Twitter Releases Recommended Algorithm Code On GitHub, Asks Users To Find Lacks

JAKARTA - Twitter has just published its recommendation algorithm code on GitHub for everyone to see, as Elon Musk's CEO promised earlier this month.

The code is a legitimate step from the company to present a Recommendation algorithm on GitHub, not like the previous hacking incident.

The microblogging network stated that users need help to find deficiencies. Where this is an opportunity for the public to be able to provide recommendations to the platform.

Seperti bagaimana mereka ingin algorithm bekerja, dan menyesuaikannya dengan apa yang diterima secara umum, bukan apa yang ditawarkan perusahaan.

"Di GitHub, Anda akan menemukan dua reposit baru (repo utama, ml repo) yang mendiri kode sumber untuk banyak bagian Twitter, termasuk algorithm Rekomendasi kami, yang mengendali Tweet yang Anda lihat di timeline For You," ungkap Twitter dalam postingan blog.

"For this release, we aim for the highest possible level of transparency, while excluding any code that would jeopardize the security and privacy of users or the ability to protect our platform from bad actors, including undermining our efforts to combat child sexual exploitation and manipulation," he added.

In addition, algorithmic codes also do not include advertising recommendations, and to ensure user safety and privacy will be protected, Twitter does not release training data or model weights associated with current algorithms.

In a blog post, Twitter describes how the Recommendation algorithm works, including ratings and filtering.

Twitter's recommendation foundation is a series of core models and features that extract latent information from Tweets, users, and engagement data. The model aims to answer important questions about the Twitter network," Twitter said.

(Like) how likely will you be to interact with other users in the future? or, What are the communities on Twitter and Tweet what are the trends in it?. By answering this question accurately allows Twitter to provide more relevant recommendations," he continued.

In a discussion on Twitter Space discussing the move, Musk said he hoped users could find potential problems in the code and help make it better.

"Our initial release of the so-called algorithm will be very embarrassing and people will find a lot of errors, but we will fix them very quickly," Musk said.

However, Twitter users have also found interesting details in the code itself. As the application researcher Jane Manchun Wong obtained, where Twitter's algorithm specifically labels the Tweet maker Elon Musk.

Twitter’s algorithm specifically labels whether the Tweet author is Elon Musk“author_is_elon”besides the Democrat, Republican and “Power User” labelshttps://t.co/fhpBjdfifX pic.twitter.com/orCPvfMTb9

— Jane Manchun Wong (@wongmjane) March 31, 2023

According to Wong, this is the answer why Musk's tweets appear so often. He also noted that the algorithm has a label showing whether tweet writers are strong users and whether they are Republicans or Democrats.

When asked about the category on Twitter Space, a Twitter engineer later clarified and stated that it was only for the purpose of tracking the stat and had nothing to do with the algorithm.

He added that the label was meant to ensure users were unbiased against one group versus another. Unfortunately, he didn't discuss why Musk has his own category, as quoted by Engadget, Saturday, April 1.