[sjwright]

And more stress on the speaker/tweeter drivers.

Of course, when you look at these high resolution recordings, the amplitude of material above 20kHz is piddling. The amount of harmonics/overtones in acoustic instruments is minuscule in the first place.

[Arainach]

You're confusing frequency and resolution. Just because a set of music fits in a certain frequency range doesn't mean that all representations of it are equal. For example, consider this in graphical form:

https://i.imgur.com/ZgU1bgI.jpg https://i.imgur.com/RZpbyyI.jpg

Both have the same 24-bit color space. Both have the same blackest black and whitest white. Both have the same resolution. And yet one preserves more detail and information. This is the nature of lossy compression.

[sjwright]

I am not confusing the two. In a band-limited signal, the Shannon-Nyquist theorem mathematically proves that the sampling rate is the frequency resolution. It also proves that when a signal is band limited, discrete time sampling can be a zero-loss transformation.

Your analogy misunderstands audio signals. The resolution components of bit depth and lossy compression are different axes and should not be conflated with or analogised to frequency resolution. They behave very differently.