Months after Elon Musk promised monthly transparency updates for X’s algorithm, crypto users say the pledge has fallen short. The official xai-org repository on GitHub shows only one commit since the code was released in January, and promised developer notes never appeared. This has left many in the crypto community frustrated about how posts are ranked and why crypto-related content seems to get less visibility.
The Original Pledge and Slow Follow-Through
On January 10, Musk said the X algorithm code would be published within seven days and refreshed every four weeks with detailed release notes. The repository went live on January 17, but since then, no further commits have been made. The code, written mostly in Rust and Python, covers four components, but the lack of updates contradicts Musk’s earlier promise. Some users had pointed to the Following tab as a non-algorithmic alternative, but that hasn’t eased concerns.
A similar pattern emerged with the older Twitter algorithm release in 2023, where activity eventually slowed after initial criticism. Now, with X’s algorithm, the silence is louder because crypto users report weaker reach. Several have noted that posts about cryptocurrency appear less often in their feeds, replaced by politics, outrage-inducing content, and engagement bait. One market watcher, Ethan, said X is losing the community structure that once made it useful for crypto discussions.
Transparency Gaps in the Code
Before the repository launched, Ethereum co-founder Vitalik Buterin questioned whether X could provide enough detail for meaningful public review. The published code does include the final score formula, but it omits the weights attached to each predicted action. Without those weights, outsiders cannot fully assess how posts are ranked or why some content gains more traction than others.
Additionally, the Phoenix module’s README states that its transformer is representative of the internal model, except for specific scaling optimizations. Critics argue this means the public code differs from what actually runs on X’s servers. Crypto users have also raised concerns about negative signals, such as reports and blocks. The model could learn from these signals, potentially making coordinated bot activity a tool to suppress certain content. However, projects like Farcaster are often cited as alternatives because they publish forkable protocols rather than limited sample code.
Related: XRP Traders Watch Gaussian Retest as $8 Fib Target Emerges
