I mostly agree with you, Rob, save for one critical point:
"The cosmos can thus 'predict' the future, by simply becoming the future, but only in 'real-time'."
If general relativity is true, there is no physically real time, just physically real spacetime. Joy's earlier work, "Absolute being vs. relative becoming" addresses that issue of dynamics and continuity.
"The prediction of the event and the event are one and the same thing. So future prediction of the entire cosmos is not possible."
Right. All physics is local, as Einstein put it.
"Only a tiny subset of all the events occurring in the cosmos can ever be predicted, namely, those events and phenomenon, which are nearly devoid of all information. Because these alone are able to have all the necessary initial conditions stored in a memory. That tiny subset is the sole subject matter of physics."
A measurement function continuous from the initial condition, as Joy's framework allows, applies to *all* initial conditions. The framework is complete both mathematically and physically. Nature makes no distinction between past and future events, only locally real choices of topological orientation. The choice function contains all information of events subsequent to the initial condition -- the difference between Bell's choice and nature's choice, however, is that Bell-Aspect results are linear and discontinuous; Joy's results are nonlinear and continuous. The former is probabilistic, the latter is deterministic. "The experiment not done" is metaphysically real, not probabilistically unknown.
" ... Joy has missed a much more fundamental problem with Bell's Theorem - the is no evidence that the correlations between components are in fact 'between components', since there is no evidence for multiple components."
Yes, he says that -- a simply connected space admits correlations continuous from the initial condition, while the multiply connected spaces of conventional quantum mechanics admits only one correlation "at a time" (i.e., for each throw of the dice).
"I pointed this out to Joy, during the 2012 essay contest, but he did not understand its significance. I referred him to the statement in Bernard d'Espagnat's Nov., 1979 Scientific American article, which is the central problem:
"'These conclusions require a subtle but important extension of the meaning assigned to a notation... it is converted by this argument into an attribute of the particle itself.'
"And if you buy that, then you should have no problem buying into the ancient idea that the color of an apple is solely an attribute of the apple itself."
It is, yet only in the sense that the apple absorbs the colors of the spectrum except the one reflected. In terms of Bell's theorem, this would amount to all possible colors being in superposition, and the color of the apple as observed being real. Joy's framework, however, contains no probability, no superposition, no nonlocality.
"The spin or polarization of a particle is no more an attribute of the particle itself, than the 'heads' or 'tails' of a coin is solely an attribute of the coin."
I think Joy's measurement framework agrees with you. Because the framework is classical and nonlinear, the randomness of a coin-toss probability is not the physically real measurement result -- the physically real measurement result is, as he has always said, the correlation of quantum measurements exactly as predicted by E(a,b) = - a.b which is what Bell-Aspect results claim is impossible.
"Irregardless of what Joy can prove about the nature of the correlations, the entire argument rests on that 'subtle' extension being true. But it is almost certainly false."
That extension is based on belief in superposition. There is no superposition in Joy's framework.
Best,
Tom