Hacker News Reader

Comment by 464192002d7fe1c | Hacker News Reader

464192002d7fe1c

I can highly recommend the Mayday (Air Crash Investigation/Air Emergency/Air Disasters/whatever its called this week) episode that deals with this, https://youtu.be/Bct1mWUp8to, and the show in general.

View on HN

I've watched this show since I was a kid — I think it's one of the reasons I went into engineering. It makes you think about the different reasons that air accidents happen, and illustrates that there's almost never a single attributable cause.

What seems initially just like pilot error may also be due also to workplace conditions, bad training, a lack of communication, or fatigue — a chain of events and circumstances that culminates in the accident. It's fascinating.

I think the very cool thing with this show is that it really illustrates how crashes are virtually never the byproduct of a single failure.

The most fascinating ones to me are the ones where there is a completely survivable mechanical failure, combined with human failures that lead to crashes. The one that comes to mind there is the plane where the engine was leaking fuel, so the tanks were not equal so the pilots engaged the cross-feed valve, causing BOTH tanks to drain leading to an out-of-fuel situation. If they hadn't done that, they could have easily made it to their ETOPS airport with minimal issues.

Also, that show makes me sound smarter than I am. I've had conversations with pilots, who will ask if I'm a pilot. Its great fun for parties when you run into pilots (GA and commercial).

You're thinking of https://en.m.wikipedia.org/wiki/Air_Transat_Flight_236, another Canadian flight.

In this case the pilots followed standard procedure for a fuel imbalance, which is the alert they got.

The fuel imbalance check-sheet also points out to make sure that the fuel imbalance isn't the result of a leak, and they did (somewhat -- it was dark outside) try to check for a leak.

They could also have looked at the rate of fuel consumption across both tanks to determine that the fuel consumption on the right tank was way higher than the engine could consume, but that was not SOP at the time (it is now, and the A330 [and probably other planes] now compute that information and alarm on it).

Its all layers, planes, with minimal exception, really don't crash from a single failure.

> how crashes are virtually never the byproduct of a single failure

There is the wonderfully named "Swiss Cheese model of accident causation" [1] elaborating on that. An accident happens when all the holes in the cheese line up...

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

Yes, I am always slightly annoyed by findings of pilot error. It is not that the finding is wrong, but the terminology leads the general public to blame the pilot when the reality is usually that the error was due to the pilot having misleading, incomplete, or just plain wrong information, or the controls behaving in a non-intuitive manner or whatever.

I suppose after watching this show, I now understand it better. But I now realize that pilot error is never just pilot error. Without some other failure or extraordinary circumstance, pilots have a damn hard time crashing a plane on their own.

With the sad exception of the Germanwings flight.

And the EgyptAir flight and the SilkAir flight if you're going to count deliberate pilot actions. (https://en.wikipedia.org/wiki/Suicide_by_pilot#By_pilots_in_...)

But Air France 447 is a recent example where it's really tempting to say that sheer pilot incompetence was the biggest factor: https://en.wikipedia.org/wiki/Air_France_Flight_447

Yes, I think they were definitely incompetent in the AF case. It's fair to question if their incompetence is a result of poor training by AF, but not recognizing a stall very squarely falls into "incompetent" IMO.

I think that's too harsh. The pilot applied what would be appropriate for "normal law" (full fly-by-wire). Unfortunately the plane was in "alternate law", where pulling the stick was the wrong thing to do. I would blame the incident partly to bad user interface and to the fact that the appropriate action is the opposite in normal and alternate law if the plane stalls (whoever came up with thinking this was a good idea). Even worse, the second pilot can't "feel" on the stick what the first pilot does.

A stall horn is hard to overhear.

Understanding what alternate law is as an airbus pilot is a very basic requirement.

>appropriate action is the opposite in normal and alternate law

I'm not sure what you are referring to here, but the fact that the appropriate action was to push forward to bring the nose down was the same regardless of alternate law/normal law. The bad UI was that the stall alarm shut off when the stall became so severe, which is why the pilots pulled up to stop the stall alarm (making the stall worse).

>Even worse, the second pilot can't "feel" on the stick what the first pilot does.

The system warns on dual input when it conflicts like that. Again, recognizing this condition is a very basic requirement of understanding the airbus control system. Regardless of that, the pilots did not communicate what they thought the problem was and what they were doing to solve it. Competent pilots don't end up in a situation where one is silently doing the opposite of the other and they don't realize it.

I agree with you that the incident is clearly attributable to pilot error, maybe even incompetence. However when searching for details I came across this: https://flightsafety.org/asw-article/stop-stalling/

I am convinced that much of human error is due to bad user interface, especially in technical settings. I think it is an important lesson that both pilots failed to recognize (in a stressful situation):

1. the plane was in alternate law 2. either actions contradicted each others

I think you're very right, and we shouldn't get into an Airbus-vs-Boeing discussion here.

Nevertheless, I wonder whether that particular accident would have happened with the physical link between yokes, rather than the side sticks; and whether there are lessons to be learned from that regarding UI/human-system-interface.

But then, how idiot-proof should planes be...

Yes it's usually a very interesting mixture of human factors and technical problems. Often in combination, e.g. some technical device not taking human factors into account and leading to usability issues. Quite often a disconnect between software developers and domain knowledge is the cause of mild to very serious issues.

One anecdote of a serious aircraft problem due to a programming shortcut I remember from a conference roughly and very simplified translates to this pseudo code for something related to landing/takeoff:

if wheels_rolling()==false: in_air=true

The problem was of course...plane with wheels on the ground but not rolling due to ice (more like sliding and there was some sensor mishap that also). And well relying on the rotation of the wheels as a height indicator and many other things...

Same here. Watched this show with my dad (who is an Engineer). Ended up becoming an Engineer myself.

I credit this show and MythBusters for pushing me into Science and Tech.

As much as I watch and enjoy Mythbusters, I rarely come out of watching that show feeling like I really learned something new or am particularly more knowledgable on a subject.

Mythbusters is fun and entertaining, Mayday is educational and entertaining.

> Mythbusters, I rarely come out of watching that show feeling like I really learned something

The earliest episodes were more scientific and educational. They spent quite a bit of time explaining how they were going to construct something and then showing how they actually built it. It became pure entertainment--devoted especially to blowing things up--once the show achieved success.

Another example of a show descending to the lowest common denominator: The first years of "Biography" on the A&E Network had stories about explorers, scientists, and world leaders--Michelangelo, Henry Ford, Julius Caesar, Marconi, Marco Polo--but now it's nothing but celebrities.

Agreed, the earlier episodes were better, but I still don't think they were great. The last season of the show (when it was just Jayme/Adam) was actually fantastic though.

I've never actually seen any of the biography series (I just associate it with shallow celebrity profiles) I'll have to dig up the earlier episodes, thanks!

I do wonder how the show by the assistants will do on Netflix. I too cannot say I'm a fan of the original. Their snowplow episode hit way too close to home and came to entirely the wrong (and a dangerous) conclusion. Testing anything having to do with snow, ice, and curved road surface on a flat, dry runway is doomed to failure with horrific reporting of results.

[I get it, no criticism of Myth Busters allowed]

The problem I have with Mayday is that there is only about 10min of footage repeated over an hour (45min + ad breaks). Most of the time is filled with summaries (after ad breaks), suspense (obtained by temporarily truncating explanations), questions and hyperboles. Also, it's very US-centric, as if no citizen of other nation were as competent as US ones. Multiplied explanations for the same fact feel like they think I'm 5, and the repetion feels like brainwashing. I come out of watching feeling like it would fit in 2 minutes, even though the original non-repeated content is probaboy more than that. I'd love if they could do a cross-season episodes: "It has been found that (spoiler alert) flight XYZ failed because of reasons A, B and C. The cause A happens over 3 accidents, let's study the 2 other causes. C happened when landing the space shuttle too and in the building collapse in another country. (etc)"

https://youtu.be/7MFtl2XXnUc

That's pretty common and makes the show way cheaper to make. Discovery did a what if dragons were real "documentary" which probably had a whole 10 minutes of footage stretched in an hour. The constant repeats after every commercial were hideous.

I see the value of Mythbusters in being able to calibrate your perception to those aspects of real-life physics that you don't typically see in normal life. How does it look like when there is a gas explosion in a house? A hand grenade blow? A bicycle trying to be ridden on the bottom of a pool? Thanks to Mythbusters, I really do understand these things pretty well now.

A more recent mini-show that gives you a similar discovery experience (albeit at a smaller scale) is the MrGear channel at YouTube: https://www.youtube.com/channel/UCo6DJdltbIub80bLiyJRv3w

The important thing about Mythbusters is that they have an idea and they test it. Their methods aren't always the best (or even good at all), but they take the right approach: they get an idea, and they try to verify or invalidate it. Pretty much anyone who watches an episode can think of a way to improve the experiment, and I think it's good that the show provokes that response.

Maybe you won't learn a lot from MythBusters, but it went a long way to imbibe a maker spirit in you. That I think is crucial to becoming a better engineer.

Replying to myself to point out that basically every single air incident that people are bringing up in this thread is also covered on this show. Its well worth a watch.

Outstanding program. Watching it will teach you a lot about flying. You'll become a more informed traveler and understand just how safe modern aviation is.

Reply on news.ycombinator.com