Two years after Air France flight 447 from Rio to Paris crashed into the Atlantic, the French Bureau d’Enquêtes et Analyses continues its analysis work. A final report of the accident’s causes is expected in early 2012. What comes out of the interim reports so far confirms my worst fears: Murphy’s Law was in the front row and eventually stepped on stage to play the main role.

But there were two other major actors involved, and they are called human element and machine entity. The hole drama was without a doubt caused by the later; the plane flew into a heavy icing layer that nearly instantly disabled the pitot tubes. From then on, instrumentation provided the crew with indications ranging from marginal to plain erroneous.

But then, as alarms were echoing through the flight deck and the two co-pilots were trying to make sense of utter chaos, a peculiar thing seemed to happen: despite all the prior training, whether adequate or not, the human element was unable to talk to the machine, and vice versa.

When pilots learn instrument flying, they are taught that in unusual attitudes, their senses will unavoidably be confused and tell them lies. I have experienced it myself. While I was wearing a hood (a visor-like device worn over my head that would shield the outside view but allow me to see the instrument panel), my instructor Paul Savoie would place the plane in a very unusual attitude by combining steep turns, heavy G’s, pitch and speed changes and then ask me to take the controls and return the plane to straight and level flight. At that point, my brain would be listening to my senses and yelling that we were in a steep diving spiral to the left, yet the instruments would reveal a slow climbing turn to the right. And they would be right, and my senses, wrong.

So after years of training, an airline pilot has learned that the instruments never lie but the senses do. He knows he has redundant systems and that should one set of instruments fail, he can rely on the next. It all happens in stages, like a cascading succession of steps leading from the normal environment to a very basic, last resort one.

But what happens when all instruments have gone crazy? One’s senses are already going mad, middle ears being fooled by sudden acceleration forces and mystifying turns, and now, against all odds, all that training is proving to be worthless: the brain cannot seem to make sense of the many conflicting readings of as many redundant systems.

This might very well be what happened on board AF 447. Two first officers were on deck and although fully trained to act individually as Pilot in Command, they had no prior training in cockpit task management without the presence of the Captain, because aeronautical rules don’t require it. In other words, while they were probably very proficient at sharing flight duties with the senior Captain, they had never rehearsed, nor been briefed towards sharing emergency procedures with one another.

Then all hell broke loose. They were probably in the clouds, in darkness, with zero outside reference. The gauges went crazy, and they were left with only mad instruments and their suppressed senses to try and piece a puzzle together. They never did. When the captain arrived on the flight deck, less than two minutes after the initial disengaging of the autopilot, he didn’t seem to bring much cohesion to the scene. From the begining of the crisis at 2:10 AM to the final impact in the ocean four minutes later, about two thirds of all recorded voice exchanges on the flight deck were either questions or comments of uncertainty. Nobody knew what the plane was doing.

As data analysis is showing rather clearly, whatever they guessed, they guessed wrong. The plane ended up in a long stall and fell towards the ocean at above 10,000 ft/min. And until the last second, the crew of three remained unaware of the actions to take to save the flight.

Were they bad pilots? Most probably not. Could they have been trained better, or better prepared for such an eventuality? Most probably so. But then again, it might very well turn out that such an eventuality had never even been contemplated.

The Air France crew, on the night of June 1st, 2009, might have been up against pure evil: Murphy’s Law teaming up with Randomness.

To recover from the extreme attitude they unknowingly fell into, the three men would have needed one piece of equipment that currently just does not exist on commercial planes: a fully integrated and independent attitude, altitude and airspeed indicator. An instrument that would give a direct reading of the plane’s angle of attack, its speed over the ground, its altitude, rate of decent and bank angle.

I know, many of you are probably thinking or even saying out loud right now: “Oh but wait, planes already have that!” Sure, there’s a secondary artificial horizon for the right seat, and all instruments are doubled and fed by an alternate source; if the main artificial horizon is pneumatic, the secondary is electric. Altimeters are fed pressure by different pitots, and so on and so forth. Redundancy is the key.

The problem is, all these instruments rely on too few, and too similar a bunch of sources. When icing hit AF447 hard, it disabled or at the very least crippled speed indications. What other indications remained available to the crew is unclear to me, and possibly to the investigators too.

In any case what we need is a completely independent device that will supply pilots with heading indication (oh, wait a minute, my smartphone does that!), attitude and rate of climb (well, my smartphone can do that too), airspeed over the ground (funny, my phone also does that) and basic navigation positioning, and yes, my phone also can do that.


Ok, end of rent.

It’s not looking good so far and the human factor seems to be taking most of the blame, despite an obvious failure of all pitot de-icing technology. Maybe the guys screwed up. But it would mean all three of them did. Four minutes is a very short time to go from routine to the ultimate plunge. It’s also a very long time when your brain is working in overdrive and procedures are supposed to kick in. Why did they not? Was the information presented so incredibly wrong that it defied logic?

A stall is a stall. All pilots are trained from day one to identify – and recover from – a stall. Yet the crew onboard AF 447 did not. Why?