Great insight; I was going to theorize something about objective vs subjective experiences, in that you can't trust a human to make an objective observation about e.g. sound quality. Best thing you can do with human listening tests is gather data and make it about statistics.

But it's never about just what's objectively or even statistically best. Certain modern music is optimized for being played and recognisable on crappy phone speakers or low quality bluetooth ones. And then there's the loudness wars, where the music was objectively of lower quality (due to clipping etc), but sold better due to psychological effects. or whatever it was about.