I would put it this way:
You simply send a kind of “level reduction command” to a specially designed and programmed processor that calculates with up to incredible 96 bits (!!!) of processing depth, thus realizing the desired channel level in the highly complex overall calculation with virtually no loss of quality.
(Please correct me somebody if this is phrased inappropriately.)
Exactly, and that is precisely the great advantage (of modern digital technology).
This wouldn’t be possible in a comparable way with analog, because at low signal levels you would increasingly approach the magnitude of interference voltages, such as those that would come from noisy components, which could degrade the signal.
The term degradation is ok, but interferences wouldn’t be the correct term.
I guess it is distortion and noise that degrades the signal.
That isn’t the same as bit-depth. I’m sure you know that.
If you lower the fader from a channel, the output will lower the volume.
Lower volume, means lower bit depth.
With 18 dB headroom you have around three bits that are not used for the full dynamic range. Every bit less, reduces the dynamic range by 6dB. That is simplified for better understanding.
24bit at the output means a dynamic range of around 144dB.
The only time I’m concerned about getting “proper gain structure” on my inputs is if I’m recording multitracks of the show and I don’t want to end up with tiny levels on my tracks. The dynamic range and s/n ratio of these desks is so high that you don’t really have to worry about it.
If you want “proper” input levels, the correct answer is you have to reduce gain somewhere late in the chain: master, matrix, or amp/ processing inputs.