Could GR's "background independence" be a theoretical artifact? ==quote from Rovelli "Unfinished Revolution" (2006) page 2== ...Others, on the other hand, and in particular some hard–core particle physicists, do not accept the lesson of GR. They read GR as a ﬁeld theory that can be consistently formulated in full on a ﬁxed metric background, and treated within conventional QFT methods. They motivate this refusal by insisting than GR’s insight should not be taken too seriously, because GR is just a low–energy limit of a more fundamental theory. In doing so, they confuse the details of the Einstein’s equations (which might well be modiﬁed at high energy), with the new understanding of space and time brought by GR. This is coded in the background independence of the fundamental theory and expresses Einstein’s discovery that spacetime is not a ﬁxed background, as it was assumed in special relativistic physics, but rather a dynamical ﬁeld. Nowadays this fact is ﬁnally being recognized even by those who have long refused to admit that GR forces a revolution in the way to think about space and time, such as some of the leading voices in string theory. In a recent interview , for instance, Nobel laureate David Gross says: “ [...] this revolution will likely change the way we think about space and time, maybe even eliminate them completely as a basis for our description of reality”. This is of course something that has been known since the 1930’s  by anybody who has taken seriously the problem of the implications of GR and QM. The problem of the conceptual novelty of GR, which the string approach has tried to throw out of the door, comes back by the window. ==endquote== Would it be possible to develop a theory of gravity that is background dependent, a fluctuating field over a fixed metric, that reproduces all known GR results where GR has been tested? What about the spin-2 field on flat QFT? I wonder if BI is a theoretical artifact that, while a valid conclusion based on GR"s field equations, are invalid when applied to nature. In otherwords, while GR succeeds as modeling gravity as encoding the geometry of spacetime, what gravity fundamentally is is some sort of emergent van der waals type interaction of something more fundamental.