The thin sheet of reality of the hologram universe is the brane of Hawking's Mind of God. ;-)

1) S-Matrix?

S-Matrix is an input-output black box model for scattering of simple beams of non-interacting particles (ensembles) on similar ensembles of targets (true each target can be a complex system like a crystal or fluid with quasiparticles and collective mode excited states of the ground state).

The inputs and outputs are external lines corresponding to poles of the Feynman propagators in the complex energy plane. In other words they are real particles on-mass-shell. If massless bosons in the classical limit they are radiative far fields of only two transverse polarization if spin 1 (EM) or spin 2 (gravity).

Virtual particles are inside the black box (internal lines of the Feynman diagrams).

However, for cosmology and the hologram universe - for horizons the S-Matrix is incomplete. Sure you can use it for collapsing matter increasing the area-entropy of the horizon (black hole) or matter flowing out of the causal diamond in the case of our observer-dependent future cosmological event horizon.

However, the stable state is completely off-mass-shell, i.e. coherent Glauber states of off-mass-shell virtual bosons like the EM near fields of electrical power engineering of our grid and our every day electrical machines and home appliances. PG & E mainly deals with near fields. High energy physicists seem to draw a blank on near fields. The most important parts of the universe are near fields.

Thus the two most important toy model SSS metrics in Einstein's GR are

g00 = 1 - rs/r  etc. black hole

g0'0' = 1 - r'^2/A

with horizons g00 = 0 and g0'0' = 0

Quantum mechanically speaking these metrics are made out of Glauber coherent states of off-mass-shell gravitons of zero frequency and a continuous spectrum of wave vectors.

GRAVITY WAVES PLAY NO ROLE in this static limit.

Yes, they do when excited states of the pixeled stretched membrane (Kip Thorne) with scrambled BITs for distant observers is included, i.e. Hawking radiation. But that is treated in perturbation theory around the above static solutions.

Now I do remember that the S-Matrix can be extended to treat bound states so maybe that is the way out of the dilemma?

2) 't Hooft's naive rejection of Aharonov's destiny post-selection &Wheeler-Feynman Cramer's transactions, CTCs in quantum computing, and invoking the Red Herring of the Grandfather paradox - see the World Science Festival 2011 video with 't Hooft, Susskind, Verlinde, Buosso - moderated by Hockenberry of PBS.