To me, this monitoring issue highlights not only that this device is fundamentally flawed, but it also emphasises how mis-designed almost all the devices we discuss here are - IMHO.
Why is it that most of these devices have level controls in the digital domain, whereas almost all other digital recording systems do not? Isn't it significant that the first question people ask about any new piece of kit here is "what's the unity gain figure"? That's because in essence people don't want digital gain - it's (broadly speaking) the audio equivalent of the dreadful digital zoom on cameras/camcorders.
If you record straight into a PC with a proper soundcard or interface, you have analog input level controls on the interface (or none, if it has no mic inputs), and the digits are fed straight to your DAW software as is. Normally that software would not out-of-the-box provide for digital input level control, only monitoring / playback level control. Similarly, take something like the Korg D888 multitracker - you've got input trim per channel, and everything else (faders, pan, eq, etc etc) only affects monitoring and playback/mixdown.
All these correctly-implemented devices are so set up that analog clipping does not occur before digital clipping, or, digital and analog clipping happen at the same level. You set the record levels entirely with the input trim controls in the analog domain - there's no point in digitally amplifying on the way to the storage medium, as you can do that safely on playback/mixdown, and if you need to digitally attenuate, well, your analog input is clipping anyway, so that's where you deal with it.
Now in the case of the Edirol R-44, digital gain should be fixed at unity, and the four knobs with concentric rings should adjust the monitoring/playback levels and pan, and analog input trim should be implemented another way. Maybe by pressing and turning the channel knob, if there's not space for more knobs. Or in software, with the pushable playback level knob flipping the rest between trim and monitoring functions. Or with an assignable nudger like the zoom rocker on a camcorder. Or whatever, it's not hard.
Sure, you could argue that during recording you are merely acquiring data and all you need do it to ensure that levels are correct, and never mind what the overall sound is like at that point. But there are many scenarios where you do need to keep and ear on the overall sound, and while you could manipulate input levels to get a balance, personally I take the view that all channels should be adequately modulated and then attenuated on playback to achieve the correct relative balance.
So - am I right or am I wrong? Who here would be taken aback if digital level controls were done away with on their kit? Does anyone find them irreplaceably useful?