so given an h-flat g-torsoroid (or something...), can we get some sort of distribution on it (or something... ??hmm, perhaps on the associated "cartanian g/h geometry"??...) by taking its "excess flatness subalgebra" (or something...)? if so then does this deserve the name "the cartan-rolling distribution" (or something)??

(i forget whether "distributions" are usually allowed to vary in dimension from point to point...

also, what about ideas about "the symbol lie algebra of a lie algebra filtration" here, and so forth? something about nilpotence here and so forth...)

which reminds me that i should try to organize a list of questions to ask derek wise about...

1 the above question...

2 something about graded nilpotent lie algebras and their relationship to distributions...

3 do you know anything about "cartan's method of curvature invariants"? (or whatever the name was that bor and montgomery (i think it was them...) called that method...)

4 something about relationships between "de sitter relativity" and segal's cosmology... or something...

5 have you thought about the (degenerate?) case of cartan geometries modeled on an arbitrary manifold considered as a homogeneous space of its diffeomorphism group?

6 something about invariant contact distribution on coadjoint partial flag variety...