Perhaps we could create a placebo bureaucracy, say an FDA like agency which just approved anything that came through it. And then we could compare the end results of that agency with the FDA to see which one approved the most 'actually useful' medicines and the one which had the most 'later found to be harmful' medicines.