Harry Waterman comments on Differential acceleration of alignment-relevant capabilities is a bad bet