Reports that say that something hasn't happened are always interesting to me, because as we know, there are known knowns; there are things we know that we know. There are known unknowns; that is to say, there are things that we now know we don't know. But there are also unknown unknowns – there are things we do not know we don't know.— Donald Rumsfeld [wikipedia]
A great philosopher creates new concepts: these concepts simultaneously surpass the dualities of ordinary thought and give things a new truth, a new distribution, a new way of dividing up the world.--Deleuze
But the virtual is not the same thing as the possible--Deleuze
We must understand that the virtual is not something actual but is for that no less a mode of being,--Deleuze
There are three general ways of thinking about the possible: (i) we take the present world as kind of a baseline, from which we project alternatives; (ii) we create a maximally plural list of 'worlds' available to, say, an omniscient being, and distinguish among these; (iii) we treat the possible as a species of quantitative probability ranging from impossible (zero probability) to necessity (probability 1) [and lots of permutations of these three]. There are, of course, ways to treat (ii) and (iii) as intimately related; and it is possible to recover (i) as a special case of (ii).
Now, Rumsfeldian unknown unknowns do not contradict any of the three approaches to probability sketched above. Even so, none of these three approaches is especially useful to help us think about unknown unknowns. For, in practice, it's our imaginations and the features of our formal machinery that constrain what enters the list of the (metaphysically/logically) possible. Logical analysis may be useful to make transparent hidden assumptions about the possible, it will not reveal an unknown unknown--genuine epistemic or metaphysical uncertainty is hard to model in it. For, all three ways of thinking about the possible have a kind of this-world bias built into them; to say this informally, while it's easy to prove the existence of God in some modal logics, it's actually pretty hard, despite the presence of monsters, to get really crazy; that is, the possible always has some resemblance to the actual.