light cone entropy

Not sure if I am buying my light-cone entropy derivation. Does a photon consider every null geodesic on a light cone an "available state"? More importantly, in flat space the angular parts of a photon trajectory are not really degrees of freedom. Once emitted the photon will not travel in these dimensions but will intersect each hypersurface with the same value.
In fact, could the entropy decrease as it moves further out? The error in perpindicular position, the position granulation will grow with time, but the location of the light ray on the cone will stay constant...