I'm most interested, as you might imagine, on how Sanctuary will implement the AI systems - which I know, is a tremendous parcel of work, not to be taken lightly. It's also not something that can just be 'tacked on' after the fact. While the design of Supcom, which was ahead of it's time, did well to lay out the data structures necessary to having a decent AI - the development on top of that structure was almost an afterthought, and was almost completely redone between the original SC and SCFA (Yes - I was already lurking about in those days and had many discussions with Sorian about moving the state of things along.
So - it interests me in how that data framework is prepared, not so much in the AI itself. I'm still highly active in the development of LOUD, where, even after almost nine years now, we're finding new metrics out of the old dataset - allowing this venerable, but technically limited environment, to provide new levels of competition which weren't considered previously.