Hi Lev,
I am not fully conversant with ETS formalism, though I am willing to learn. Nevertheless, as I read your essay, I can discern no difference between a "temporal stream of interconnected primitives" which form structs, and a bit stream of primitive states.
Relevant to (1). What I mean is, suppose that time gives structure to space:
Then the orientability which defines the stream (indeed, even allows us to give meaning to the word "stream") is a space-time structure _because_ it is oriented. If we find that all of these primitive spacetime structures -- assuming scale invariance and infinite self similarity -- are oriented the same, we have to allow a 2-dimensional object just on the principle of orientability alone.
So primitives that are featureless unless oriented (streaming) implies both time dependency and spatial domain. Because this domain is 2 dimensional and therefore bounds an infinity of points, and because we can further assume the nonlinear evolution that time dependency implies, whatever structures that emerge in d > 2 are already oriented in the same direction. That is, we see that the 2-dimension topology (S^1) on the manifold of a two ball (S^2) is a three dimension object from any point of S^3, the four dimension sphere on which we live. This leaves the 2-ball itself undefined! We know there's correspondence between a point on the 3-sphere manifold and an interior point of the 2-ball, though, because we map an internal plane self similar to the S^2 manifold, of zero curvature. That's not the "real" structureless interior of the 2-ball though, is it? -- we only know that whatever forms may result on the plane are structured by our singular assumption of orientability.
I won't go into Riemannian geometry (every Riemann surface is orientable), but I think the implications are more or less obvious, and interesting.
Anyway, physical information theory follows the same rules as thermodynamics (Shannon). So I think you can make the connection between the Jacobson-Verlinde treatment of gravity as identical to information, and the conclusion that you and I have both reached by different routes: time is identical to information. If gravity is time dependent, then, there exists at least one quantum gravity model in which time is identical to information, because the classical gravity spacetime trajectory is reversible and the quantum information time trajectory is not. That the time metric can orient to all points of the space gives structure to an otherwise formless object. The field of information in the n-dimension stream, n > 1, is ordered by the 1-dimension time metric.
Therefore, time is more primitive than space -- I would have to be convinced that time is more primitive than the integers, before I would give up on a mathematical ontological model.
Regarding (2): I'm not convinced that we can't treat structs in a topological model using scale invariance. More later?
Tom