Hollinger will reflexively point at his favored stats rather than admit that maybe they don't accurately capture the value for all kinds of players. It's just as bad as being immediately hostile to numbers in the first place. And I saw someone on RealGM post some pretty convincing numbers that according to the normal metrics favored by Hollinger types that Rose is clearly rating out as a good defensive player this season--it's just that the second unit is historically good (literally one of the top defensive units in league history) makes him not look as good by plus-minus measures. Of course, the 2nd unit has the luxury of feasting on other second units and playing at full throttle for the short periods of time they're in the game.