That's how i understood GP's "half of such groups" statement - as a 50-50 chance (half of all groups, not half of the people in the group). But maybe I was biased because I heard of "Birthday Paradox" before.