I believe the fundamental cause was that the computers flying the plane used invalid air data inputs (altitude and airspeed) with pilot control inputs that were based on the assumption that the computers had valid air data inputs (altitude and airspeed). I believe the FBW was not designed or tested against this particular set of inputs. That is based on my software engineering expertise and not so much what I know about flying.
Nope. The AP and AT disconnected exactly as designed and the PF reported "I have the plane".... at this point the aircraft went where the human pointed it. It was pointed up (and left I believe, in response to a right roll). The computers at this point applied the design protections (where possible) based on Alt Law being in effect, but did NOT control the aircraft in any rational sense of the word - that was the bloke moving the joystick around. Unfortunately the human reaction does not seem logical to (most) pilots here and the aircraft ended up in a pretty bad attitude. (Nose high, AoA very out of shape, and minimal airspeed).
Why is this so hard to understand?
