as an example of an amateur getting gang raped by a GM down a rook:
A relevant quote from the comments section on that page:
"keypusher: <tatarch> A chess club in Manhattan had an odds tournament many years ago, which Andrew Soltis wrote a column about. If the rating difference was over 600 points, the stronger player gave odds of a rook. (There were lesser odds for smaller differentials in playing strength.) The odds-givers lost only one game out of about 40, when what Soltis described as an underrated A-player beat Arthur Bisguier at rook odds. So, I would guess an 1800 player would lose to a GM at rook odds most of the time."
Id like a little more verified proof before I go off of anything that article says. Who knows, maybe the time control was game 10 or 15 or something like that? Maybe Soltis simply has the facts wrong etc etc