1) Because Miller magnifies a small caps effects and makes it the dominant
pole in the response curve, eg. controlling phase margin. Stated another way
if no miller is done the parasitic poles due to stray C control the phase margin,
and leave us with pole locations close to axis with no + real damping loss.
eg. equal oscillation or partially damped oscillation. We could use a large
external C to modify response pole location, but then we have to supply large
current to slew that C, eg. to minimize OpAmp slew rate. But if we magnify the
small C, eg. miller effect, we only have to supply the unmagnified level of
current to slew that cap/gain stage.
2) The appendix has a good explanation....
Regards, Dana.