We need your help to keep this website free. You can 
xAI has updated the public code behind X's recommendation project, adding a field for people mentioned in posts. Creators should resist turning that finding into an instant tagging strategy. The code shows where mention information can be carried; it does not show that adding names to a video post earns more impressions.
Source: xAI's public X algorithm repository on GitHub. Cropped code-diff screenshot in an editorial layout; it is not an X feed screenshot or a measurement of post performance.
What changed in the published code
The October 9 repository update touches 16 files. One addition is an optional mentioned_users list in the post data definition. Another piece of the diff maps mention metadata into that list when the metadata is available.
We also checked the home-mixer parameter file in the same commit. Its displayed change is to the synchronization header. The diff does not change the parameter values there. Putting those two observations side by side is useful: a file can appear in an update without a ranking setting inside it being altered.
A mention field is not a reach score
Readers can inspect the new list in the post data definition. What they cannot get from that definition is an impressions comparison between tagged and untagged posts. A recommendation system needs information to work with, but the presence of one input does not disclose its influence on a recommendation.
Keep the publishing workflow in perspective
There is no announcement here of a thumbnail uploader, a new image-size requirement or a cover-testing feature. For someone preparing a video, this is code news rather than a change to the cover-selection workflow. Our guide to cinematic YouTube thumbnails deals with visual presentation on YouTube, not a newly discovered rule for X.
The contrast with YouTube thumbnail and testing tools is worth keeping. That report concerns an announced publishing tool. This one concerns a visible code change. Neither should be used to promise an X reach boost, and a tool on one platform should not be assumed to exist on another.
A platform explanation or a performance study could give creators firmer grounds for a change in strategy. This diff alone does not. Mentioning a genuinely relevant person is still an understandable editorial choice; padding a caption with names on the assumption of guaranteed distribution is not supported by the evidence inspected here.
