On the last day of the IMPRS summer school, Bovy gave a detailed talk about dynamical inference in the Solar System, our April Fools' paper. He gave a very nice description of not just what we did but also why it works and what are the limitations. I am getting steadily more pessimistic about dynamical inference with Gaia, at least with phase-mixed populations. Populations that are not phase mixed, such as tidal streams, are seeming much more promising at present.
All my presentation materials (somewhat disjointed) are available in this tar gz file. For a non-attendant at the summer school, the most comprehensible part will be the problem sets.