jacques,
I discuss it here -
www.badastronomy.com/phpBB/viewtopic.php?topic=3364&forum=1&18
.
Here's some qualitative explanation, a crosspost from there:
Consider a very large chunk of space uniformly filled with tiny specks of matter so that the space remains essentially transparent. Let the peculiar speeds of those matter specks be negligible. Then let one of the specks turn itself into photons - annihilate. The photons leave their place of origin in shape of an expanding spherical shell of radiation. For all matter specks outside the expanding shell nothing has really changed, the gravity of the annihilated speck is still there. For the specks inside the expanding shell the case is different - the annihilated speck is gone altogether with its gravitational field. Accordingly, as the expanding shell covers new space, the potential at the positions of specks inside the shell gets a boost of Gm/r (m is mass of annihilated speck, r is distance from source speck former location), so the matter specks getting inside the shell acquire energy boosts.
Now I need to prove that the photons in the expanding shell are actually shedding that energy to the surrounding matter as they pass it by. Here goes - imagine we arrange that the photon shell gets fully absorbed at some distance from the source point - the mass/energy of the source is then spread between potential/kinetic energy of the matter specks covered by the expanding shell and the rest goes into the absorber as energy of photons. Thus for the greater distances from the source, gravity is not changed; it's simply that mass/energy within the shell got through some local redistribution, nothing more than that. Now if we absorb our photon shell at a greater distance from the source, the greater area with more matter specks in it acquires the potential/kinetic energy boost - hence, simple conservation laws require that the absorbed photons conveyed less energy to the absorber.
There is a yet simpler qualitative explanation - since photon emission by the stars is accomplished by shedding the mass, every such photon is accompanied with a gravitational perturbation that spreads with the photon, i.e. space is getting "uncurved" behind the photon - this is as if the space where the photon travels experiences a bit of localized transitional expansion exactly when the photon is there. Thus such photon always travels through a somewhat expanding space area, but this expansion is purely transitional and associated with the photon itself.
Now that's as much explanatory as I can possibly imagine, and it's very simple physics indeed.
