Twitter on Friday open-sourced most of the code running its recommendation algorithm on its main "For You" feed.
Much of the source code was published in two GitHub repositories, along with some of the AI models powering the algorithm.
- In a blog post today, Twitter said the move was "taking the first step in a new era of transparency."
- According to the post, its recommendation pipeline has three main stages:
- It first gathers "the best Tweets from different recommendation sources."
- Then it uses a machine learning model to rank the tweets.
- Finally, it filters out tweets that are NSFW, those a user has already seen, and tweets from accounts that a user has blocked.
- CEO
Elon Musk discussed the move in a Twitter Space today, saying he hopes
people could use the code to unearth "issues" and improve it.
- "Our
initial release of the so-called algorithm is going to be quite
embarrassing, and people are gonna find a lot of mistakes, but we're
going to fix them very quickly," Musk said.
- Separately, portions of Twitter's proprietary source code were leaked on GitHub earlier this month and later taken down.
- A federal court has now ordered GitHub to identify the user who posted that code.
- Twitter executives believe that the leaker was a former employee who left the company last year, sources told The New York Times.
