Twitter: Elon Musk publishes recommendation algorithm code

While it’s true that Twitter has been under the microscope ever since Elon Musk took over the reins, the conversation has focused on the For You tab over the past few weeks. The recommendations in this section convince few. Either because it shows posts that are of no interest to the user, or because at some point they considered only showing tweets from people who follow Twitter Blue.

To try and calm down a community that is getting louder and louder due to sudden social media changes, Elon Musk has published the source code of the algorithm which, in fact, is responsible for the recommendations of publications. All files are available via GitHub.

The idea, obviously, is that the experts review the code to make sure that there is in fact no anomalous or manipulative activity as intended. But knowing how Elon Musk has changed since he took over Twitter, Can we really believe that this very code is implemented in a social network? Everyone will have their own opinion on this…

Steps Twitter takes before recommending a tweet

Elon Musk’s executives explain in general terms how their recommendation system works. First, it must be said that different services are involved with specific goals. One of them, for example, analyzes the popularity of tweets based on the user who posted them and the interactions they received (likeretweets, replies, etc.).

“Twitter’s recommendation algorithm is a set of services responsible for building and offering the Home Timeline.”

However, before showing it to you on the timeline, another system evaluate the possibility of your interaction with this account directly in the future.

In addition, Twitter analyzes the topics and communities in which you most often participate. If you are a person who constantly seeks and enters popular topics When it comes to movies, the algorithm analyzes trends in that topic and shows you posts that might be of interest to you.

The entire process described above is known as “Selection of Candidates”. Given this information, the Twitter algorithm proceeds to classify each tweet with a machine learning model. Those at the top of the rankings are more likely to be recommended.

But before we move on to the For You tab, there is one final step that needs to be completed. It’s about filter according to your preference. If you have blocked an account in the past whose posts you might be interested in, there is no point in recommending them to you. Tweets that contain content classified as NSFW (unacceptable for work) or that you have seen in the past may also be hidden.

Of course, this is a pretty simple explanation of how Twitter’s recommendation algorithm works in general. Although there is a whole network of information and rather complex analysis behind this.

You can be sure that over the next few days there will be people checking the code for some strange feature or drawing inspiration from it to implement in their projects. We have to trust that the code is the same as the one that runs on the Twitter servers.