Einstein's ponderings on gravity led in 1907 to his principle of the equivalence of gravitation and inertia, illustrated by his famous elevator prop. For Newtonian gravity, it is impeccable. But his elevator illustration of gravity's bending of light implicitly allows light to be accelerated to a speed greater than c, violating his 1905 Lorentzian relativity. In 1913, Einstein decided that general relativity supplants Lorentzian relativity, and proceeded to his famous gravity theory. But his Einstein equation has no unique solution, hampering his effort to explain Mercury's perihelion precession. He overcame that obstacle in 1915 with a simple Lorentz-invariant coordinate condition which isn't mentioned in textbooks, honoring his insistence that coordinate conditions are irrelevant. Friedmann, also honoring that insistence, found a new coordinate condition in 1922 that can make the Einstein equation easier to solve; it was eagerly adopted by cosmologists. But it makes gravity essentially Newtonian, which is drastically wrong at high redshifts. In 1962, Feynman elucidated general relativity's physical role in gravity theory; it generates the gravity field's self-interaction in the context of an action principle, which is a fundamental prerequisite for quantization. But Feynman missed the very large constant in that action, the key to gravity's classical-field nature and the absence of gravitons, so he tried quantum-gravity perturbations, which are a physical dead end.