The thread title says it all...also feel free to participate if you have a strong reason not to record at sample rates higher than 96 kHz.
 

Well Craig, I have just been through the fire in regards to higher sample rates. I tried to record a solo acoustic guitar project with high end pres and mics so I recorded at 176 and 192KHz sample rates (for utmost quality). I have a Lynx Aurora Thunderbolt converter and a screaming computer but found that using higher sample rates forced me to set my buffer too high to 512 or 1024 (and this was before adding any CPU intensive plugins)! When I lowered my sample rate to 96KHz I could have a 256 buffer which gave me room for 2 more buffer increases if needed when I add lots of processing. Note: I  do the Archiving/Freezing thing after I've recorded and bounce my mic tracks to new tracks. I then hide the original mic tracks in case I need to go back to them later or start over etc. So in my experience with my computer system, 96 KHz is the sweet spot for me in terms of performance and still achieving high audio fidelity.

Do you think there is a difference in quality between 192 and 96 kHz?

I can record 2 ch's at 192 but I do not. I normally record at 48. I have seen no definitive proof that even 96 is recognizably superior to even 44. 192 vs 48 requires 4 times the disk space and 4 times the processor power. This means 1/4 the available plugins/effects. If your system can handle 16 instances of Kontact at 48 then it can handle 4 at 192. The only advantage currently is 192 yields 1/4 the latency of 48.
I see no real benefit in recording at 96+ and downsampling to 48 or 44 in order to mix. Most folks can get <= 10 ms latency even at 44.
If distribution methods other than CD or MP3 become popular I will move to that sample rate but I do not see that anytime soon.
I mean what is the point of 96+ when the distributed material is 128K MP3 with earbuds?

What I don't get is why no one seems to be recording at 88.2 kHz.  I don't because my current rig won't support it...but the downsampling from 96 k just puts you in the "bad math" zone.

Actually Craig, yes I can hear a difference between 96KHz and 192Khz on my Focal Be6 monitors. But it is VERY subtle. I actually can hear a bit more of the squeaks and slides of my acoustic guitar, yet this is not such a good thing. This whole issue of higher sample rates has gotten out of control. Yes higher sample rates DO give higher fidelity and this is why major orchestras and choirs demand to record at 192KHz (and some now at 384!). But it must be understood that they want to have a tremendous dynamic range and can afford multiple high end converters and computers connected daisy-chained via ethernet cables. My wife has commented that she could "smell" the rosin of a violin bow when it was recorded at 192 and not at 96. It is a trade off between higher fidelity and computer resources strain. Believe me, it's usually only people with average monitors that say they can't hear any fidelity difference from higher sample rates.

Dan. The "bad math zone" as you call it is now old school (and probably now a myth). Any modern computer with decent CPU power can handle a conversion from 96KHz down to 44Khz mathematically accurately. It is a "human" idea that it is more accurate to "halve" a sample rate e.g. 88.4 to 44.1 or 176.8 to 44.1 etc.  Math is math and any modern computer can compute the conversion from a non-multiple of 44.1 (192 or 96) down to 44.1. I've read research on this and will try to find it again and post it in this thread

dantarbill
but the downsampling from 96 k just puts you in the "bad math" zone.

No. In theory it might mean "harder math", but in the modern world it makes no difference because the CPU power is more than abundant to convert from any sample rate to any other without any artifacts when using a decent SRC routine (which Sonar most definitely has).

MiltonYes higher sample rates DO give higher fidelity

 
Not for recording/playback they don't. 
 
They give higher frequencies and potentially lower latencies. You can't get any "higher fidelity" once you've already captured everything.

 I find that it is better NOT to add dither when converting a higher bit depth (24bit) down to 16 bit using Sonar. I find it better to use Ozone's dithering to do this. I wonder if this also holds true in regards to converting down from a higher sample rate (96 etc.) to 44.1 as well?