It does run contrary, given "pick a percentage number, preferably one that's above COLA, and just give everybody that" which explicitly doesn't (a) differentiate improvement from non-improvement; and (b) doesn't address significant salary increases due to professional growth and changing responsibilities.