Thanks, Daryl! Since I complimented you on your essay site for following clearly stated premises to a logically closed judgment, this is a good place to insert a link to a paper by Tobias Fritz recently posted to the arXiv that I think shows pretty clearly the consequences of defining things in a way that guarantees validation by observation, in effect rigging the game -- or loading the dice, if one prefers -- toward a preconceived reality. Logicians call this method of reasoning "double negation," meaning that one fits a result to a nonconstructive argument so that there is no means of escaping the conclusion for which the argument was prepared.
Fritz wants to obviate the free will hypothesis (i.e., " ... agents conducting measurements can choose between different measurement settings in each run of the experiment, and this choice is random and independent of the source") for Bell's Theorem, thereby strengthening the theorem's conclusion that no local hidden variables theory can reproduce quantum predictions even given free will, i.e., closing a loophole to the main conclusion that no model of a physical system can be both local and realistic.
He starts with the assumptions:
"(I) Realism: Any physical system can be described in terms a probabilistic mixture of states in which all observables have definite values ('hidden variables'). Measurements reveal these definite values.
(II) Locality: Physical systems have spatial components which can be described independently. They do not interact across spacelike separated events."
Who decrees that because a physical system *can* be described in probabilistic states that it *has* to be? Certainly no classically continuous model is described in "a probabilistic mixture of states." The disguised assumption here is that probabilistic reality is a physical law, that because quantum theory cannot model classical phenomena, we can therefore leap to the conclusion that classical physics cannot model quantum phenomena. Joy Christian has falsified that notion by accomplishing the required feat -- with a local, realist and constructive argument that agrees with all but the "probabilistic mixture of states" in Fritz's definition. Remove that assumption, and free will comes back into play with exactly the degrees of freedom required to reproduce quantum correlations locally, and no value can be assigned to nonlocality -- as Bell-Aspect results are compelled to do.
(N.B. Fritz wants replace the free will hypothesis with his own: "if an experiment contains several sources, then the theory describes these sources as independent. This means that the joint distribution of hidden variables is a product distribution." In our essay, it is clear that because the source of *all* information is a point at infinity, a measurement result realized from a continuous range of variables guarantees a locally real event without obviating equally real results of any experiment not performed in that same local time interval.)
Tom