I chanced upon an argument in Misner, Thorne and Wheeler to the effect that the energy/momentum of the gravitational field cannot classically be localised. Basic idea: you can make the Christoffel symbols vanish at any point, and hence the gravitational field at that point will vanish, taking any sensible conception of gravitational energy/momentum with it. What does this mean for attempts to formulate gravity as a local quantum field theory? Can higher-order terms in some effective lagrangian contain derivatives of the Christoffel symbols that are non-vanishing, even when the Christoffel symbols themselves are? Thanks in advance.