There are a lot of other methods to approaching a TOE other than String Theory. For example I'm playing around with Minimal Quantum Surfaces in De Sitter Space as a renormalization technique for quantum gravity (it's a mouthful but I'll do my best to explain). A minimal surface is a geometrical object embedded into space that minimizes its area given some boundary, formally a minimal surface is defined as if and only if it's mean curvature is zero at every point on the surface. Soap films are naturally occurring minimal surfaces. The Mean Curvature of a surface is an extrinsic measure of curvature and is half the sum of the two principle curvatures where the principle curvatures are the eigenvectors of the shape operator. The appeal of minimal surfaces is that they are solutions to the Lagrangian Equation which describes the dynamics of a system and is time-invariant and thus from Noether's Theorem all Lagrangians conserve energy. So from a Quantum Field Theory perspective specifically perturbation theory it may be possible to represent Feynman Diagrams as minimal surfaces where at every point on this surfaces there exists a vertex, a creation and annihilation operator. This would allow you to replace the interior lines of Feynman diagrams otherwise coined "virtual particles" (which don't really exist beyond being mathematical approximations in perturbation theory) with a local geometry compensating for missing energy-momentum due to the the very curvature of spacetime on quantum scales. Finally, if one integrates over all these minimal surfaces over all space you still have flat spacetime on cosmological levels precisely because minimal surfaces have zero mean curvature.

The idea is to replace virtual particles with local spacetime curvatures and thus the problem with gravitons simply go away due to their unnecessary existence. These surfaces will be embedded in a De Sitter Space not Minkwoski Space. De Sitter Space is a solution to Einstein's Field Equations with a positive cosmological constant and allows spacetime to be curved without the presence of energy or mass. This is appealing because "virtual particles", again mathematical artifacts are off mass-shell meaning they violate E^2=M^2 + P^2 the energy-momentum relation. It would be like rolling a ball on a curved surface rather than a flat one, obviously the ball on the curved surface will have more momentum than the one on the flat surface.