Comment by dkislyuk

(meant to say, scale-invariance of probability ratios, or shift-invariance of the inputs)