With respect to USB AC I don't agree, that is supported natively in nearly all computers and many portable devices. I have seen a much higher rate of driver difficulties over the years with Firewire interfaces, for example. I've never had a problem with a USB AC device; a USB device with its own flaky drivers, sure I believe that, but USB interfaces would potentially have the same difficulties. Again, if the physical circuit components are the same--which they largely are--why would there be a difference?
Then you're lucky. USB AC 1.0 sucks. Badly. Among other things, it provides no information about how different timing sources (sample rates) are related. If the input changes sample rates, should the output change? Maybe? Probably? The driver has to guess. This means that if you have a USB AC 1.0 device with a sample rate switch/button instead of it being entirely driven from the software side, your odds of problems just went up to approximately guaranteed.
And a lot of USB AC 1.0 firmware sucks, too. The spec says that the number of samples per frame must be within -0/+1 samples between consecutive frames. The reason for this is that audio drivers have to guess what to do when the aforementioned devices change sample rates. Getting too many samples by a couple of samples per frame (or too few by one or more) usually hints that the device just switched from 44.1 to 48 or vice versa. At this point, the OS has to guess how to handle it, and assumes that the same rate changed. If buggy device firmware causes this to occur when a sample rate change is not happening, then that device basically has to be hard-coded into the driver as a workaround. Unfortunately, this isn't uncommon with cheap silicon.
I've had long conversations with USB audio driver engineers on the subject, and the general consensus is that everything should move to USB AC 2.0 descriptors as soon as possible, and should provide 1.0 descriptors only as a fallback for Windows. USB AC 2.0 just makes life a lot easier.
As for USB mics in general, my biggest problem with them is that *most* people will eventually need to do at minimum stereo recording, and in most cases, dual-mono recording (e.g. for acoustic guitar). When they realize that need, if they started out with a USB mic, they end up buying the interface twice. That doesn't apply to everybody, but it's common enough that USB mics set off red flags for me.
Another big problem I have with them is that there's no real room for trying out different mics. Every time you buy a new one, you're buying not just the mic, but also the interface. If both the mic and the interface are of high quality, that cost adds up a lot faster than buying just the mic each time. And if the cost doesn't add up fast, odds are good that corners were cut along the way.... More importantly, you have no way to know whether sound quality problems are caused by the mic or the interface. If one piece of the chain sucks, the whole chain sucks. This makes it hard to upgrade later.
Also, there's no real room for modding. Most of the mics I tend to pick, with few exceptions, are mics that you can then improve (or at least change) by modding. I've never seen a USB mic that wasn't pretty much wall-to-wall surface mount electronics. That doesn't lend itself to modding unless you're good enough to design your own boards, in which case you probably also can afford to buy a traditional interface.
Finally, USB (at least 1.0) doesn't provide much of a power budget. You'll never, for example, see a USB-powered tube mic (or at least not anything based off a 12A*7 tube). Lots of interesting microphone designs just plain aren't possible. And you'll probably never see a ribbon mic, either, because you'll never get the analog noise floor low enough. So it is pretty limiting in a lot of ways.
That said, if you really are interested exclusively in vocal recording, have no interest in modding, have the opportunity to try out a bunch of mics in a store, have a good enough ear to pick the best choice on the first try, and are careful to buy a USB mic that has a built-in headphone jack (without which real-time monitoring isn't likely to work very well), I suppose they're an okay choice. That's not most people, though.
Nevermind, ghost of Jobs, go back to blowing up Chinese workers . . . )
