I agree that SLOP should be used whenever possible. Of course all offsets should be cancelled when aproaching the first waypoint of a published procedure.
Rather than building in a fixed offset perhaps a random innaccuracy when operating in RNP > 1 airspace?
Just a thought.