We've discussed core technical issues like the component model abstracted around "Bricks", with some AOP-like cross-cut concerns added as "Natures". This was in Java.
Them some discussion on data-model separation. Currently it is tied to the component model, by defining POJOs that flow in-and-out the Brick mandatory public interface.
Them meta-programming, to substitute current "Nature", run-time injection of behavior with more debuggable Macros, but as that means using another language (Boo migrated to JVM as Boojay) it is not clearcut if the Sneer coders will go down this route.
After suffering some "Rock Band"-ness, I swayed to the sovereign games development session, where discussions just kept the big question unanswered, what kind of game would be a sucess in a Sovereign landscape.
Finally, we got back to the final all-voices discussions on the merits of pursuing such a project and how to make it effectively grow up. Most people feel that a killer app is needed, and that portable protocols should be defined to enable developers not to be forced to use Java or more generally just some specific language/platform to develop Sovereign pieces and apps.
I, for example, still prefer to implement my sovereign bits in C#,Boo, or probably some Mono-based DSL I'll cook in the vein of Intentional Programming or Reinvention Of Programming