If General Relativity were only a theory of accelerated frames, would it contain the Equivalence Principle?

Question

In Relativity: the special and general theory, Einstein motivates General Relativity by the need to extend relativity not only to inertial frames, but also to accelerated frames. The laws of physics should have the same form in all frames of reference, be they inertial or even accelerated.

But, later in his exposition, his attention seems somehow to shift from a theory of accelerated frames to a theory of gravitation.

The bridge seems to be the Equivalence Principle (" happiest thought of my life" as says Einstein).

I've tried to reconstruct Einstein's train of thought in the way below. In this reconstruction, the Equivalence Principle appears as a way to turn a theory of gravity into a theory of accelerated frames.

My question: is it really the case that the function of the Equivalence Principle is to turn a theory of gravity into a theory of accelerated frames?

My attempt at reconstructing Einstein's project:

1. The goal is mainly to study frames of reference that are submitted to a gravitational field

2. but such frames can be seen equivalently as accelerated frames : studying in a purely theoretical way the behaviour of objects in an accelerated frame will therefore yield ( at no experimental cost) a correct theory of gravitation

3. for example, a purely mental experiment shows that light is bent in an accelerated frame ( such as an elevator accelerated upwards), which implies ( by the Equivalence Principle) that gravity bents light

4. the road to a theory of gravity is therefore to take an inertial frame S and to consider another frame S' that is accelerated relative to S : studying the effects of acceleration on S' will be the same thing as studying the effects of gravitation on this frame.

Answer 1

The equivalence principle only applies to local frames; gravity is manifestly different from an accelerated reference frame if you look at a large enough volume of space that tidal forces (e.g., spacetime curvature) become important. Therefore, general relativity is more than studying accelerated reference frames in a flat spacetime. I would actually classify accelerated frames in a flat spacetime as part of special relativity (even though normally accelerated reference frames are not taught in courses on special relativity).

Having said that, I basically agree with the gist of your four point list at the end of the question. By understanding accelerated reference frames in a flat spacetime, we can understand the effects of a local gravitational field, and by patching together different local accelerated frames we can build up a global picture of a curved spacetime. Weinberg's famous book on gravitation essentially takes this approach, rather than starting from the assumption that gravity = curved spacetime.