tuga
Legal Alien
From an Information Theory POV 'dither' is expected to be 'randomised'. However a systematic form of dither has been tried in the past. e.g. the first NICAM ADCs that the BBC designed and used applied a half-LSB step up/down between alternating samples.
But so far as human perception is concerned, just-inaudible noises like hum, rumble, etc, might do something similar by affecting the hearing process. This is just conjecture though.
I think that background noise like you get with vinyl playback can help mask deficiencies elsewhere in the system that can become more taxing with say CD.
The same can perhaps be happening with harmonic and intermodulation distortion and channel bleeding.
But in a very transparent system my experience leads me to believe that noise detracts from realism. This can be subject to testing by comparing recordings from the early 60s with contemporary ones.