X Tweaks Algorithm to Prioritize Mutual Follows

X has rolled out a change to its recommendation system designed to give users more visibility into posts from people they follow, particularly mutual connections. The adjustment addresses a gap that had reportedly caused content from familiar accounts to get buried under algorithmically suggested posts from strangers.

Why the Change Was Needed

According to X’s head of product, Nikita Bier, the platform noticed that data reflecting mutual follows was missing from its algorithm, which caused friends to show up less often in reply sections. Bier noted the fix should also make it easier for interest-based groups of users to form organically, something people had reportedly asked for. The change is intended to make replies feel more like conversations among people users actually know, rather than exchanges dominated by unfamiliar accounts.

A Broader Pattern in Social Media

While the update may look like something that should have always been standard, since a follow is an explicit signal of what a user wants to see, it highlights a wider trend across social platforms. Engagement-driven algorithms increasingly favor behavioral signals, such as likes, watch time, and shares, over users’ direct preferences. Research has indicated that platforms often see higher engagement when they lean on usage-based recommendations rather than strict chronological or follow-based feeds. This has created recurring friction, as AI-driven suggestions can crowd out updates from people users have deliberately chosen to follow.

The rollout suggests X is attempting to strike a better balance between algorithmic discovery and user-defined preferences, at least for now, in a landscape still largely shaped by automated recommendation systems.