Jim, I am not a recording engineer, so I defer to the engineer that I hire. His feeling (and certainly there is disagreement on this point) that one should record at the highest density possible for a number of reasons:a) it may keep your options open for transferring to future media--say if DVD-A became the norm (doubtful,s but clearly the Redbook standard won't last; b) you can never add resolution, so always start with the best source; c) if you are doing much processing in the digital domain, it will be cleaner when peformed at the higher sample rates; d) since I'm recording acoustic music (improvised jazz) he finds it captures the real time/space nuance of the event as well as a truer tone of the instruments.
I have made 5 cd's to date. The first was recorded to a Tascam D-88 digital tape machine through a device that doubled the sample rate to 88.2. The second was recorded to the Masterlink at 24/96. The third was recorded to half inch analog tape (all of these projects were recorded straight to two-track stereo) with the Masterlink as a safety at 24/96 (it may have been at 88.2, now that I think about it.) The fourth was the only one we have multi-tracked, and it was recorded to ProTools at 24/192 and mixed to the new two track Tascam DSD recorder (it has no hard drive--it records to DVD-R's. And the session I just did this week was recorded straight to two track DSD, with the Masterlink at 24/96 as a back-up. Actually we use two Masterlinks--one is mine and one is the studio's. With digital, redundency is a must. And I like having it on the Masterlink so I can listen in my system to evaluate the tracks and work out sequences for the cds. I believe we used Apogee converters for all the digital formats, except of course the DSD.
When the cd is mastered is goes into the Sonic Solutions mastering suite for final sequencing, editing, and level matching. In the projects I have been involved in, the minimal signal process we have done has been analog (minimal eq, usually with a Manley Massive Passive equalizer, and perhaps a tiny bit of compression with a tube compressor) so this is another reason for maintaining high resolution. The data is only dithered down to 16/44.1 in the last step, when it prints to cd-r. I don't have the experience to say this is the best way (and I'm sure nothing is the best way in all circumstances) but I trust the people I work with (my mastering engineer is John Fischbach, who was the lead engineer on Stevie Wonder's Songs in the Key of Life, so he has a wealth of experience to draw on.) And I have been very happy with the results.
The analog tape probably sounds the best, but the cost and inconvenience make it difficult to work with. What attracts me most to tape, however, is knowing that it won't be any more obsolete than it is now. There may no machines to play it back on, but at least the concept will remain viable. Digital is a big leap of faith . . .