Dozy:
At this point, BEA and Airbus know what was being displayed on at least one set of instruments, they know which versions of the software was installed on the aircraft's computers and they know what the timestamped ACARS messages were.
That's enough to gain a reasonable idea of what that particular aircraft did in terms of systems behaviour, over and above what control inputs were from the pilots.
Up until stall, yes.
How they determine validity of airmass data will be an interesting revelation, given spurious airmass data as contribution to the event.
This should be enough to run simulator tests with crews to gauge their responses to the situation that presented itself, in much the same manner as the Birgenair investigation did. The systems are not so complex that recreation of the incident is impossible.
Yes, except for the simulation of what happens once the aircraft stalls ... which goes back to "how do you train to get out of a stall once in?"
Up to that point, both stall prevention and error/trend detection processes and habits may be usefully analyzed.
There is a danger in making assumptions about what was on the PF's display set ... maybe BEA can figure out and publish more of that between now and when their final report comes out.
Also, crews put into the scenario ought to be unalerted, if you want to collect some data on the man-machine interface. <--- That appears to be a critical performance node that must be examined.