Brace yourselves for gravitational waves

[Alex K]

So if you now take the *deviations* from the flat space metric tensor, and treat these numbers as a field akin to the electromagnetic field, the rules of quantum theory suggest that for any given frequency, there is a smallest deviation from flat space which corresponds to a massless bosonic particle -the graviton -with spin 2, just as from the EM field we get photons. If you plug this into the usual formalism to obtain a kind of more complicated QED, the leading resulting Feynman graphs give you the usual gravitational attraction via exchange of gravitons, in analogy to the em force from feyman graphs with photons. The trouble only starts when one has to consider how the gravitons interact with each other at very high energies...

In this case one naively gets results like the probability of radiating off one more graviton always being higher , which would lead to probabilities larger than 1, and so the notion of individual particles with well-defined interactions breaks down. Suddenly, an infinite number of free parameters would be needed to specify all the possible ways in which 2,3,4.... gravitons interact, and this kills predictivity