The new X algorithm changes EVERYTHING. And most people still think they’re posting on 2021 Twitter. They’re not. xAI basically open-sourced the architecture behind the feed and if you actually read it carefully… X is no longer a social network. It’s becoming a behavioral prediction engine. The algorithm does NOT care if your post is “good”. It cares whether your post changes human behavior. That’s a completely different game. The system is likely tracking things like: • Did people pause scrolling? • Did they reread it? • Did they click your profile? • Did they reply? • Did they send it to friends? • Did they come back later? • Did they follow after seeing it? • Did they spend more time on the app because of YOU? That’s what’s being ranked. Not likes. Likes are cheap. The algorithm wants psychological grip. Which explains why some posts with mediocre likes suddenly get pushed to millions. Because the post created tension. Curiosity. Debate. Identity alignment. The best-performing posts aren’t “informative”. They create cognitive interruption. Posts like: “The economy isn’t crashing. It’s splitting into two realities.” “Most people aren’t unemployed. They’re algorithmically invisible.” “AI won’t replace programmers. It’ll replace mediocre thinking.” Notice the pattern? These posts force your brain to resolve tension. Your brain pauses. And pause time is EVERYTHING in modern recommendation systems. That 2-second hesitation before scrolling? That’s probably one of the most valuable signals on the platform. The craziest part: The algorithm also tracks NEGATIVE reactions. The repo literally references: • mute probability • block probability • report probability Meaning the system isn’t just asking: “Will this go viral?” It’s asking: “Will this person slowly damage the platform experience?” That means pure ragebait eventually gets suppressed. Which explains why so many engagement farmers explode… then disappear. The algorithm learned they create addiction without retention. Another thing most creators completely miss: The algorithm wants to UNDERSTAND you. If your account posts: • crypto one day • anime memes next day • politics after that • gym content later • AI threads tomorrow …you confuse the model. You become recommendation noise. The biggest creators often aren’t the smartest. They’re the clearest. The algorithm can instantly classify: • audience • topic • emotional style • engagement pattern • retention value Clear identity = easier distribution. This is why anonymous “main character” accounts grow absurdly fast. They create predictable emotional outcomes. At this point creators are basically becoming: part writer part psychologist part product designer part machine-learning input stream The people winning on X are not tweeting. They’re engineering attention loops for a recommendation model trained on human behavior. Most people are still optimizing for likes in a system optimizing for retention physics.