Your argument is that nothing should be done about the absurd gains of the vertical part of a parieto distribution because differences in the long tail of the horizontal part exist.