Re: noise. Do you mean SEM, reliability, longitudinal studies, and the like? If so, yes. The first two are in the manuals. I know there's been some research on the second. Check out the Lothian birth cohort. Try googling "IQ regression to the mean" for some amusing (?) discussions on that topic from a statistical perspective.

Doing a quick survey, I did find that it appears available evidence suggests that high scores not only revert, they revert more than sub-mean scores (3-5 point on re-administration if IQ above 109, vs 1-2 if below 90, in a study where children were re-tested three years apart).

...pronounced like the long vowel and first letter of the alphabet...