Shows that scientific fraud such as the cobbling together of parts of different datasets to create a curve that suits your purpose is amply rewarded if you're doing it to foster the propagated narrative.