Hmmm I see your point but it seems naive to assume that google can't see how many times a link has been shared on facebook. It seems like a pretty simple bit of code to hook into their API and return the total number of shares. Who's to say they wouldn't use this data?
Let's assume for a moment that there are T number of total shares, and X visible shares, while I invisible shares for a given link. We could say that T = X + I.
Further more if we assume that each share on facebook has an associated Page authority, lets call it Pa. And let's also assume that each page that is invisible to google has an unknown page authority Po. Then it would follow that the total link juice passed by a facebook share would be something like:
Link Juice = SUM-of-Pa-from-0-to-X + SUM-of-Po-from-X-to-T
If we were to assume that on average Po was equal to the average page authority of the visible share pages, then we could conclude that
Po = (Sum-of-Pa-from-0-to-X) / X = PAverage
So you'de get: Link Juice = SUM-of-Pa-from-0-to-X + PAverage*(T-X)
Since they are capable of caluculating both of these quantities (all known variables) then i think it follows they they would be able to calculate the rough value of a link shared via facebook.
If i can do this pseudo math on the back of a napkin, who's to say that Google's engineers couldn't figure this one out?
So can we still safely assume that google treats facebook shares the same way they treat other links? I highly doubt it. Espcially because of the fact that most of them, like you say, are nofollow, yet despite their nofollow and invisibility, they still effect rankings (at least they do in my experience).