The FCC has the famous "protections" built in who restrict the inputs to stay within the programmed envelope. To restrict and direct, the computer needs sensors and those sensors can fail or ice up or etc. etc. The "feeling" of the computer is therefore wrong, just as the resulting restrictions of the inputs will be wrong.
So in short: HAL is still fiddling with flight commands even with the AP OFF.
I understand the fact that Airbus designed a "protection" software into their flight computer's program which is fine by me, however are you telling me that if a pitot tube of some other probe (a la AF447) freeze up or just fails to function properly the FCC will still take its inputs and get confused instead of disregarding them and work in a "less than ideal" situation?
If this is a fact then Airbus is pretty stupid in letting a "compromised" sensor render their "protection envelop design" and/or aircraft controllability vulnerable. The compromised sensor should automatically be taken out of the loop and a warning to the crew annunciated immediately.
Furthermore, if the whole FCC is somehow compromised, there should be an easy and quick way to disable it so that the pilot can regain full control of the aircraft (like in a conventional aircraft) albeit without the "protection envelop".
Finally, what I would like to know from an Airbus driver is assuming this crash was related to another stall like the AF447 flight, can a pilot regardless of the FCC's confusion just say "I have control", switch the AP off, the AT to off and FD if need be, fly straight and level and set a given N1 or EPR value to keep the aircraft from falling out of the sky?