I share Linus Torvalds' view on AI for programming: it is a tool, a very useful one at that, but like all tools it needs someone who knows how to use it to use it effectively.

This response is approaching AI as a technology, not "Big AI" and it's miriad of ethical issues. A purely technical viewpoint.

There is ambiguity on this blog post as to whether Claude wrote the whole thing from scratch or it took an assistive role; I have nothing against purely agentic coding, or vibe-coding, but for something as crucial as your display server, I don't trust even the leading frontier models for that yet.

This person's concept of developing, or at least designing (as is my preferred term when referring to agentically-written programs), all their software is very cool and has inspired me to contemplate some of my own tools. But again, I don't think I'll be writing my own X server or Wayland compositor anytime soon.