My opinion is that you both need to reread a text on how light is measured and stored by a sensor. Since the sensor (measurement) is linear and the light scale is logarithmic, there is far more data (discernible gradations) in the shadow end of the spectrum than in the highlight end. This is precisely why Dlighting works better pulling detail out of the shadows than the highlights.
I have always shot digital in the same way I learned to shoot transparency film: expose to protect the highlights. If you blow the highlights they are gone. If you loose some shadow detail, some can be retrieved in post.
Noise is more obvious when pulling data from the shadows but this is due to the signal to noise ratio (charge from photons hitting the sensor vs. electrical noise) when the data is collected. The more photons, the higher the signal and the better the signal (photon charge) to (electrical) noise ratio and therefore less visual noise in the images.
The moral of the story is to make your exposure such that the histogram is used edge to edge without saturating at the far right and you'll have the maximum amount of data the raw file can hold.