Hi Olaf,
I know about Milgrom's result but I have not seriously looked at the consequences for SD. I had one crazy idea recently (if you'll allow me to indulge ;-). The conformal group in 3d is SO(4,1), which is the isometry group for de Sitter. Thus, I think a natural action for SD is the one of Stelle and West that uses an SO(4,1) connection because it can be decomposed into a conformal geometry in 3d.
Now, how do you couple fermions to this action? You can't use normal spinors because the space is locally de Sitter NOT locally Minkowski. Thus, you shouldn't use spin 1/2 reps of the Poincarre group but rather spin 1/2 reps of the de Sitter group (which, is isomorphic to the conformal group). But, because the cosmological constant is small, these "dS spinors" should be effectively the same as standard spinors, at least for particle physics experiments. You would only notice a difference in the dynamics at cosmological scales related to the cosmological constant (because this is what distinguishes the dS group from Poincare). But the MOND scale is the cosmological scale! So maybe you would expect MOND like behavior from the SO(4,1) spinors?? And maybe the relation to scale invariance is because of the isomorphism with the conformal group??
I don't know... but I'd like to look into this at some point! Did that make sense??
I take you're point about the cosmological constant problem. You're right that the story might change once we have a good theory of quantum gravity. Now I understand your point. Thanks!
Cheers,
Sean.