Elon Musk Changes X Algorithm Again To Focus on Read Time Instead of Just Retweets and Likes

Elon Musk has outlined changes to X's recommendation system designed to improve predictions of how long users view posts and identify content they read from beginning to end. The changes could affect what appears in users' For You feeds, although their precise impact remains unclear.

In a post on X, Musk described two developments in the platform's recommendation system. One concerns predicting the time users spend viewing posts, while the other aims to identify content people read all the way through. His announcement does not establish whether the changes have been fully rolled out to all users.

The updates could influence which posts appear in users' feeds, including content from accounts they do not follow. However, X has not disclosed how much influence the new models will have compared with existing engagement signals.

How the X Algorithm Will Prioritise Reading Time

Musk described a ranking model intended to predict time spent viewing posts more accurately. He also outlined a retrieval model designed to identify posts that users are likely to read to the end.

The two systems perform different roles. Retrieval identifies posts that could be included in a feed, while ranking helps determine their order. Changes to either stage could influence which content users see.

X's public recommendation algorithm combines posts from accounts users follow with recommendations from outside their networks. Its documentation describes models that predict how users might interact with posts, including liking, replying, reposting and spending time viewing content.

The announcement does not establish that existing engagement signals are being removed. Instead, it points to an effort to improve predictions of how long users will spend viewing posts.

A post with fewer likes might still attract attention if the system predicts that users will spend longer viewing it. However, X has not published enough information to determine how the new models will rank specific types of content.

What the X Algorithm Changes Could Mean for Users

Users could notice changes in the content appearing in their For You feeds. Detailed explanations and threads might benefit if the new models favour posts that people are more likely to finish reading, although X has not confirmed that outcome.

Posts from accounts users do not follow could also appear in their feeds. However, the announcement does not establish whether smaller accounts will gain more visibility or how the retrieval model might affect creators of different sizes.

Reading time is not a measure of accuracy or quality. A useful explanation may hold a reader's attention, but so could a provocative claim or heated argument. Time spent viewing a post does not establish whether its information is accurate, balanced or useful.

The changes also do not guarantee that longer posts will dominate. A short post could perform well if users are likely to read it in full, while a lengthy thread might lose their attention. X has not published enough detail to determine how post length will affect recommendations.

Musk Claims Time Spent on X Is Rising

Alongside the announcement, Musk claimed that user active seconds were growing exponentially and reaching new highs almost every day. He shared a chart showing the metric, but the figures have not been independently verified. The available information does not establish the rate of growth or whether the announced recommendation changes contributed to it.

Time spent on a platform can create more opportunities to display content and advertisements. However, Musk's announcement does not establish that advertising revenue is the reason for the algorithm changes. Longer viewing time also does not necessarily mean users are seeing more reliable information.

Users who want to adjust their feeds can provide feedback on individual posts, mute accounts or keywords, or switch from For You to Following. The Following feed displays posts from accounts they follow in reverse chronological order, according to the platform's described functionality. Users should check X's current help documentation for the latest instructions and menu labels.

The precise impact of the changes remains uncertain. Until X provides more details about the models and their implementation, it is too early to determine how significantly they will change what users see in their feeds.