Elon Musk Open Sources “Dumb” X Algorithm

Unbelievable: He Struck Gold Again with a Fortune!

The moment the Github link went live, it felt like walking into a digital ghost town. Empty promises of transparency echoed in the code. You could almost hear the faint, sardonic laugh of Elon Musk himself bouncing off the server racks: “Yeah, it sucks.”

The Algorithm Dump: A New Era for X?

Think of those old treasure hunt movies, where the map is released one tiny piece at a time. Elon Musk appears to be making good on his promise to open up the X recommendations algorithm for public scrutiny. He teased this back in 2022, but the initial snapshot of code quickly became outdated. This latest release could be a real step toward open source—*if* the repository sees regular updates. Musk pledged updates every four weeks.

But even with regular updates, keep your expectations measured. Musk initially promised to release “all code used to determine what organic and advertising posts are recommended to users.” It feels like there’s still a ways to go.

Last November, Musk and Grok accounts indicated that Grok sorts posts on the “Following” feed by default (though users can toggle to chronological order). This algorithm seems absent from the current release. Plus, both “Following” and “For You” feeds include ads, supposedly served via another algorithm Musk vowed to make public. It seems like several more releases are needed.

What data does the X algorithm use?

Judging by the plain text readme files, the X algorithm update is attempting to refine the TikTok method for user retention. From what I can tell, it prioritizes engagement above all else, trying to pinpoint the posts that will stop you from scrolling. It draws from accounts you follow, and also from accounts deemed similar to those. The goal isn’t necessarily to inform, but to captivate. It wants to feed you whatever holds your attention.

Conflicting Signals: Pride vs. Disdain

I watched a street performer once, beaming with pride one minute, then cursing his act the next. Earlier on the same day Musk trashed the algorithm, X head of product Nikita Bier seemed proud, noting that daily engagement time from new users jumped from under 20 minutes to the mid-30s within six months. So, is it good, or is it bad?

The problem might be Musk’s perception of lingering “wokeness residue” from the Twitter era. His “it sucks” tweet responded to complaints that the algorithm penalizes posts from heavily blocked accounts, with some suspecting a bias against right-leaning accounts. This could also simply penalize accounts engaging in harassment and abuse. You decide.

How often will the X algorithm be updated?

Replying to complaints that the algorithm favors those who jump on news events, Musk stated that improvements would roll out monthly—hinting at the four-week GitHub code dump cadence. This open-source release is like a dam breaking. The flood of user feedback could reshape the platform’s future.

What programming language is the X algorithm written in?

Perhaps users with sharp ideas will dig into the code, identify the real issues, and offer suggestions to Musk. Perhaps the algorithm will evolve into something more rewarding and profitable. On the other hand, maybe the goals of a company focused on ad revenue and user engagement and the desires of informed, happy users are fundamentally incompatible. Maybe making the algorithm open source to satisfy both is a fool’s errand. Which outcome do you foresee?