A few theories:
a) Noise is there you're not looking hard enough. As Helen said the image is nothing but noise, however that doesn't mean that the noise covers the entire tonal range. It will still only be in the order of a couple of bits worth of noise under reasonable conditions.
b) Most lossy compression algorithms favour bright red / green channels for information and dump as much data as possible from blue and black areas. This causes huge blockyness when using moderate compression and then after bumping the brightness up. The noise may be eliminated when the file is saved.
c) You may not be able to reproduce it. LCDs are imperfect beasts in a lot of ways, but all screens except top quality IPS panels suffer from lack of graduation in the dark making it often difficult to distinguish near blackness from absolute blackness, and also suffer from branding, sudden jumps in luminance which may mask the 1-2 unit difference in brightness, again most pronounced in the black areas.
d) You may not be picking up noise. Noise is a function of shutter speed, and other inaccuracies. It is reasonable to assume that at ISO100 at 1/1000th of a second you shouldn't really be picking up any noise in the blacks at all.
If no light hits the sensor, then it should not record noise.
Noise isn't caused by light, it just exists in inaccuracies of analogue digital conversion and random currents. Infact a measure of noise in a photodetector (not counting influence of any other circuitry) is called dark current because no light hits the sensor.
Why is this an issue anyways? digital noise isnt something i would want to produce, generally.
I hope when your kids ask you one day why the sky is blue you don't reply why is it an issue, you don't want a green sky generally. Where's your sense of discovery, learning and science
