I'll just weigh in that having read many of the posts, there are lots of things being confused here.
1. Confusing Upsampling vs magically creating more bit depth2. Confusing why one upsamples (its not to get more info)3. A vague notion that digital is suddenly good. Some of it has been good for ages, much still is awful, and most of that from the studio, and in Rock/pop4. Confusing chips with implementation. Differences are in the latter. Gimme some of the old 18/20 bit R2R DACs any day. Hell, don't give them to me, but please sell them to me. :-)
5. And just to make my point, it still doesnt matter because i can make both PDM and R2R sound good or sound bad. (note: PDM = 1 bit = sigma delta, it works like your fuel injection)
But yes, overall things are improving, will continue, and red book is good enough But it does make all our engineering lives very difficult due to the proximity of the sampling rate to the nyquist limit. Give us some filter slop room, please! Why does auto-correct want ot make "Nyquist" = NyQuil?
G
ps: i have info on some of this on my blog at sonogyresearch.com