we can talk about bell's inequality later.

but, for now just chew this through - just because a statistical analysis that models a phenomenon as random works relatively well in deriving correct answers doesn't mean that that the phenomenon is truly random.