The counterpoint is that we already exploited a lot of the cases where simplifying the problem helps. We've spent approximately two centuries doing that. That leaves a lot of processes left over where that kind of specialization was either not feasible or not economically attractive