Interesting discussion, though perhaps quickly moved beyond the level of the OP I think
My .02 is I am not sure there is a right or wrong way to look at this, workflows are different and often have different purposes. Personally, I choose to make all basic "global" adjustments in the raw files, then move to the editor (CS4 in my case) to make "local" adjustments. And often, those local adjustments include a final "global" curve, hue and/or saturation adjustment on top of them to balance effects from the local edits...
And... What is "accurate" color? If shooting a product, we can measure and confirm it. But if shooting a landscape, is it what we really saw or what we remember seeing, or what we wanted to see?
So, in the end I have to agree with the 16-bit TIFF comment -- I can almost always eek more out of a carefully converted raw in the image editor than I ever could directly from the raw itself. However, it does require first generating an "optimal" file from the raw that will take the further editing. Thus, my goal for my fine art prints is to use the raw converter to generate the optimal TIFF for further editing in CS, which is usually not the same settings as I would use for getting as optimal a final image as possible out of the raw processor...
PS: And my .02 on files sizes offered only FWIW: Drives and drive boxes are really cheap right now and setting up a 4 x 1TB RAID 5 array costs half of what a typical high-end DSLR zoom costs, so it should be well within the budget of any serious photographer. Call it $800 for 3TB of redundant storage, or about 26 cents per Gig. Assume 140MB converted size for for any current top-end 22+MP DSLRs 16-bit tiff, and you're at around 3-1/2 cents per image to save it that way.
Cheers,