Cohl
You've written a refreshingly thought provoking essay. I too, am very much drawn to the algebraic approach. From the matrix model perspective, particles indeed seem to have a purely algebraic description. For example, in Smolin's D=27 matrix model, fermions can be identified with rank one elements of the observable algebra, consisting of only self-adjoint operators. These rank one elements are eigenmatrix solutions of a characteristic equation in a Jordan formulation of quantum mechanics. When normalized, these rank one elements are primitive idempotents of the observable algebra.
In Smolin's model, there is an F_4 symmetry, acting on sets of (at most) three orthogonal idempotents, which are easily seen when a given element of the observable algebra is expanded in a spectral decomposition as a sum of (scaled) rank one elements. Geometrically, the picture is in a projective space, where a given operator takes values on three charts. Imagine this as a 3-sphere generalization of the two charts on a 2-sphere. These charts, in Smolin's model, are 16-dimensional, reflecting the 16-dimensional structure of the fermions in his model. The F_4 symmetry arises from the octonion division algebra underlying the observable algebra. Using F_4 transformations, one can diagonalize a given element of the observable algebra, and ascertain the (eigen)values the operator takes on the three charts.
In string theory, for coincident brane configurations, one has a U(N) symmetry. If we restrict to the case of three branes, we have a worldvolume with U(3) gauge symmetry. Think of this U(3) as the freedom the strings connecting the branes have, on the worldvolume. There are 3 scalars coming from the modes with coordinates transverse to the worldvolume, which are the positions of the 3 branes in the transverse space. The process of giving some of the scalars expectation values i.e., moving the branes away from each other in the transverse space, breaks the gauge symmetry from U(3) to a subgroup. This is the Higgs mechanism, arising from diagonalizing the hermitian operator for the D-brane system. The hermitian operator's eigenvalues give the classical positions for the D-branes, while the off-diagonal entries of the unitary operators (short strings) give fluctuations about classical spacetime.
Applying the D-brane intuition to Smolin's model, he seems to have a type of octonion three brane model, with F_4 symmetry. As there are topological restrictions to any projective space higher than OP^2 with the octonions, Smolin's model is maximal, over the octonions.