To SyEng and others
As far as the unlikeliness of a pilot not pulling to idle, the T/L of the engine with the deactivated T/R: it has happened twice before CGH with accident consequences, and has happened even more times with only slight uneasiness for the pilots as a result, as, for instance, described by 4 Holer Poler on 23 or 24 august (page 22 or 24 of this thread).
As for the post which described how much design effort has been put into the development of the non-moving thrust lever system - effort is not yet a guarantee that it is right.
Look at the FCU (Flight Control Unit; in Boeing called MCP; for others, the glareshield panel). Airbus philosophy there is - push a knob for computer generated values (managed modes), pull same knob for pilot values (selected modes). Well, within that one panel they managed to make the function of the baroset knob just the other way round: push for pilot value (QNH), pull for standard altimeter setting!
Also, the vertical speed selector: in order to conform to the above stated philosophy, it could not be in the form of a "wheel", it works on the same principle as the heading selector. For a heading selector that is a natural control, turn left for a left turn, but for vertical speed there is no direction of the selector that is naturally connected to the aircraft reaction. Turn left for descent, turn right for climb is a learned trick that requires cognizant activity of the brain every time the control is used (not sure whether cognizant is a real word, hope though that the meaning is clear).
I know that these controls have nothing to do with the accident at hand, I have only used the examples to show the scientists and designers that wrong things can very well exist in an aircraft, even if it used succesfully for millions of hours.