Here's an article that was written quite sometime ago. I think in overall a lot of it's content basically supports what a lot of us are saying. I'm sure there are lots of other good articles out there ; however this one is written in plain English as in oppose to gibberish that doesn't make sense to us mortals.

"... VS: Most analysts feel that DVD is a sure thing, not only as a replacement for tape as a delivery medium for film, but also as an enhanced audio format. However, audio facility owners buying equipment today are purchasing with the understanding that DVD is not ubiquitous today. How much of a factor should posting audio to specs higher than the current Red Book (CD) standard be? Is archiving to the assumed standards of tomorrow important?
Elen: The Red Book standard (44.1kHz sampling and 16-bit word lengths) was established in the early '80s. We have been able to do better for many years. Today, there are several consumer distribution media that offer quality superior to Red Book, notably DVD and HDTV. DVD in particular offers 20-bit and 24-bit word lengths, and the DVD-Audio specification allows for multichannel recordings at 96kHz sampling and stereo up to 192kHz. Very likely, video productions created today are going to be released on media that offer higher quality than Red Book. In addition, it is important for production facilities to operate at a higher level of quality than the consumer distribution media because higher production standards offer a degree of headroom that can give you more flexibility when re-purposing the content for future release configurations.

Pirali: The answer depends largely on the format you're mixing down to, the kind of music, and the taste of the producer. If you're recording down to a CD, then there really is no need to track at anything larger than 16-bit. Mixing down is a different story because you're summing multiple streams of data there. You'll probably want to mix at higher resolution to preserve everything in the input streams because digital mixing is more of a mathematical process than electrical as in an analog system.
Recording at higher bit rates [greater than 16-bit] provides better dynamic range at the source, allowing for finer detail and more headroom during recording without losing data. Ultimately, the final mix is heard most often as 16-bits, but for most of the process [recording to final mastering], it is desirable to use 24-bit dynamic range, with even higher bit counts for intermediate results during summing and processing. This is especially important when you consider that the level of signal may be lowered and increased in various stages before the final summation and that it is typical to compress the dynamic range of the final mix before truncation to 16-bit. Because of noise levels, amplifier, and the quality of consumer converters, most consumers do not listen to the final results in environments that allow the full dynamic range of a 16-bit recording, so higher bit resolution is not needed.
Braksick: I agree. Certainly the effect of 16-bit wordlengths is easily demonstrated, and I think it is pretty well accepted that engineers want to conserve as much resolution as possible during the recording and mixing process, even if the final deliverable is a 16-bit, 44.1kHz CD. As far as the higher sample rates, a lot of people are just beginning to explore that. If DVD-Audio gets off the ground with consumers, it'll become common. The question that always arises, however, is whether you want to spend your CPU or DSP cycles on running a bunch of tracks and plug-ins at 48kHz, or half as many tracks and plug-ins at 96kHz.
The thing that has really changed is the ubiquity of digital I/O. People now tend to believe that it is a good idea to avoid multiple A/D and D/A conversions and keep the connections between components digital whenever possible.
O'Neill: 24-bit conversion offers real-world, audible improvement over 16-bit or 20-bit. Sampling rates above 96kHz have a less audible improvement but reduce the cost and complexity of the digital filters. Because digital audio can be preserved or published only as well as the initial conversion, it makes sense to capture that audio and perform all processing and intermediate storage at the maximum resolution in terms of both signal level and sample rate. For audio post, use of these higher conversion standards allows music recorded today to be published at higher standards than the current CD RedBook, both now and in the future. "