> the halting problem can just be figured out experimentally in production That's a wild statement