You should see all the misallocation that was put towards valve-based computing. Why didn't they just all arrive at the correct answer without investing in discovery first?