: compared with a chronological feed, the algorithm gave everyone a higher fraction of conservative political content.

Lasting Effects Through New Follows

Another key piece of the puzzle came from examining which accounts people chose to follow. Among users who started with the chronological feed, those switched to the algorithmic feed began following more conservative accounts and, especially, conservative political activists. These new follows then reshaped their feeds going forward, including what appeared in the chronological view. Even after the experiment ended—and even when the algorithm was later turned off—the accounts people had chosen to follow because of algorithmic exposure continued to supply them with similar content. Meanwhile, switching the algorithm off for people already used to it did not noticeably change which kinds of accounts they followed, helping explain why their attitudes did not shift.

What This Means for Everyday Users

In plain terms, the study suggests that X’s feed algorithm does more than merely show you what you already like; it can gently, but persistently, steer your political views on concrete issues by serving you more engaging and more conservative content and by prompting you to follow new political voices. Once those connections are made, they keep shaping your feed even if you later opt out of algorithmic sorting. The work helps reconcile earlier research that found no effect from switching algorithms off: by the time people flip that switch, the algorithm may already have done its work. For citizens, journalists and policymakers, the results underscore that invisible choices about how our feeds are ranked can play a quiet yet important role in how we see politics and the world.

Citation: Gauthier, G., Hodler, R., Widmer, P. et al. The political effects of X’s feed algorithm. Nature 652, 416–423 (2026). https://doi.org/10.1038/s41586-026-10098-2

Keywords: social media algorithms, political attitudes, X platform, online news feeds, digital polarization