I quote from the 2nd BEA report:
"The reappearance of the flight directors on the PFD when two air speeds will
be calculated that can lead the crew to rapidly engage the autopilot. However,
these speeds, though of the same order, can be erroneous and weak and thus
lead the autopilot to command movements of the flight control surfaces that
are inappropriate for the real speed of the airplane.
In case of automatic disconnection of the autothrust with activation of the
thrust lock function, the absence of appropriate manual adjustment of thrust
can present a risk of an attitude/thrust mismatch, especially when this
disconnection occurs with a low N1 value."
The same report, speaking of 13 cases of A330 airspeed anomalies:
"In seven cases, the autopilot was reconnected during the event. In two of
them, the re-connection occurred when the two speeds were consistent
with each other but were erroneous".
The A330 stall recovery procedure is to apply takeoff thrust and full aft stick, relying on the computer to maintain maximum allowable AOA. Without FBW, no pilot uses full aft stick to recover from a stall (this is usually reserved for entering a spin!). This practice is unique to an aircraft with FBW and essentially puts the pilot in a position of relying on the computer to prevent the stall (because the pilot's control input is actually trying to cause the stall). In this case, the computer may have been relying on airspeed and altitude data from 2 consistent but erroneous air data inputs and computed a maximum allowable AOA that corresponds to a lower altitude. This would be a higher AOA than should be allowed at FL350 and if full aft stick were applied would lead to a full stall. Depending on the aircraft's actual indications of a stall (something the pilots would not have experienced in training since the aircraft is "unstallable") and the fact that it was night in IMC and turbulence the crew may not have recognized the nature of the full stall and believed that the computer was preventing a full stall. I think this could have set up the situation of a deep straight-ahead stall all the way down. By the way - in the stall the drag is much higher than normal and lift much lower. This combination creates a very high sink rate, much higher than an idle power nose down descent from cruise. I think a rate of >10,000 fpm is very possible. Someone calculated a rate of 18,000 fpm from the reported information.