It is just a theory, in the sense that not only do we lack experimental evidence, but the construction of an experiment to observe them is actually impossible.

You may know that there are four fundamental kinds of forces, which, in decreasing order of strength, are: the strong nuclear force, electromagnetism, the weak nuclear force, and gravity.

The strong nuclear force holds quarks together to form baryons (which include protons and neutrons) and holds protons and neutrons together in atomic nuclei (something has to overcome the huge electric repulsion of all the positive charge).

The electromagnetic force includes static electricity, attraction between permanent magnets, and what’s called “normal” force (the reason why things don’t just fall to the center of the earth: the electrons in every atom repel and keep things from passing straight through each other).

The weak nuclear force is responsible for radioactive decay and neutrinos. Finally, gravity attracts different masses and results in solar systems, galaxies, and black holes. Unlike electromagnetism, there is no repulsion for gravity. It may be weird to think of gravity as weak, since, after all, stars keep planets in orbit through gravity, but it’s 100,000,000,000,000,000,000,000,000 times weaker than the weak nuclear force, which in turn is about 100 times weaker than electromagnetism.

The other three forces are explained by a theory called the Standard Model of particle physics, which we have tested to extreme precision with accelerators like the Large Hadron Collider (LHC). However, even though we discovered the theory of gravity first, we actually understand it the least. We cannot explain it with the Standard Model, and if we try, we end up with unavoidable infinities in the equations. The best theory we have for gravity is Einstein’s general relativity, which successfully predicts things like gravitational lensing and the slight discrepancies in Mercury’s orbit. The Standard Model models forces as particles that carry the effects of the force back and forth between matter particles, but general relativity says gravity is the result of curved space and time, and not the result of particles. These two theories explain our data to very high precision, but they model forces in completely different ways and more importantly do not agree with each other, so at least one must be wrong.

To resolve the contradiction, we put together theories of quantum gravity, such as string theory, which replaces the particle and curvature pictures with strings for all of the forces. In particular, it predicts a particle (or rather, a string) called the graviton.

Remember how I said gravity is so much weaker than all the other forces? For practical purposes this means that to test any of our theories using our current methods, we would need an accelerator larger than the Solar System just to collect the energy needed just to produce a graviton. To detect it, we’d need so many detectors that they would collapse into a black hole.

So we don’t have evidence from experiment, and actually cannot even build (much less raise the money for) such an experiment. What do we do when theories can’t be tested with experiment? We derive them as results of existing theories that we have tested, using mathematics and logic to make the jump. One of the most enduring and successful theories is the purely mathematical Noether’s Theorem, which relates conservation laws to symmetries. Using it, particle physicists predicted the existence of different particles to guide accelerator experiments. However, we seem to have exhausted its usefulness after applying it so many times. There are still a lot of open questions, and no experiments to resolve them, so for now we just have a lot of competing theories derived from what little we know.