Thanks for the thumbs-up.
The 44/88 thing is like you remember: If you convert from 88 to 44, for instance, the conversion is really just a matter of the DAW using every other sample of the waveform. On the other hand, 96 to 44 requires gearboxing, which degrades things a bit. I never really thought it was that bad, to be honest. But if I would still stick with the 96k → 48k and 88.2k → 44.1k conversions, like you said.
The difference between 24 bit and 16 bit is huge in production. Everyone just figures “16 bit = CD quality,” which is totally true. But a finished master is at a maxed-out level, and it’s not getting altered after that. Think of it with the “poster” analogy from the post above, and figure that the master limiting on a “loud” CD mix puts ANOTHER 5-10 dB onto the mix (while in mastering, still at 24-bit), using “brickwall limiting” techniques (thank you, loudness wars! 
). And the master limiters add dithering/noise shaping, too, so you’re really getting some information over and above what a master would have if it was only truncated from 24-bit to 16-bit. 16-bit is great for CD’s, but I would really say it’s an inherent compromise for recording, unless you are a talented engineer with GREAT gear who can REALLY manage the dynamic range on the way in and commit to more level than you’d have to in a 24-bit situation. That’s the thing: everyone figures, “Alanis was recorded on blackface (16-bit) ADAT’s, and that doesn’t sound half bad.” Sure, and it was recorded with a vintage AKG C-12 by a guy who studied under Quincy and Bruce Swedien, and it was done on a Euphonix, and I think the A/D converters were upgraded, too. I still wouldn’t want to be staring at those meters thinking, “if it falls below ___, it’s gonna suck in the end. And if it goes above 0 dBFS, it’s gonna get scorched.” Each single bit DOUBLES your resolution. Think about it that way: 17-bit would be TWICE 16-bit, 18-bit would be FOUR times 16-bit, etc.
Absolutely check your DAW and keep it in 24-bit (or more) on all sessions. I’ve gotten sessions to mix which were in 16-bit, and I’d be hitting a wall for an hour or two wondering why everything sounds so plastic-y when I start to use some processing. Just doing “Save Session Copy” in 24-bit to change over the mix engine would improve things drastically, as far as giving me the effective option to do some aggressive EQ’s. The trap on Pro Tools was always that, when you start a New Session, it asks if you want to use “Current Setup” or whatever – that means using the setup from whatever session you opened last. That’s bad news, if your previous session was from some demo studio with a crazy I/O setup. Next thing you know, you’re flipping through your buses going, “’‘Multiverb?’ it’s been DECADES since I used a Multiverb…” Just check “Stereo Mix” and check your options each time. That will keep you from accidentally using Chucky’s Demo Factory’s sample rate, too (this is a PITA with external clocks/converters, when you have to switch them every time you open the ONE session that’s at 48k or whatever).
