Hacker News Reader

Comment by FabHK | Hacker News Reader

> What's more, over the course of his 55 years in the profession, he [lawyer Marc Moller] [has] learned that every plane crash can be traced back to a single, simple cause.

I don't know, that strikes me as nonsense. It's precisely the interplay of many factors that lead to an accident - and if any one of them had been different, the outcome might have been different. See Reason's "swiss cheese" model of accidents - all the holes have to line up [1].

How you can go and designate one of these holes the hole, the "single, simple cause" is beyond me. This doesn't do the complexity of engineering justice. Might help to win cases with juries, but I don't see how it could help to make flying safer.

[1] https://en.wikipedia.org/wiki/Swiss_cheese_model

View on HN

Forgive me a great deal of cynicism here. I suspect that a more accurate statement is "he learned that it works well for him to argue that every plane crash can be traced back to a single, simple cause."

We don't ask lawyers to do surgery. We don't ask them to design bridges. And we don't ask them to do root cause analysis. We leave all of those things to those who actually know what they're doing in those areas.

Yeah, but the work of lawyers has often forced industries to do things differently, even when the best science says they shouldn't do it that way. For example, lawsuits over botched baby deliveries that argued "you should have c[aesarean]-sectioned" resulted in extreme overuse of the procedure in the US, with higher resultant risks.

So ... I wish it worked like you described.

I suspect the reasons for increased c-sections are manifold.

When we medicalise pregnancy and childbirth, then fill hospitals with profit driven policy, and surgeons who fear malpractice accusations, is it any wonder c-sections can become over used?

Fear of malpractice alone can’t explain it all, as it some other lawyer could demonstrate c-sections are over used and lead to complications.

I don't think the parent is arguing malpractice fear explains it all. But clearly, that fear is a significant element in many decisions.

We don't ask lawyers to do surgery

No, but when they're in a court debating about e.g. doctor's negligence, we do expect them to have a rudimentary understanding of what's going on.

Generally, they find, and pay for an expert to argue whatever point they want.

The great thing is, if you pay enough, you'll find an expert being willing to argue anything you want.

Yes, but that works for both parties, though.

The expert more capable of conveying their message will be the one who gets the most credibility, and this often (though unfortunately not always) aligns with the expert telling the factual truth.

> learned that every plane crash can be traced back to a single, simple cause.

That's silly, but if I were playing that game, I'd say the single, simple cause was a failure of Boeing and the CIA to keep abreast of A320new development.

Boeing were caught flat-footed by the launch of he A320neo and, facing a loss of potentially thousands of sales, were forced to quickly develop and release a modified 737 rather than create a new plane altogether.

Corporate espionage would have prevented all this.

> Corporate espionage would have prevented all this.

I believe Boeing knew exactly what was going on with Airbus. US does corporate espionage to favor domestic companies, and creating a new plane is such a big project that there's no way to miss that.

> US does corporate espionage to favor domestic companies, and creating a new plane is such a big project that there's no way to miss that.

Pardon my ignorance but I thought the Chinese government was the one doing this sort of thing. Surely the US government does not engage in espionage on behalf of it's companies

Friend of mine pointed out.

Scientific facts aren't the same animal as legal facts. And I'll add not the same as political and engineering facts either. Root cause analysis facts are social and engineering facts.

You don't need to be able to analyze causes in order to count them.

No, but if you want to prevent future incidents, it usually helps to have a more systemic view of failure. E.g. questions like: “what kind of management lead to decisions that made such misbehaviour possible?” are usually completely uninteresting for a lawyer, as they either play a legal role or not.

In real life a incident may be the emergent behaviour enabled by complex circumstances and a certain culture, all culminating in one crystalized point of failure. Think of causal chains: Managment pressures engineers and fires experienced older engineers, engineers are clearly not up to task, do something. Qualitt control might catch it but is circumvented by other managers who fake documents. The thing is not up to speck but wouldn’t have failed if the pilots... etc.

This is what I mean by systemic emergent behaviour. All of these things were wrong or wreckless, but pinning it to one cause isn’t doing reality a favour here.

And that's all well and good ... but his role isn't about preventing future incidents, it's seeking restitution for the victims and their families, nothing more, nothing less.

It is the job of regulators to figure out how to prevent future incidents, and it seems that the EASA will be taking the hardline lead on that as there seems to be a general mistrust of the FAA this side of the pond.

No. But I've (as an amateur) looked at crash investigation reports, and they usually state more than one cause.

I also wanted to mention this after reading the paragraph, but it is a pretty unimportant side note in the article.

The framing here is obviously from a lawyers perspective. A lawyer that has been very successful at extracting money from companies.

Making the issue appear simple and traceable to a single problem seems like a great way to convince a jury.

> Making the issue appear simple and traceable to a single problem seems like a great way to convince a jury.

Or, an electorate. Maybe oversimplification is the path to popularity.

Telling a good story. Sometimes that means oversimplifying.

I think, as a lawyer, he's using the phase a "simple cause" to mean "a singular responsibility, culpability, and liability."

Probably. Goes to show that the lawyer representing the victims and the safety folks (NTSB etc.) trying to make sure an accident does not repeat have very different goals, incentives, ways of thinking, etc.

Given how vague the language is, I don't know that we can necessarily draw a conclusion.

That "single item" might be the component-failure, that which fell through the aligned holes in the swiss-cheese, or it could be "the single layer of the system that was left out" that would have mitigated or otherwise precluding the class of failure.

One could castigate Boeing for not having a more redundant use of sensors the MAX case, or one could say "Why don't you have 'hackers' as a part of your process, whose job it is to find these classes of problems and raise all hell when they do."

Then again based on the phrasing "single, simple cause" I'm more inclined to say you're on the money. That the lawyer is looking at the component-failure rather than "avoiding non-redundant systems" as a class of problem to solve.

The single simple cause he may be alluding to is greed.

Lawyers aren't interested in the technical minutiae except in as much as it can be useful in communicating intent to the jury.

"Why did you decide to only use one sensor to feed such a critical system?"

"Why did it not occur to anyone to reevaluate the safety rating when the software was changed?"

"Was there any pressure to not do as thorough an analysis as possible?"

What I would do as an attorney is cross examine everyone involved to establish that there was a clear and present intent to put the completion of the product on time ahead of completing the project in total compliance with federal regulations. It's proving intent to the jury that matters. Not arguing over the finer points of aircraft design. In fact, I'd be surprised if anyone with an aviation background would even make it passed voir dire unchallenged.

Yes. He's certainly not going to approach this like an engineer going bug hunting.

The goal will be to establish that the problem was cultural and political. An avoidable cascade of engineering failures and regulatory misrepresentations was the result, not the cause.

Yeah, I think the language used there just isn't being that precise, by engineering standards, that's all.

When you're assigning blame for something like an aircraft crash, what you want to single out is errors which were preventable. So, for instance, icing on critical sensors probably isn't preventable 100%, so that's why we do things like have redundant sensors, algorithms about how to use that sensor data to avoid letting a bad sensor reading cause a big problem, etc. Another thing that isn't preventable is weather, of course, so we engineer things to handle that. But preventable errors, especially ones which we've already learned how to handle in the past, are generally considered unforgivable and worthy of assigning blame. A crash caused by a Russian AA missile isn't really avoidable for a passenger jet, so we don't blame the pilot or the aircraft maker or airline. A crash caused by using a single sensor that's known to ice up and is used for a system that's only there because someone thought it'd be a great idea to slap some oversize engines on an obsolete airframe from the 60s in the wrong position, however, is clearly a preventable error.

"It's difficult to get a man to understand something when his salary depends on his not understanding it."

The quote from Marc Moller seems counter to what I have seen as the prevailing thought regarding aviation accidents, which is: accidents are the result of a "chain" of failures; if any link in the chain does not occur the accident can be avoided. It is easy to say "software bug" or some single thing caused this failure, but there are many other factors that need to be considered: a/c certification requirements, training requirements (sim time, transitioning to new variants of an a/c), design philosophy for the Max, etc...

(Not an expert, but private pilot, have read many NTSB accident reports).

A lawyer's job is to win the case.

A forensic engineer's job is to prevent the next accident.

Cases are won on clear, persuasive argument. A simple, single-cause narrative works.

Accidents are avoided by removing or mitigating contributing factors, based on significance and costs.

Put another way: truth and science are not rhetorical activities. Politics and business are.

Note that the aviation industry (as you are likely aware) operates on a blameless basis in the interest of preventing a repeat of the same tragedy. That part has come and gone. The cooperation has happened. The investigation has been conducted, the physical, objective nature of the issue; it's implementation, and the physical chain of events that led to the crashes is established.

Boeing survived that.

What they haven't survived is the full process of peeling back the layers of paperwork, and subsequent investigation into what enabled the physical foundation of these crashes to come about.

Considering there was apparently a whistleblower willing to testify that decisions material to enabling the crash were made specifically to meet deadlines by dodging regulator scrutiny, I have the feeling Boeing will have a rough time of it.

> that strikes me as nonsense

It is nonsense. It takes multiple failures to bring down an airliner, and the NTSB accident reports recommend fixing all of them. It's why airliners are so safe today.

The 737MAX issue is no different.

Actually, they've been required to completely rearchitect the computer system due to being able to prove in an engineering simulator that a certain combo of single event upsets could end up causing loss of the aircraft.

They had three pilot's attempt the scenario, and one of them lost the plane. This never got looked at well enough before. If they had, they wouldn't have been able to certify before their deadline.

https://www.seattletimes.com/business/boeing-aerospace/a-lac...

It's different in that it starts to look as though there was a fairly concerted effort to keep the planes flying even though it was known to have a pretty big flaw.

the flawed product was rushed to market to "beat" Airbus there, but hey it's only ~350 people killed, and weeks of delay to ground the fleet. Sorry Boeing, you messed up hard. Personally I don't care to fly, but I sure as F am not getting on your planes. Fish rot from the head down, and I see that in this tragic case.

I guess you won't fly Airbus either, as they were recently slapped with an AD to address pitch instability problems.

"The FAA is adopting a new airworthiness directive (AD) for all Airbus SAS Model A321-251N, A321-252N, A321-253N, A321-271N, A321-272N, A321-251NX, A321-252NX, A321-253NX, A321-271NX, and A321-272NX airplanes. This AD was prompted by analysis of the behavior of the elevator aileron computer (ELAC) L102 that revealed that excessive pitch attitude can occur in certain conditions and during specific maneuvers. This AD requires revising the airplane flight manual (AFM) to incorporate updated procedures and operational limitations, as specified in a European Union Aviation Safety Agency (EASA) AD, which is incorporated by reference. The FAA is issuing this AD to address the unsafe condition on these products."

http://rgl.faa.gov/Regulatory_and_Guidance_Library/rgad.nsf/...

In the world of avionics, the concept of a “single point failure” is widely known and greatly focused on eliminating.

The failures that take down aircraft will likely involve multiple things going wrong, because the engineers (mechanical, aeronautical, system, software, etc) work hard to ensure a single thing going awry won’t down a plane.

For decades accident analysis stopped at the component part. Nowadays it may extend to organizational issues. Even more recent safety paradigms stretch to complex links between product, organizational, business, governmental contexts. Depending on how Moller has worked in the past he could mean a lawyer's language of culpability or an outdated view of safety.

I mean, if the crashes were due to a single cause, that cause should probably be either the failure of the last backstop, or the failure that started the chain. IMO the start of the chain would be the AOA sensor failure, and the last backstop would be the pilots running through the stab trim runaway checklist...

That doesn't get him settlement money, though.

The journalist writing the article may have distorted what he said. Is their a dogma of accident investigation that says there is always a root cause, or is it just a conventional way of working? There are certainly domains in which root cause analysis is inappropriate—psychiatry, for example.

https://en.wikipedia.org/wiki/Root_cause_analysis

What if every accident shares one common hole among all the holes that aligned, and had that hole been eliminate then none of the accidents would have occurred?

Well the probability of having two or more independent, crash resulting faults occur at the same time is remarkably low.

> independent

That assumption and the birthday paradox is what leaves people walking funny.

Reply on news.ycombinator.com