X has open sourced its recommendation algorithm again, releasing new code and documentation that explain how posts are selected and ranked in users’ feeds, following a transparency pledge made by Elon Musk last week.
Background Of Earlier Algorithm Release
In 2023, the platform then known as Twitter partially open sourced its algorithm for the first time. At the time, Elon Musk had recently acquired the company and said he intended to restructure the social media service to operate more transparently.
That initial release drew criticism from researchers and technologists, who described it as incomplete and argued that it provided limited insight into how decisions were made internally or why the algorithm functioned as it did. Some critics characterized the effort as transparency theater rather than a meaningful disclosure.
New Open Source Release And Promises
The platform, now rebranded as X, followed through on a promise Musk made last week to release an updated version of the algorithm. Musk said the company would make all code used to determine which organic and advertising posts are recommended to users publicly available within seven days and would continue releasing algorithm transparency updates every four weeks.
On Tuesday, X published the materials on GitHub, including an accessible technical explanation of how the feed-generation system works and a diagram outlining the process.
How The Feed Algorithm Works
The documentation shows that when X selects content for a user’s feed, the system evaluates the user’s engagement history, such as posts they have clicked on or interacted with, and reviews recent posts from accounts within their network.
The algorithm also analyzes posts from outside a user’s immediate network using machine learning to identify content it predicts the user may find relevant. Content from blocked accounts, posts tied to muted keywords, and material classified as violent or spam-like is filtered out.
The remaining posts are ranked based on predicted appeal, with the system factoring in relevance and content diversity. The ranking process also evaluates the likelihood that a user will like, reply to, repost, favorite, or otherwise engage with a post.
Role Of Grok And Automation
According to X’s GitHub write-up, the recommendation system relies entirely on the company’s Grok-based transformer model to determine relevance from sequences of user engagement. The documentation states that there is no manual feature engineering for content relevance, meaning human reviewers do not directly adjust how relevance is calculated.
X said the automated approach reduces complexity in its data pipelines and serving infrastructure while allowing the system to learn directly from observed user behavior.
Transparency Claims And Platform History
X did not explain why it chose to release the new materials at this time. Musk has previously said he wants the platform to serve as an example of corporate transparency. When Twitter first released algorithm code in 2023, Musk said the disclosure would be uncomfortable initially but would lead to improvements in recommendation quality and help build user trust.
Since Musk acquired the company in 2022, some aspects of transparency have changed. The acquisition shifted Twitter from a public company to a private one. X did not publish its first transparency report until September 2024, after previously releasing multiple reports each year.
In December, European Union regulators fined X $140 million, citing violations of transparency obligations under the Digital Services Act. Regulators also said the platform’s verification check mark system made it harder for users to assess the authenticity of accounts.
Ongoing Scrutiny Of Grok And Content Practices
X has faced increased scrutiny in recent weeks related to how its chatbot, Grok, has been used. The California Attorney General’s office and members of Congress have examined claims that Grok was used to create and distribute sexualized images of women and minors.
The renewed release of algorithm details comes amid that scrutiny, adding to ongoing debate about the platform’s approach to openness and governance.
Featured image credits: Wikimedia Commons
For more stories like it, click the +Follow button at the top of this page to follow us.