X has taken a dramatic step toward algorithmic transparency by releasing a significantly expanded portion of its ranking codebase and introducing new tools that show users whether their posts or accounts have been affected by visibility‑reducing labels. The company’s move centers on open‑sourcing the “For You” feed’s core ranking engine—now 10 to 15 times larger than earlier releases—and making the underlying model configurations, filters, and scoring parameters publicly accessible on GitHub.
The update goes beyond symbolic openness. X’s VP of Product Keith Coleman explained that users can now inspect the same ranking code that pulls posts, scores them, and assembles the feed for each viewer. Some components, including the ranker and scoring systems, can even be run outside the company, giving developers and researchers unprecedented visibility into how posts rise or fall in the feed.
Alongside the code release, X is rolling out a new transparency feature inside an “Under the Hood” settings page. Users who have posted at least ten times in the past month can download a JSON file containing aggregated stats that reveal whether any labels—such as spam, policy violations, or other ranking penalties—were applied to their posts or accounts over the previous calendar month. This tool is launching first to a pilot group of accounts at least a year old before expanding more broadly.
The change directly addresses long‑standing user frustration around “shadowbanning,” a term often used when reach suddenly collapses without explanation. While the open‑source code does not show a secret switch that silently suppresses accounts, it does reveal several real mechanisms that reduce visibility: post‑level filters for spam or rule‑violating content, per‑viewer filters based on blocks or mutes, and negative ranking weights for posts the model predicts users might dislike or report. These systems collectively shape distribution and can now be inspected rather than guessed at.
By making algorithmic behavior partially inspectable, X is shifting the conversation from speculation to evidence. Users, developers, and researchers can now analyze how ranking decisions are made, challenge them, and even propose changes through GitHub pull requests. It’s not full transparency—live enforcement and real‑time model behavior remain opaque—but it marks a significant cultural shift for a major social platform.