So, viewing the EPR system as two particles from the moment of apparent separation doesn't work. This note presents an alternative view, namely that when one particle "splits in two", it's wave function doesn't. It continues to be described by a single wave function until the properties of the split wave function are established by an interaction (measurement). Aspect proposed that wave functions be considered 'global', independent of space-time extent, that wave function properties exist with respect to a central local space-time reference for that wave function. I propose that it is clearer to consider that wave functions are in the eye of the observer, not the particles cf. On The Interpretation of Quantum Mechanics.
A nomenclature problem is worth considering here: when the state of a particle is determined by measurement, the wave function is considered to have "collapsed". But, a wave function isn't a physical entity, it's a theoretical construct that models external state of knowledge of a particle. Collapse implies physical movement, but since wave functions aren't physical entities, nothing moves when a function "collapses". I recommend the term "resolution" to describe the process by which a wave function becomes observed; this term doesn't imply physical movement.
This view does not propose that wave functions exist instantly throughout all space time. EPR, and everything else we observe, is consistent with wave functions expanding at the speed of light, it is only their resolution that occurs 'instantly'. The wave function of a particle that has the capability of splitting in two expands at the speed of light from the points at which that possibility exists. This expanding wave function is 'cut off' upon an interaction; the residue of the pre-interaction wave function continues outwards in space-time, but is not affected by subsequent interactions or resolutions.
With this view, EPR becomes a local experiment. First, we set up conditions so that a particle can "split into two". This constitutes an observation, and both halves of the split particle remain local to this first observation. When the properties of one of the split particles is determined, a second observation, the joint wave function is resolved to describe both particles, simultaneously with reference to the space-time at the observed point of split.
If we were not aware of (observe) the condition that provided for particle splitting, we would not be aware of any correlation until we inferred the existence of that condition by repeated observation. That would then be in effect become a first observation.
Wave function resolution that is instantaneous relative to a single space-time might have applications in several cases of extreme space-time density.
Consider, for example, the big bang theory. In order to model the observed uniformity of the universe, it's necessary to introduce an "inflation" period, during which the universe instantaneously expands by a huge factor in order to model its observed uniformity. But, if wave function resolution is independent of space-time, might correlation of states of the early universe be able to model the observed outcome without requiring the introduction of the inflation hypothesis?
It's believed that black holes smaller than a size dependent upon the background temperature of the universe at the time must evaporate, ending in a "puff" of extinction. But, everything else in our universe is quantized - why not black holes? Might the independence of space-time of wave function resolution permit the evaporation of black holes to end in a ground state? If so, what might this ground state be? Could it be a Higgs, thus providing a link between quantum and general relativity theories?
When matter approaches a black hole, it slows to zero speed at the event horizon as we view it from outside the hole, so never gets inside it. But, there must be a way in which additional matter can be added to a black hole, because we observe such huge ones. (The method might be the merging of two black holes, when the distortions of space-time render simple time views moot.) However it happens, when more matter is incorporated into a hole in equilibrium, string theory is invoked to provide that the resulting increased radius of the hole doesn't "swallow" previous matter, that the wave functions of previously incorporated matter expand along with the black hole horizon. Would space-time independent wave function resolution give an answer more consistent with the fundamental essence of quanta than string theory does here?
Of all the questions in physics, if I had the tools, this is where I'd be looking.