Theshadow
It's a lot of words but if I read you correctly your thesis boils down to:
(1) Plane runs into problems gradually.
(2) Autopilot deals with problems gradually
(3) gradual problems accumulate until autopilot is overburdened and disconnects.
SURPRISE!
(4) befuddled pilots who didn't even know anything was wrong get disorientated, or confused, or lose spatial awareness, etc.
(5) Plane crashes.
With respect, that's not exactly a new thesis or a new problem. I don't know if it's true in this case; it wouldn't surprise me given the facts we know.
No, it's not new, agree. But it has not been tackeled, at least not effectively, only by lengthy and a$$ covering bulletins.
I have for a long time criticized the described chain of events. Once the autopilot can no longer cope with a situation, it simply throws the aircraft back at the pilot. Good thing for the company and manufacturor, because they can always detect pilot error as a cause for the accident.
This happens on B's and A's. What bothers me is that on the A's, even when throwing the problem at the pilot, the automatics still interfere with commands, through protections or limitations etc.
With the B's at least you can very rapidly oversteer with some force. That's what many have asked for an eternity now to be implemented in A's: A rapid automatics disconnect button.
To what extent this might have helped in the AF accident is doubtful. The main problem is that a manual recovery is poorly designed and even more poorly trained. The sims are simply not programmable for such manoevers.
I have witnesses sim sessions with windshears, upsets and RAs that uncover a frightening lack of flying skills in many pilots. Just take away the flight director and autopilot on a simple take-off, give an early level-off combined with a turn and observe .....
Having worked in three companies that all propagated manual flying skill sessions, the outcome was almost consistently very poor, but ..... They all reacted by simply abolishing these sessions.
Why?