A good analysis.

What I am struggling with here is an intuition that the whole idea of unpredictability in "the theoretical/philosophical sense" is a bad, ill-formed idea. I know roughly what it means to have predictability as a two-place predicate. P(E, A) means that person A (a person equipped with the theory and empirical information that A has) is capable of predicting event E. Fine. But now how do we turn that into a one-place predicate. Do we define:

• P1(E) == Forall persons A . P(E,A)

or is it

• P1(E) == Forall physically possible persons A . P(E,A)

or is it

• P1(E) == For some hypothetical omniscient person A . P(E,A)

or is it something more complicated, involving light cones and levels of knowledge that are still supernatural.

The thing is, even if you are able to come up with a precise definition, my intuition makes me doubt that anything so contrived could be of any possible use in a philosophical enquiry.