Conclusion 

I have made no more progress in the general theory of relativity. The electric field still remains unconnected. Overdeterminism does not work. Nor have I produced anything for the electron problem. Does the reason have to do with my hardening brain mass, or is the redeeming idea really so far away? 
Einstein to Ehrenfest, 1920 

Despite the spectacular
success of the theory of relativity, it is sometimes said that tests of Bell's
inequalities and similar quantum phenomena have demonstrated that nature is,
in some sense, incompatible with the local realism on which relativity is
based. However, as we’ve seen, Bell's inequalities apply only to a strictly
nondeterministic theory containing "free choice", and only under
the assumption of strong temporal asymmetry. Bell himself noted that the
observed quantum phenomena are not incompatible with "local
realism" for a fully deterministic theory, nor for a temporally
symmetrical theory. The entire framework of classical relativity, with its
unified spacetime and partial ordering of events, is founded on a strictly
deterministic and temporally symmetrical basis, so the relevance of Bell's
inequalities is dubious at best. Furthermore, since no energy or information
is propagated faster than light, even when Bell’s inequalities are
violated, quantum mechanics actually provides a striking confirmation
of the principles on which Einstein chose to base special relativity.
Admittedly the phenomena of quantum mechanics are incompatible with at least
some aspect of the intuitive metrical idea of locality, but this should not
be surprising, because (as discussed in the preceding sections) the metrical
idea of locality is already inconsistent with the pseudometrical structure
of spacetime, which forms the basis of modern relativity. 

It is often said that general relativity, rather than being a further development or generalization of the concept of relativity, is actually “just” a theory of gravity. Similarly one could argue that the relativity of Galileo and Newton was not a further development of the purely kinematic concept of relativity (Heraclides), but was actually “just” a theory of inertia, which in a sense is a restriction (rather than a generalization) of the concept of relativity. Likewise general relativity teaches us that the principles of special relativity are strictly applicable only over infinitesimal regions in the presence of gravitation, so in this sense the general theory restricts rather than generalizes the special theory. However, we can also regard special relativity as the theory of flat fourdimensional spacetime, characterized by the Minkowski metric (in suitable coordinates), and the general theory then generalizes this by allowing the spacetime manifold to be curved, as represented by a wider class of metric tensors. It’s remarkable that this generalization, which is so simple and natural from the geometrical standpoint, leads almost uniquely to a viable theory of gravitation. Echoing Minkowski’s comment about ‘staircase wit’, it’s tempting to claim that if we hadn’t already known about the existence of gravity from our direct experience, some fancyfree mathematician might have predicted it, based purely on abstract considerations of the class of intelligible spatiotemporal metrical relations. (Interestingly, a similar claim has been made about modern “string theory”, i.e., that it “predicts” the existence of gravitation.) 

Just as the special theory is sometimes interpreted in the Lorentzian sense, an analogous situation exists with regard to the general theory, which is sometimes interpreted as a field theory rather than as a geometrical theory. Those who hold this view usually discount or downplay the equivalence principle, and regard the fact that gravitation can be represented by curvature of spacetime as merely a mathematical oddity of no fundamental significance. One of the motivations for this approach is the desire to make the theory of gravitation fit into the same “linear” mold as the theories of the other fundamental forces (albeit with nonlinear corrections to match the field equations), usually with the hope that this might help us to see how the theory could be quantized like the other theories. From considerations of the spectrum of gravitational waves in equilibrium inside a container, analogous to Planck’s analysis of blackbody electromagnetic radiation, one would expect that the gravitational field must be quantized, and yet no convincing quantization of the nonlinear field equations of general relativity has emerged. This may not be so surprising when we consider that a quantum of gravitational radiation would also have to be a source of gravitation – analogous to a photon with electric charge. Only a few physicists (including Einstein himself) have tried to approach the problem from the opposite direction, taking the nonlinear geometrical basis of general relativity as the starting point, and seeking some further generalization or restriction that would yield quantum phenomena. 

Of course, it's entirely possible that the theory of relativity is simply wrong on some fundamental level where quantum mechanics "takes over". In fact, this is probably the majority view among physicists today, who hope that eventually a theory uniting gravity and quantum mechanics will be found which will explain precisely how and in what circumstances the classical theory of relativity fails to accurately represent the operations of nature, while at the same time explaining why it seems to work as well as it does. However, it may be worthwhile to remember previous periods in the history of physics when the principle of relativity was judged to be fundamentally inadequate to account for observed phenomena. Recall Ptolemy's arguments against a moving Earth, or the 19th century belief that electromagnetism necessitated a luminiferous ether, or the early20th century view that special relativity could never be reconciled with gravity. In each case a truly satisfactory resolution of the difficulties was eventually achieved, not by discarding relativity, but by reinterpreting and extending it, thereby gaining a fuller understanding of its logical content and consequences. 
