If MS wants this to be a succes the it'll have to be:
- As easy to expand as the original. I don't agree with the author that it has to work with the old ESP framework as anyone who worked with it will know it was clunky to say the least but it can improve on this with actual documentation (Free and openly available) and use of real world units instead of weird scalars that affected 3 different aspects of a model at the same time (an not mutually exclusive of course).
- Run offline without requiring an internet connection
- Provide "good enough" flight physics out of the box. No-one into serious simming ever accused the MS flight models of being accurate, but that's where point number one, add-ons, comes back in.
- Provide the same scenery wise: Good enough elevation data, some detailed large airports around the world, the full Jeppesen database reproduced procedurally (not super accurately, leave that to the community and industry)
- Priced attractively enough for the general public (80 Euros max for a single not-for-profit license) and for businesses (Offer cheaper single seat, single user commercial licenses and the like)