a-cosmology-group/acg#120

A longer quote from that Wikipedia article:

A 2010 analysis of entropy states, “The entropy of a general gravitational field is still not known”, and “gravitational entropy is difficult to quantify.” The analysis considers several possible assumptions that would be needed for estimates and suggests that the [observable universe]( has more entropy than previously thought. This is because the analysis concludes that supermassive black holes are the largest contributor.[31] Lee Smolin goes further: “It has long been known that gravity is important for keeping the universe out of thermal equilibrium. Gravitationally bound systems have negative specific heat—that is, the velocities of their components increase when energy is removed. … Such a system does not evolve toward a homogeneous equilibrium state. Instead it becomes increasingly structured and heterogeneous as it fragments into subsystems.” This point of view is also supported by the fact of a recent experimental discovery of a stable non-equilibrium steady state in a relatively simple closed system. It should be expected that an isolated system fragmented into subsystems does not necessarily come to thermodynamic equilibrium and remain in non-equilibrium steady state. Entropy will be transmitted from one subsystem to another, but its production will be zero, which does not contradict the second law of thermodynamics.

This is exactly how it works in my own model. Energy is first removed from the CMB to drive gravity. This increases the entropy of the CMB photons, but reduces entropy in the graviton filaments which make up spacetime. The energy in those filaments is equal to the system’s gravitational energy and that energy is increased and concentrated due to the input from the CMB. Then when those filaments are in turn redshifted, energy is returned to the CMB photons, creating a blueshift in them and reducing the CMB entropy back to where it was.