The LLM Trilogy

Signal in the noise

A three-part series examining the AI hype machine from the inside. What's actually breaking, what's being destroyed, and why the core problem will never be fixed — no matter how many billions get poured into the next funding round.

Total reading time: ~44 minutes. Best consumed with coffee and a healthy distrust of LinkedIn prophets (more likely bots).

Part I — The Problem
Vibe Coding: A Monkey Behind the Controls
Why AI assistants make terrible architects. Cognitive debt, the procurement manager problem, and what happens when the person evaluating AI output understands less than the AI "understands" itself.
READ →
Part II — The Fallout
The Open Source Trap: How LLMs Are Eating the Web Ecosystem Alive
The open source motivation collapse, the web ecosystem death spiral, and why the future might belong to languages that LLMs can't touch. Flash was right. C is patient.
READ →
Part III — The Ceiling
The Shape Is Perfect. It Still Tastes Like Shit.
Hallucination isn't a bug — it's how pattern matching works. The $10 billion scaffolding illusion, and why the entire business model depends on pretending the next version will fix it.
READ →

"The question isn't whether LLMs hallucinate.
The question is how long the industry can keep pretending they'll stop."