i agree that you can't learn a lot from simply looking at the waveforms. for example, it tells you nothing of the signal in relation to the noise floor. in which case the smaller waveform could be the inferior one.
however the methodology isn't completley invalid.
this image tells the story. the squared off tips of the peaks means clipping, pure and simple. a non-clipped signal doesn't square off.
so in this case, the point is valid. but you're right, you can't make a purist argument against louder mastering simply by comparing the size of waveforms alone.