Einstein, Podolsky and Rosen argued that quantities that are conserved absolutely such as momentum must correspond to an objective element of reality. Because QM does not model these elements it must be incomplete. In the subsequent Bohr/Einstein debates Einstein argued as if there were additional hidden variables associated with individual particles that determined experimental outcomes. Over time massive experimental evidence has accumulated that experimental outcomes are determined at the moment of observation in a way that appears to be completely random. At the same time there is massive experimental evidence that the conservation laws are absolute and not statistical.
One can update the EPR argument by noting that there can be an objective process that enforces conservation laws and individual observations that appear to be random if those observations result from a chaotic process. The existing experimental record does not distinguish between a local chaotic process and irreducibly random observations. Fortunately we know through the work of Bell and others how to construct practical experiments that can distinguish between these possibilities.
One can make the argument of EPR more compelling by noting that all of the seemingly strange and metaphysical claims about QM depend on the assumption that probabilities are irreducible. All of this strangeness disappears if probabilities come from a chaotic physical process. The strange aspects of QM that are a necessary consequence of irreducible probabilities include the following.
1. Quantum entanglement
2. The need to use configuration space to model state evolution. This cannot be done with relativistic evolution in physical space.
3. Experimentally state evolution appears to be completely reversible but the actualization of probabilities defined in the wave function is an irreversible process.
4. There is not and cannot be an objective solution for the measurement problem in the framework of standard QM.
Quantum entanglement is necessary because once two particles interact and move apart conservation laws related to that interaction must be enforced absolutely even though observations of the individual particles must be completely random. Thus the observation of one of the particles imposes a constraint on observations of the other that did not exist before and cannot be explained through hidden variables.
The need for configuration space models and the inability to use physical space models to describe state evolution is a direct consequence of quantum entanglement. One must model the probabilities for all the combinations of outcomes that are consistent with the conservation laws.
The actualization of probabilities is an irreversible process that destroys information. Yet all the evidence suggests that physical state evolution is reversible and preserves information.
The inconsistency between the observed reversibility of nature which seems to be accurately modeled in reversible wave function evolution and the absolute irreversibility associated with actualization of a probability prevents any objective physical solution to the measurement problem. There is no way to define objectively what a measurement is or when it is complete.
In contrast, if measurements are determined by a chaotic physical process then quantum entanglement exists but is enforced though a local physical process that also enforces the conservation laws. Quantum collapse is something like convergence to an attractor in chaos theory. As particles move apart the degree of quantum entanglement between them fades until they become essentially independent. The actualization of probabilities is a physical process that is reversible.
Had chaos theory existed at the dawn of QM these issues might have been understood in this way from the time one first observed the apparently random nature of individual observations. It is far easier to apply a new conceptual framework to a new set of problems than it is to alter ones perspective and see a set of problems that one is intimately familiar in a different conceptual framework.
I do not expect anyone to change their conceptual framework based on the above arguments but I do think they are a compelling reason to give greater thought and attention to experimental tests of Bell's inequality.