It's really not comparable since nobody knows the extreme outliers the algorithms produce, that are out there. For newspapers we know them.