Every time I'm about to jump into one of these discussions, Fons says
pretty much what I was going to say. But there is one additional
textbook observation I was going to make:
a) any non-linearity introduces harmonics, some
non-linearities
introduce an infinite amount of harmonics, which will cause foldover
distortion. the large the sampling-rate, the lower the foldover.
Although this is mostly true, it's difficult to believe it matters in
a practical audio application. If it does, in most of the cases that
are not deliberately using a highly nonlinear transfer, adding one
additional bit of depth will get you the same noise/distortion benefit
as doubling the sampling rate. Once you're using 32 bit floats,
you're already 160dB down... If you need more, I'd love to know what
you're doing, and I'm not being sarcastic!
Monty