Interesting paper; now more than ten years old. It's most recent and vivid accidents being Colgan Air and the Hudson River ditching. I would hazard a guess and say that the same people interviewed today would have an even more pessimistic view of the industry.

The problem with increased mechanical reliability, more advanced automation and lowering of T&Cs is that the job doesn't need sharp people anymore. However when things go wrong; those same people are quite clearly out of their depth - this incident and the French Bee A350 and now perhaps the Spicejet in 'turbulence". How is the regulator supposed to get ahead of these 'black swan' events that are by their nature unpredictable and infrequent?

It's simple; the cost to do something about it is is higher than having a hull loss every x million passenger miles. If that relationship reverses; expect something will get done.

Final BEA report

Apparently both crew members applied simultaneous control inputs for 53 seconds resulting in the controls being desynchronised for 12 seconds.

but also :
and under Safety measures taken after the incident :

In your opinion, what was the classification of this incident according to the ERC matrix ?
In my opinion, it lies between 2 and 21.
https://www.galiot.aero/images/resou...sification.png
The flight path was never really dangerous. One call from one pilot was enough to resolve the issue.
There are other systems that help prevent escalation into more serious trouble (EGPWS, TCAS,..)

I have the same question, regarding crashes this time. Ask the exact same questions, however take the accident scenario for question 1, and what should have happened for question 2.
Then go into the matrix. For example AF447 should have been a 502, at worst. Concorde crash was a 2500.
However, some deadly crashes should have been classified 50 or 102. There are many examples where lots of barriers synchronised to be all defective together, whereas they shouldn't have been.

One interesting thing in the analysis is that the incidence of dual control input in the AF fleet was broadly similar between the A & B. There was much talk around AF447 that it wouldn't have happened with coupled controls, but this incident suggests that it might not be as clear cut as that - that once communication and situational awareness break down, the coupling of the controls may not rescue the situation.

From the report:

The first two points pretty much sum it up IMO.

I'd argue that the simultaneous inputs were perceived almost instantly by both pilots, but due to the general level of arousal attributed to anomalous behaviour of the system rather than input from the other side. I would also point out that it’s fairly obvious when someone has their hands on the controls as they’re sitting next to you - I feel straight away when people use the yoke switches while I’m flying and they get asked to use the ones on the coaming or centre console.

Pretty thorough report, glad to see the 2-crew night Atlantic 24hr trip getting a bit of a spotlight. Try doing six of those a month for six months. Overall though, no surprises - decision making can be collaborative but manual flying is a solo operation pretty much all of the time.

I am surprised that the deviation from the stabilised approach which is at the inception of this incident is not analysed in this report.
The report states: "with a 6° left bank, and the wheel 17° to the left, the PF called out 'go around'", 6 seconds after stating that the bank angle was "strange".

I guess it means that the PF suffered a somatogravic illusion possibly coupled to some fatigue.
And I understand that, since this deviation has been followed by the correct response, going around, it is not considered as serious as the double input on the commands.

However, in my opinion, it is worth analysing if it was indeed a somatogravic illusion or if it was something else, and checking what might have triggered the illusion at that time, 2.1 nm from the threshold.
Because there has been and there will be other losses of spatial awareness in short final followed by an incorrect response and more tragic consequences.

It's weird isn't it? Literally no discussion of how the approach originally went out of kilter.

I don't think we share the same understanding of how the ERC works, but I stand to be corrected. It doesn't work for accidents that have happened. It starts with Q1 (funnily enough) which asks what would have been the likely outcome if it escalated into an accident. The most likely outcome of disorientation IMC and at low level is LOC-I, which is going to be catastrophic. Only then do you start to look at the effectiveness of the remaining barriers. For this event, the barriers operated but it was limited effectiveness in my view as the control wrestling went on for a full 53 seconds.

If the BEA didn't think it could have escalated into an accident scenario, I suspect there would not have been a full-blown Annex 13 investigation.

Now we know the reason why it started. PF thought 5G had disrupted his control of the plane. It all went downhill from there.

​​​​​​Boeing ​connected control columns didn't help confusion over who was doing what.

All discussions on connected controls "breakout" feature can be boiled down to training IMHO. I cannot believe pilots were not aware of the existance of this feature. However this feature is "presented" in training only as a "solution ico jammed controls". I personally admit I never thought it would be possible that "fighting pilots" would create a similar effect and breakout could be achieved in such a scenario as these forces are fairly high.

A simple demo of both pilots giving opposite inputs in training would create the necessary awareness and avoid future problems.

​​​​​​​Egypt Air B767?

Maybe the wording ”fighting pilots” was a bit too strong. Wasn’t referring to 2 pilots really fighting for death vs living, but 2 normal yet confused pilots simply not aware of eachothers actions. As the recorded deviation in bank/pitch of 767 vs 777 is quite different, you would expect the ’applied forces by the pilots’ to achieve a correction to be quite different as well (hence a lot lower on the 777 flight)

Reminds me of the training exercise we did on the DHC8. Came with PULL-TO-DISCONNECT for pretty much all controls in case of flight control jams. This was something of a CRM exercise. Instructor would jam the ailerons, disable the spoilers and the rudder (I think total hydraulic failure). Only thing working was the elevators. If you disconnected the pitch control, and had one pilot pitch and power for altitude, and the other pitch for angle of bank you could fully control the aircraft. Bonus points for differential power in the turns for turn coordination. Don't think it was ever needed in real life, but great to see how important working as a crew is vs not knowing what the other guy is doing.