not to argue with Mike, but i use an echo audiofire interface, cool, not the lowest of round trip latency. now, i use guitar rig 4 all the time, and my signal chain is guitar>soundcraft mixer>audiofire>pc>sonar>guitar rig>sonar>pc>audiofire>mixer>monitors. i run at 128 samples, around 2.9 secs reported latency, but roundtrip is more like 11ms, and i can easily play guitar in real-time...also soft-synths. of course, sonar delay compensates for the playback audio, but even so, if this round-trip was such an issue i'd be in trouble, huh?

if sound travels approximately 1 ft/ms then standing 10 feet from a drum kit would give me 10 ms delay alone? how do bands ever play?!

yeah thanks for taking that the right way Mike, i know from experience some people are a lot more aware of latency than others, and i know that as an engineer (which you certainly are) it's your job to pinpoint the technical nuances of a mix, and certainly detecting off-timing is a part of that. i don't know, but has the introduction of ad/da conversion using computers increased, decreased or not affected latency when compared to using hardware channels/comps/eq etc and tape?

It's easy to compare the feel of playing thru 5-6ms vs. playing thru 12-18ms Round-Trip-Latency.
If you have an audio interface that delivers low RTL (5-6ms at a 64-sample ASIO buffer size).
Use software based input monitoring to listen to the feel at 5-6ms. 
Now... insert a delay plugin (100% wet) that adds 6ms (simulating 12ms RTL).
Finally, increase the delay to 12ms (simultating 18ms RTL).
Compare the feel of playing thru all the above.
Report back which feels best.
 
Anyone *can* play thru 10+ms RTL...
But once you've experienced significantly lower RTL, and how much tighter the timing feels, you're never going to go back.  Also, you need to note that when playing soft-synths, some are slightly latent.
Thus, if you're playing back at 6ms one-way (playback) latency... and the soft-synth adds 2-3ms (or more) of extra latency... you're getting close to the 10ms mark.
Playing Kontakt samples at 10ms latency definitely feels sluggish.
Much more so than playing a Motif, M3, Fantom G, etc...
 
Take the OrangeTree Guitar samples for example...
Excellent library... very advanced use of scripts (Good stuff and loads of fun)
Setup Kontakt to playback at a 64-sample ASIO buffer size.
Play some fast runs and you can feel the latency caused by the scripting.
Now imagine how it feels playing the library (or similar) starting with double/quadruple the playback latency.  It's going to feel very sluggish.
 
As native DAWs have progressed (speed/power), the issue of low round-trip latency has become more significant.  10 years ago, effective monitoring thru (native) software processing was not possible.
Thus, we were happy to use workarounds (monitoring straight off the console, etc)
Today, we have the processing power (and audio interfaces) to allow effective monitoring thru native software processing.  With the plethora of excellent sounding plugins (like Ampeg SVX), some folks are going to want this capability.  There are inexpensive options that provide low round-trip-latency... as well as more expensive/elaborate units.  Make a good choice... and you're set.  Make a bad choice... and you either live with high RTL... or you buy another audio interface.
 
 

My eyes are peeled for USB 3.0 interfaces.  But I'm wondering how the new spec labeling will affect production? From a purely speculative, theoretical standpoint, how long might it be before we even see a USB 3 Audio interface? Or will it be more a case of being able to maximize the performance of current USB 2.0 gear? Probably more the latter to start. Manufacturers will now have to prove they can deliver the theoretical 5GBPS... Hell, half that would be amazing.  It would be interesting to hear  thoughts on where this is all heading as USB 3.0 gets set to roll out.

I think that's beside the point though. Manufacturers will take advantage of the new protocol in some form or fashion... even if it's merely being able to maximize the performance of existing 2.0 interfaces. I realize we don't need to access 5GBPS or even half that... but I can't see the industry not taking advantage of the newer protocol. And I also wonder what this might mean for Firewire.