On the paradox of n random variables