Twitter has kept up with its CEO Elon Musk’s promise and released its algorithm on GitHub, providing transparency to users about how it determines which tweets appear on their timeline. The algorithm goes through three stages, gathering tweets from various sources, ranking them through a machine learning model, and then filtering them before appearing on the timeline. Twitter’s blog post also gives additional information about the algorithm’s process, including optimizing for positive engagement and ensuring users don’t see too many tweets from the same person.
Musk has promised to make the algorithm the least gameable system on the internet and as robust as Linux, which is one of the most successful open-source projects. He aims to maximize unregretted user minutes, and by releasing the code, he hopes independent third parties can determine with reasonable accuracy what will be shown to users.
However, Musk has also prepared users for potential disappointment, citing the algorithm as overly complex and not fully understood internally. Nonetheless, he promises to fix any issues discovered through code transparency. While Twitter has released the code for transparency, it still needs to earn the open-source label. This involves deciding which pull requests to approve, what user-raised issues deserve attention, and how to stop bad actors from sabotaging the code for their purposes.
The move to increase transparency is not happening in a bubble, as Musk has been critical of Twitter’s previous management and their handling of moderation and recommendations. Since he took over, he has faced backlash from users annoyed with the For You page shoving his tweets in their faces and conservative boosters increasingly concerned with their lack of engagement. Musk has argued that the algorithm deboosts negative and hate content, but analysts without access to the code have disputed his claims.