Ladies and gentlemen, dear fellow PPRuNers.
First I'd like to remind you that "blame" and "pilot error" are not phrases included in any properly made accident report, (except in the legal notice clearly stating it is not the job of investigation board to blame anyone or anything). Fault is also seldom used and then only in mechanical or electronic sense. Accident reports only state that pilots did so-and-so and occasionally can add it was in contravention of such-and-such rule or procedure. This is what folks not very well acquainted with aviation safety, or more often with aviation at all, wrongly condense into term "pilot error", which by itself wouldn't be so wrong if it didn't always come with the notion that they who erred are the ones to blame for the calamity. Such a ignorance-based mistake usually comes from media or lately, bless internet, from anonymous fora.
Since AF447 crashed into international waters, BEA was appointed as the official investigator as the country of registration was France. In its final report, it has thoughtfully provided CVR transcript from the autopilot disengagement till AoA went over 40°, superimposed on some FDR parameters (page 60 English report, page 64 French), to make it more readable than it would be the case if one would need to constantly switch from CVR transcript to DFDR graphs in order to get exact chronology of who said and did what. NTSB style animation would be even better but I guess we have to do with what we have for the time being.
What can be seen is interesting, to say the least. CM2 has promptly arrested the roll, while unnecessarily pulling and properly announced he has controls. Next thing CVR recorded is stall warning, followed by exclamation of surprise from the CM1, next both pilots commented they have no display of speeds. To digress a bit: there was a theory put forward that they were unconcerned about sudden massive indicated speed drop but rather with characteristic protection speeds being removed from from the displays. To accept it as plausible, one has to be massively unaware of the importance of the IAS in any flying, let alone airline one. Basically: speed is life. For advanced users: while in itself it is life, we can live without it being properly measured and displayed.
So far so good, there are two pilots who promptly and correctly diagnosed the problem so what should have ensued is application of proper procedure, life goes on, no one notices except perhaps FDM, etc.
However, with the benefit of 20/20 hindsight we know that it took less than five minutes form the first sign of trouble till the aeroplane impacted the ocean.
Ten seconds after he concluded there are no good indications of speed CM1 warns CM2 to watch his speed. CM2 replies with "Okay, I'll redescend" as if he were warned of climb but he doesn't. CM1 now realizes they are climbing and only then prompts CM2 to descend. CM2 agrees but doesn't reduce the pitch enough to stop climb so busts practical ceiling. Aeroplane loses energy, stall warning goes off, CM2 attempts to pitch the aeroplane even more up so seals his and 227 other fates.
Not a reaction expected from professional pilots. Perfectly understandable if we assume ones acting so are scared mindless.
There was an occasional "we steam gauges pilots did so-and-so while these new EFIS kids on the other hand..." which usually made some sense but was more often than not garnished with a plenty of nonsense as it would compare theoretical perfect pilot of yesterday with the tragically underperforming modern crew, while the unspoken but presumed narrative would be that what was discussed were the average representatives of both species. There was many a classic cockpit crew who lost the situation awareness and ended smashed against the mountain or at the bottom of the spiral dive following disorientation as there is many a wonder-electric-jet crew that turned potential catastrophe into incident (e.g. QF32). So yes, there are still pilots flying those machines overhead us and given the number of incidents that did not end up tragic (or in newspapers or PPRuNe at all) I'd venture a guess they still outnumber mere system operators by quite a large margin. Now take this statement with a grain of salt: circumstances can catch any pilot out of his depth. Trick is to be so skilled, knowledgeable, conscientious and calm to make chance of it astronomically small.
It was mentioned there was thirty-something other incidents very similar to AF447 that ended without any damage or injury. While it can be used as a definite proof that Airbus is not lethal by design, conclusion that other crews knew what happened and what they are supposed to do is so far-fetched to be patently untrue. This is the tragic part: not every did but they did manage to avoid the traps that AF447 failed into. Many survived by doing nothing while trying to figure out what is going on and so exited the area of ice that clogs the Thales pitots in the process. There were those who pulled but they didn't ignore stall warning so pushed. There were those exposed to brief stall warning as they hit updraft. After every updraft must come a downdraft so they, wrongly but not fatally, assumed warning was false.
There was many a heated argument of how this or that automated feature should have been incorporated into Airbus to prevent the CM2 from wrecking the aeroplane. Well, Einstein once observed that the good thing about thought experiments is that they always succeed. In real life, safety devices have to be designed (and demonstrated) to acceptably cope not just with the occurrences thy are supposed to deal with but also the two failure modes: 1) failure to work when required 2) activating when not required. Number 2 is dealt with by system being overridable (stickpusher) or shutting itself down when risk increases (Airbus protections and control laws). Certification authorities assume that even the crew that has barely passed the obstacles of the type rating course has good situational awareness and will recognize failures and react to them properly. There is never an assumption that crew will get totally incapacitated for prolonged period. Pilots that get detached from reality are faced with simple choice: timely regain SA and act correctly or meet thy maker. Airbus protections (which incidentally started with A300, only got more sophisticated with introduction of FBW) can only buy a bit more time for crews to regain their wits but it was repeatedly demonstrated that you can still crash while staying away from protection parameters.
What are the chances that the pilot who gets so freaked out to forget just about all the basics of flying - pitch & power, performance ceiling, that heavy buffet and failure of the aeroplane to respond to controls is indicative of stall (since warning just got ignored) would pay attention to AoA gauge or use manual pitch trim? In real world: zilch. We can indulge in wishful thinking if we find it emotionally satisfying but it won't prevent the recurrence of AF447.
There was even mention that there is no feedback from aeroplane to pilots in Airbus as sticks are not backdriven and this is supposed to be major design flaw. Well, we have mostly abandoned the feedback fifty years ago when we made the switch from power-boosted to power-operated controls. If you fly aeroplane with hydraulically operated controls and believe what comes through yoke is feedback, I am sorry to disappoint you but you have been lied to. It is synthetic pitch feel. It is there to prevent you from overstressing the airframe. It can provide clue how far you are from the trimmed state but it is not by design or purpose and folks who are mislead to believe they can use it to tell the speed error can get bitten when things get a bit pear shaped as almost was the
temporarily hapless crew of G-CPAT. They were so concentrated on yoke feel they at one point believed they had unreliable airspeed - despite all three indications agreeing. Good thing they maintained healthy respect towards GPWS.
ECAM complications are another red herring. No abnormal checklists, no ECAM actions, no memory items, there's nothing that has be done before positive control is established (to be nitpicky: except items that prevent loss of control such as overcoming control jam or feathering the propeller that went into ground fine, nothing similar was involved in AF447) which seemingly never was.
It was mentioned that AF447 is a proof modern pilots have insufficient manual skills and that we should practice raw data manual flight more often. While I agree with the recommendation, I don't think it can be derived from the accident we are discussing. Indeed, most common contemporary accident scenario is not involving a crew that knows what goes on and what needs to be done but lack of manual flying skills prevents it from carrying out the plan. More common is the crew who loses SA, often through some minor and trivial distraction and does exactly the wrong thing (sometimes with astounding manual dexterity) so loses control. I am afraid manual flying when everything goes right does exactly nothing to prevent such a calamity and introducing distraction during simulator won't be much helpful either as basic limitation of it is that it is still a sim, it cannot simulate the feeling that you are flying for your life. IMHO problem here is the pilot that has many an hour and many a simulator session under his belt. He goes to the sim, does the motions, passes the checkrides while secretly harbouring deep mistrust of what he has been taught as he "knows" better than some manual written by the lawyers for the aeroplane made by the pilot-haters. So one day proverbial hits the fan and that's when he realizes that he has nothing to fall back upon.
Question that remains is how to prevent AF447 from recurring. Human factors are definitively the key but methinks in that aspect, BEA report leaves a lot to be desired. While I'm no expert in psychology so I can't meaningfully comment some findings from that aspect, I do know that notion that night makes maintaining attitude in passenger jet more difficult through lack of outside horizon to be utter tosh. You can combine it with the best psychological expertize and still you won't get anything true out of it. Even worse is lack of background data for the pilots involved except the very basic information. BEA should really follow the NTSB example with its background checks. We got informed that F/O of AA587 really misunderstood AAAMP to imply that it is proper to use rudder for any disturbance, was already warned by a captain on one of his previous flights, but was unable to comprehend. NTSB told us that captain of Aloha Islandair 1712 regularly scud-ran, he made wrong estimate of his position only once and that was enough. As for AF447 pilots, we have no idea whether there were precursors noted during their careers that would make their reactions more comprehensible or - far more scary option - their de-structurization and consequent disaster struck out of the blue.