(It would also seem that Airbus now has at least two sets of pretty good data on entry to stall, stall behavior, transition to deep stall, flat spin, 10,000+ per minute altitude loss.... to add to their simulator's knowledge base.)
I was thinking about this too. Does the FDR from a crash that involves a stall help improve the simulated responses to a stall?
The answer is obviously yes, but how many sets of data about performance outside the envelope (i.e. flights) do you need before you can program a simulator to correctly simulate performance there? What does it take to make simulations significantly better? My gut feeling is "a lot".
Any folks who build simulators out there? Any test pilots who can speak to this?