Grep beats LSP? Why were OpenAI, Claude, Codex and self-referentiality

This is brilliant. I think it could stand to be a little more imaginative in their stories. It seems like it is the bitter lesson of the emergence of intelligence. It seems like the only possibility where you might be able to fully articulate it, but... What exactly is the fundamental difference that creates this subtle distinction?