Hi Jonathan,
Thanks for the paper. In looking at it I see many things which are in my notes and which I have in other papers and the book "Sphere Packing, Lattices and Codes" by Conway and Sloane.
The graininess of spacetime is something which I think only comes about with the measurement of black hole states. As I indicated on Giovanni Amelino-Camelia's essay blog site there is an uncertainty principle,
ΔrΔt ~ (2Għ)/c^4 = L^2_{Planck}/c.
which is commensurate with equation 1 on Giovanni's paper . Spacetime appears grainy depending upon the type of measurement one performs. In the case of a quantum black hole a measurement involves spatial and temporal coordinates in a null congruency called an event horizon. If one makes another type of measurement spacetime is then as smooth as grease on an ice skating ring. The measurements of delay times for different wave lengths from very distant gamma ray burstars indicate that space is smooth down to a scale 10^{-50}cm --- far smaller than the Planck scale. This then ties in with some interesting work by Torsten Asselmeyer-Maluga on the role of exotic four dimensional space in quantum gravity. These are homeomorphic spaces that are not diffeomorphic. In 11 dimensions the 7-dimensional is dual to the 4-dimensional space. The exotic 7-spheres found by Milnor are simpler, with only 7-distinct non-diffeomorphic forms, rather than an infinite number.
The octonions are a system of 7 quaternions. The exotic system in 7-dimensions I think might be connected to the automorphism G_2 in E_8 or SO(O). This would then connect with a physical meaning of octonions and nonassociativity in physics. The Polyakov path integral
Z[A] = ∫δD[ψ]/diff(ψ) Ae^{-iS[ψ]}
"mods out" diffeomorphism or equivalently gauge changes on a moduli. Yet with exotic spaces this definition becomes strange. However, if there are 7 quaternions which are related to each other by nonassociative products (ab)c - a(bc) =! 0, then the measure can maybe be realized according to associators δD[ψ]/diff(ψ).
I discussed octonions a bit with Lockyer, but he seemed a bit put off. As I see it, and from some experience, presenting a gauge theory with nonassociative brackets and stuff falls pretty flat, I am not necessarily saying this is wrong, but doing that sort of work has a way of getting people to present their backside to you. I think the role of nonassociators is best advanced by other means so that in the future they may simply be too convincing to ignore.
Cheers LC