Twitter announced on Friday that it has made the code responsible for the recommendation algorithm that its platform uses to determine the contents of users' timelines open source.
The code, which is now open source, does not include parts responsible for ad recommendations, or those that threaten Twitter's ability to keep in check attempts by threat actors to manipulate the platform.
Twitter said in a post on its blog: “For this release, we aimed to achieve the highest degree of transparency possible, while excluding any code that would compromise user security and privacy or the ability to protect our platform from bad actors, including: undermining our efforts to combat exploitation.” child sexual abuse and manipulation.
The company added: “Today's release also does not include the code that supports our ad recommendations. We have also taken additional steps to ensure user security and privacy are protected, including our decision not to release training data or model parameters associated with the Twitter algorithm at this point.”
Twitter has published two separate repositories on GitHub that contain the source code for its recommendation algorithm , and some of the machine learning models that run it.
As revealed by the company's engineering team, the tweets displayed in the For You timeline are selected through a service known as the Home Mixer, and it comes with tweets from various sources for recommendation in a process called candidate sourcing, and then It classifies each tweet using a machine learning model and then applies detection aids and filters such as blocked tweets, previously viewed tweets, and inappropriate tweets.
Twitter explained the process by saying: "For each request, we try to extract the top 1,500 tweets from a pool of hundreds of millions through these sources." "We find candidates from people you follow (within your network) and people you don't follow (outside your network)," she added.
The recommendation algorithm's goal for each user's For You timeline is to show 50 percent of the relevant and recent tweets from their followers, and the other 50 percent from people who aren't in their network based on what the user might find interesting.
It is noteworthy that Twitter recently removed the private source code and internal tools that were leaked on GitHub and have been available to the public for at least several months.
In the DMCA infringement notice , the company also asked GitHub to provide information regarding access history to the leaked code, and will likely find out who downloaded the code when it is available online.
It is noteworthy that Twitter's announcement came on Friday in the wake of tweets by its CEO (Elon Musk) in which he promised to make the Twitter algorithm publicly available.