On June 1, 2009, Air France flight 447 from Rio de Janeiro (Brazil) to Paris (France) suddenly entered an aerodynamic stall and crashed into the Atlantic Ocean, killing all 228 passengers and crew onboard. Although the initial reports on the possible causes of the disaster evolved around the malfunction of the plane’s pitot tubes, the investigation concluded that the main reason for the tragedy was pilot mistake.

Tom Dieusaert, a Belgian journalist and writer, does not believe it was quite so. In his new book “Computer crashes: When airplane systems fail” he argues that pilots are commonly blamed for tragedies like the Air France F447, but something more subtle – the plane’s computer software – is often overlooked.

AeroTime had a chance to talk with Tom about his book, plane crashes and the computer systems that control modern airplanes. 

Your book is titled “Computer crashes”, but the accidents you cover, most notably the Air France Flight 447, are officially blamed on the pilot. Why did you choose to focus on the technology rather than human factor?

Air France 447 which was actually the case that caught my attention and inspired me to write the book. There were a lot of on-going problems with the airplane – and the pilots just were the last ones to solve those problems. In my opinion, the pilots should have never been put in the position to solve those problems at the very end. There is a lot of emphasis on what the pilot should have done. But there is not enough emphasis, which I tried to put, on all the troubles the machine already had.

The concept of “human error” is still used a lot in the press but the human operator is only one of the many defenses the system should have.

The pilot is not flying the plane anymore. The pilot is just there to solve the problem, when everything else fails, so he is the last resort. I think it's a bad analysis to pinpoint the pilot and just forget all the other trouble that machine had. But of course, the industry doesn’t want a critical eye on the problems their planes have, and they find it easy to target the pilot, the one who is flying the plane.

I’ve spoken to quite a lot of pilots about this accident, including the pilots that fly the same plane – the A330 or the A340, which is almost the same thing), and they said if you put 10 people in the same situation, 9 out of 10 would have done the same mistake. That’s what is called retrospective bias. We look backwards and we say that the pilot should have done this or should have done that. It’s not very easy to put ourselves in the place of those people when they had a lot of contradictive information on their instruments.

Tom Dieusaert is the author of “Computer crashes: When airplane systems fail”

In your book you say that computers behaving erratically or malfunctioning are quite common, but the reason the public doesn’t know about it is that it is not advertised (or even acknowledged) by the manufacturers and more importantly, the safety boards. What is the motivation for authority bodies to ‘keep a blind eye’ on the technical issues found on planes?

Let’s take the example of the Air France accident, where, I believe, personally, there was a pilot error, but I do not want to say it was the cause of the accident, it was just one of the whole chain of things that went wrong. And the biggest part that went wrong were all the previous events, also the design of the airplane. For instance, those pitot tubes were not functioning well and had not been functioning well for a few months before, two years before the crash. Airbus had a lot of issues with those new pitot tubes on their A330s. There had been events, 20 or 30 events in 2 years. One of these events went wrong, which was the Air France one. Air France took 6 months to change these pitot tubes on their plane, and the plane that went down was going to have them changed the day it was supposed to arrive to Paris. So it is actually a mistake of the company. That plane should not have been flying at all. But afterwards the blame was put on the pilots.

What about the sort of flying agency that should be supervising errors on planes? In France there is an organization that is called BEA (Bureau d'Enquêtes et d’Analyses), which investigates air accidents. First of all, they don't have all the information. A modern airplane has like 500 computers, a lot of software inside, and the agencies don’t have all the information. After the crash they can only listen to cockpit voice recorder (CVR), or the flight data recorder (FDR), if they find it. They don't have all the information on the codes of the software, so they don’t know exactly how everything works. That’s one reason why they cannot be very critical.

And there is another factor, which is also very important. One lawyer I spoke with, who is actually advising the families of the deceased people on the Air France crash, told me, there is a co-dependent relationship between these organizations and the big airplane constructors. For instance, there is one investigator who was on the Air France case, and who is now working for the BEA. Many people who now work in such organizations first have worked in the aviation industry, because they know it. It’s a small world, so it’s obvious that they will not make a very hard decision against the airplane constructors. For me, one of the most striking things about the history of aviation is the fact that for instance, Airbus has never been convicted in a trial. For instance, car constructors have been, like in the case of Volkswagen’s Dieselgate. It never happens with the airplane constructors, because they always get off the hook and always say “this or that accident was a pilot error”

If you look deeper, you can see that after the Air France 447 incident the software was changed on the pitot tubes and the pilot training was changed also, by which they admitted that they had wrong conception there. Personally, I think this sort agencies are quite blind on the bigger aircraft constructors, because there are just too many problems at stake.

You have mentioned that air safety agencies do not have all the information on software and codes. Planes are heavily loaded with tech, but does it get proper maintenance? For instance, does the plane check involve supervision on the computer software?

I have issues with my computer, you have issues with your computer, so it is completely imaginable, that with 500 computers on one plane, you are going to have issues, I mean they are there all the time. What I see, is the big companies never admit it, and they do not do it because they don't want to scare the public. And I think many times there are also minor issues, there are issues that probably do not affect the flight, but when they have an issue they simply don't admit it, and I think that's the reason of my book. I made an investigation and I want to make it public that they should admit the problems they have to enhance safety in general.

Tom Dieusaert, "Computer Crashes: When airplane systems fail" available to order at amazon.com

For instance, the problem they had with the pitot tubes that were installed on the A330s, they were working before very well from the beginning of the year 2000 until 2007-2008. And then suddenly they started working badly, the heating system, so very probably it has to do with the software, which operates the heating system, because the pitot tube itself is just a mechanical device, which doesn't change a lot, it hasn't changed a lot over the history of the aviation. So probably software changed there. And neither Air France, nor Thales want to comment on that. I would suppose that Airbus would say the problem was Thales' fault. And one would suspect Thales to say that the problems lie with the Airbus software, but none of the two companies issued any statements about the recorded flaws on the machines. Both company spokesmen I spoke with say that they “cannot give any information” because there is a court case going on, which is supposed to end in 2019-2020. That's strange for me, because it’s even incorrect - they refer to an ICAO agreement which has nothing to do with the criminal case. And there is no certifying agency which digs a bit deeper and really looks where the problem did occur.

Of course, software is checked every time, they also have redundant computers, which means, if one computer fails, another one takes over. But there are always situations, which I would call the worst case scenario that in some point something happens when all the bad conditions come together and latent error, which was not apparent before, comes to the surface. For instance, with the Air France crash, the flight directors behaved erratically because of the false speed reading. And while most of the younger generations of pilots fly a lot with the flight director they also should look at the standby instrument, they let themselves be led by the flight director. When the FD gives the wrong information, that's understandable that the pilot makes bad, decisions, like that pilot on the Air France made and gave too much pitch up and put the plane into the stall.

This is the first part of the interview. Stay tuned for the second part, which tackles the question of whether computers will replace pilots in the future.