Peter,
The loss of unitarity requires a somewhat different way of thinking. I think it is a great thing that our educations give us rigor in our thinking and abilities to solve problems. There is an unfortunate flip side where with regards to foundations this can lead to a sort of "rigor mortis." Trying to think of physics that is outside of what is accepted is very difficult. It is very easy to come up with something that is completely wrong, where one has to abandon that ship as soon as possible. I have spent years doing that. It is very difficult to come up with something which is maybe correct and hard to get it accepted.
I would need to think about what you are asking. This seems in some ways similar to causal set theory.
In my essay and reference #13 I illustrate this with noncommutative geometry. The small scale structure of spacetime is then wildly strange near the Planck scale. The best and most interesting about these essay contests is communicating these results with other people. I have found that Amelino-Camelia's work has interesting connections to twistor space, which my paper is based on in part with the BCFW recursion. The work on shape dynamics and causal sets and so forth of Gryb, Dribus and Barbour are also interesting. What in some ways is the most interesting is the work by Asselmeyer-Maluga and Krol on exotic manifolds, a topic I was taught near the end of a course on differential or algebraic geometry. This turns out to imply physically smooth space or spacetime. This is beginning to look like another door into this area, one where a wild chaotic microstructure to spacetime is dual to a smooth structure.
Donaldson, Drinfeld and Freedman showed that this results in an infinite class of manifolds which are homeomorphic but not diffeomorphic, called exotic manifolds. These substructures one may look at are things like of Casson handle R^2xD^2, D^2 = two disk, or T^2xD^2. T^2 = two torus. These many be removed from an R^4 and replaced with CP^2#CP^2, and this space is homeomorphic to the original space, but not diffeomorphic. If we then consider R^4 = R^3xΠ_n{t_n} (an infinite product of time intervals), the R^4 is a huge stack of four dimensional "slices." Each stack may then be exotic. The infinite product is a form of "time operator."
The stack of exotic spaces is a foliation of 4-spaces mapped to each other by knot operations, or more generally by the Yang-Baxter equations. These are general braid operations, where each line is a Wilson line integral with a gauge field content. Further, the braid operations are quantum groups, or the generator of Hopf C* algebras. So in this setting we may consider quantum gravity as a process of quantum homeomorphisms on both the AdS and its boundary dS, which have opposite orientations. The gravity content is contained within the AdS, but conformal symmetry breaking of the CFT on the boundary means there is some gravity content on the boundary.
I was stuck on this problem of conformal completion of AdS, but recent communications with Asselmeyer-Maluga and Krol have illuminated how this process can be considered with exotic 4-spaces. There are some parallels with C* algebras, and it now turns out that quantum gravity is a Yangian system of conformal dual gauge symmetries --- in part giving the conformal completion of AdS.
It is strange to think spacetime might have a wild and chaotic structure on a small scale, and then at the same time a structure that is perfectly smooth. The Yangian system constructs a dual geometry, and this is reflected here. Experimentally it has been found that gamma ray burstars that are billions of light years away have no dispersion of light. If spacetime is grainy on a small scale it is expected that short wavelength light would interact more strongly with this "foam" and would be slower. However, the observational data does not bear that out. This is an experiment which involves a huge distance, and momentum p = ħk is such that if we consider this distance the "probe momentum" is near zero. Another experiment which might involve enormous energy and large probe momentum might then measure the noncommutative or foamy structure to spacetime.
Cheers LC