Thanks for all of the great thoughts. Firstly most analogy is specious
I see a lot of mention of 'latency.' In another thread we see struggles with latency issues in a USB setup. USB hubs almost guarantee latency issues and unless you're lucky enough to have a deck bristling with USB ports, the issue looms. My personal experience is that I get more latency from a USB connection than I do from a line in, at least to that point in processing. That's a no-brainer; a line level signal on a wire is going to get there faster than USB, not slower. Depending to some degree upon where 'there' is, I suppose.
Keeping to the point though, we are talking about 2 track signal transport to the DAW, and whether using a line in is notably inferior in this regard. It is logical fallacy to bring 4-tracking into the equation at all. There are plenty of 2-track USB external interfaces out there, and this is the realm of expectation that I also work in, at this point. So a 2 channel USB device is going to have the same problems mixing that I will have, that are described in the thread. Yet, many are happy enough with 2 tracking with an external audio interface, as I am as well with my internal audio chipset. For singer/songwriters, 2 tracking is an okay solution.
About money spent, etc., keep in mind that I've never spent any money on an external audio interface so my investments in boards have already happened for other reasons. If I am taking in multiple channels on a board and then recording that to a single channel, that is the fixed mix of that channel, yes, and generally that is two mics for a single vocal track, or it could be two mics on a guitar. If those aren't right to start with, then I get what I get. I might have been able to fix it in the DAW had the inputs been split each to a track, or I might not have. It's all work and I would have been better off getting it right before I started recording/singing/playing the song.
To say that I am stuck with pre-mixed levels on tracks though, is not correct; I still have two tracks and can set up automation, reduce noise, apply compression, limiting, effects, plugins, dupe the tracks, all of that. In Garageband I can put live effects on the tracks and monitor either the raw input at the board (zero latency) or I can monitor the Garageband effect-laden capture (little appreciable latency even then). Now if I am 2-track recording whilst playing a keyboard controller, there I can see some latency surfacing in the software instrument (generally it will cut out entirely though if I keep playing, it's always recorded okay, for some reason). Now, if I had added an external USB audio interface and then tried 2-track recording whilst playing that keyboard controller, also USB, what would happen then with latency in my USB bus? It would likely hose - but I've never done it, so don't know. I can monitor my mix in Adobe Audition CS6 while singing into two mics as two more tracks, then add that as two more mono tracks into Adobe Audition. In that case the transport is the DR-05; I transfer the files via USB storage connection, align the vocal manually then continue. But, none of this procedure stuff is relevant to the question of signal transport.
My setup is nothing to brag about, and with many limitations. Those are facts. The larger point here is that my line in works just fine taking signal from the berrys. I don't see the latency issues that keep coming up here as complaints found in using an internal sound card. Where would they be surfacing, exactly? And if you're using live effects in your DAW whilst recording, how's the scarlett or whatever gonna help you with that? The effects run in the scarlett itself?
Finally ec33 points out that my mac line in is -10dB, I've read that is the case with some line in's, though there are balanced/unbalanced line in's, but I see no issue with an unbalanced signal at this point in the chain (output of the berry). I have several outputs that I can pull from; line out, tape out, mains out at line level so there's flexibility there. This however is the closest semblance of a reason to switch to a USB interface, that I have seen yet in this thread. I will look into this further. I do have a special cable that attenuates, that I use to go from the UB802 to the DR-05's line in; I think that this cable loses 10dB. Its purpose is to connect the DR-05 to DSLR's I have. Anyhow I believe that the berry pmp4000 is giving the mac line in just the signal that it expects.
My tentative conclusion here is, no compelling reason has been presented thus far for me to migrate to an external USB interface. Without evidence to support any other conclusion, I am still at EXT USB INT < necessary for 2-tracking. The caveat that I have to add now is, 'when one has a board.'
And I can't function without a board. If I had an external USB interface it would have to connect to an output of a board, otherwise I would be plugging stuff all of the time.
Thanks again for all of the replies, I appreciate it!