Aah, but that's the rub. The result isn't actually random, it is only "computationally indistinguishable" (or, some lower threshold, since it is only in the last few decades of the multi-millennia history of crypto that this has been true). I'm not denying that pseudo-randomness (or even in very limited cases, real randomness) is used in most varients of crypto today, rather, I am suggesting that despite this, the end result is precisely the opposite: a message so highly ordered that it is "computationally indistinguishable", merely masquerading as random.