@PolyWolf So this system has worked like this for a looong time. It used to be based on a manually trained model that worked similar to spam detectors. Before that it was manual. It only switched to LLMs a few years back.

The real issue is that, much like with debian, kernel developers don't see themselves as collaborating on a greater whole, they see themselves as rulers of their own small fiefdoms, which creates constant adversarial relationships between different areas of the project.