Venturing off topic along with you guys, I have found that in practice, it does not sound the same with 24bit to have maximized the bit depth where the peaks are hitting close to zero during the recording, versus running soft levels at the show peaking below -6 and then boosting them close to zero in post. Theoretically it should be true, but my experience has been that it's not.
I am not talking about just noise. I am talking about the overall quality of the signal itself. I experimented last year running one set low, peaks between -12 and -6, and the next set much hotter, peaks between -4 and zero. I boosted the first set in post up to match the second set that was left as it was recorded, dithered/resampled both to 16/44.1, and they were not identical sounding. The second set where the levels were optimal to begin with sounded much better to my ears, smoother, cleaner, or something along those lines.
Except in very select situations where the source is wildly dynamic, or in cases where the situation does not afford you the ability to easily access your deck to adjust, I think it's missing the mark to run soft levels when you are in an open taping situation. I sure don't want to walk out of a show with 16bits worth of information when I had a 24bit recorder. This is just my opinion, of course. To each his own..........