Hi Hal,
I would put it that delta-sigma is a **less direct** way of accomplishing D/A conversion than R2R. However, the results of BOTH approaches are approximations, in the sense that what comes out does not correspond to what goes in to an infinitely precise degree. And of course the same goes for everything else in the recording and reproduction chain, both theoretically and in terms of the various practical implementations, from the recording microphones to the speakers in our listening rooms and everything in between. Even the
Nyquist-Shannon sampling theorem, which underlies A/D conversion, D/A conversion, and essentially all digital signal processing, is an approximation in any real world implementation. (For example, note in that Wikipedia page the various equations involving integrals or summations that are taken from minus infinity to plus infinity).
So in that sense **everything** in the recording and reproduction chain is lossy, to some degree. Which of course tells us nothing whatsoever about which design approaches and which practical implementations of those approaches are to be preferred. But my perception and belief, as evidenced in part by various posts from highly experienced audiophiles in this very thread, has been that depending on the specific implementation both approaches can provide results which are satisfactory to many, at pretty much all price points that can be considered to be audiophile-oriented.
It’s as simple as that, IMO.
Best regards,
-- Al