Hacker News Reader

If you think that's the entire premise, you didn't read much of the article. Keep scrolling, it's a well well reasoned discussion of why Malcolm Gladwell's commentary on previous KA crashes was misinformed and possibly even distorted to prove his point. Gladwell's book was the source of the "Korean culturalism is incompatible with emergencies" myth you cite.

> With Airplane pilots, Crashes are so exceedingly rare, and the backups, both mechanical, human, and training so well designed to prevent them

Air travel is exceptionally safe, but it falls prey to the law of large numbers: 99.9% success still means some people die every year when there are thousands of planes in the air. This has everything to do with individual decision making on the part of the pilots, which will necessarily be faulty occasionally. It's basically just random chance that some planes every year will crash because of pilot error.

> the issues associated with a crash are typically systemic, and little to do with an individual.

What sounds better in an NTSB report: "the pilot fucked up and flew into the ocean, whoops" or "we recommend adjusting the height of the foo so the pilot can better see it while barring"? Obviously everyone wants actionable outcomes from post-crash investigations, even when they're just trying to explain away a random accident. Y

> It's basically just random chance that some planes every year will crash because of pilot error.

When you design a system where errors can cause a large number of deaths (airplanes, trains, nuclear reactors,...), your objective is to have very few accidents (nobody can guarantee zero). Having someone to blame when an accident happens -- the pilot did not follow procedure -- is not enough.

Your system design must be human error tolerant, because you know these will happen. To take an extreme example, if a drunk pilot error causes an accident it's not only his fault, but there probably also is a systemic error which caused this criminal behavior (which is a possibility you expect) to cause an accident: maybe the cockpit culture was such that the co-pilot was frightened to stand up to the drunk pilot.

I guess my point was that airline safety hits a sort of horizontal asymptote as you put in effort. You'll go years and years without accidents, have one, and then struggle to generalize a rule from this incredibly freak event. My biggest complaint is that the whole 'East Asian culture causes deference to authority, crashes planes' argument is a huge generalization with very little data to back it up. In any other context, the number of events we're talking about would be anecdotal.

re: "Entire Premise" - sorry, I didn't mean to say that was the "Entire premise" of the article, I was just trying to say that his "entire premise regarding the golf game analogy" was off. Landing Planes is systemically guaranteed to be safe, it's no longer a function of individual performance, whereas there is nothing systemic about golf games - it's all about individual performance.

The OP was trying to compare and contrast a "System"'s behavior with an "Individual's" behavior, which was where the article went awry.

Re: Gladwell - Gladwell was most definitely not the source of "Korean Culturalism" regarding Air Safety. He simply wrote a popular chapter on an issue that already well documented.

The cause of any pilot error is not law-of-large numbers - pilot error has been, for the most part, systemically eliminated from flying. What's left now is system-flaws.

One of the obvious flaws at SFO was it's GlideScope Indicator was out of service. I have to believe that this will turn out to be a major contributor to the accident, and, in the future, Airport's may be much more cautious about disabling such safety equipment unless absolutely unavoidable.

Air travel is exceptionally safe, but it falls prey to the law of large numbers: 99.9% success still means some people die every year

99.9% success would mean ~475 plane crashes during take-off and landing on Heathrow alone each day =)

Wait, that can't be correct, can it? :)

Was going off this: http://en.wikipedia.org/wiki/World%27s_busiest_airports_by_a...

Edit: Silly me, the stats are by year not by day as I was expecting. Guess it means only ~1.3 crashes at Heathrow each day

I doubt there are 475000 take-off and landings per day from Heathrow despite how busy it is!

Actually, air travel doesn't really fall prey to the law of large numbers.

Air travel in the US is so safe that it's actually getting difficult to measure it and difficult to identify problem areas. The last fatal airline crash before Asiana 214 was years ago. Some people are most emphatically not dying every year anymore, despite thousands of planes in the air. We go years at a time without a single airline fatality in the US.

And yet, those few accidents that do happen all have identifiable causes with clear remedies. This Asiana crash may have been pilot error, but it was definitely not just random chance. Pilots letting their airplane get too low and too slow on landing with no adverse factors is a clear failure of training and proficiency. One fairly obvious remedy for this one would be to require all airline pilots to hand-fly a purely visual approach at least X times per year to ensure that the skill stays fresh.

I dare you to find a first-world airline crash from the past couple of decades that has no actionable outcomes from the investigation. Shouldn't take you too long to go through all the crashes, since there won't be many to look at in the first place. I guarantee you that every one had actual, useful recommendations come out of the accident investigation, not just bureaucratic CYA "recommendations".

> With Airplane pilots, Crashes are so exceedingly rare, and the backups, both mechanical, human, and training so well designed to prevent them, that the issues associated with a crash are typically systemic, and little to do with an individual.

That's a weird assertion. Generally, that events are extremely rare outliers in a system is an indicator they are the result of unique special circumstances applicable to that event (of which, the combination of involved individuals would certainly be a plausible candidate for an event where a small group of individuals had a major direct role the way the flight crew does in aircraft operations), rather than systemic problems. This is pretty much the basis of statistical process control.

It turns everything on its head to consider that extremely rare outliers in a process are a particular indication of systemic problems rather than special causes.

It's systemic because the whole point of procedures and policies is to avoid the failure of any one part of the system.

Imagine the power going out at your company's data center and then one of the generators doesn't turn on, and as a result your website goes down and you lose $10k/hour. Would management suggest that the generator is an individual and an outlier and we'll just hope it doesn't happen again? My guess is that a set of policies would go into place such that even if a single generator did fail, the website would not go down. You now have a systemic fix to what was entirely likely a very rare systemic problem. Just because it's rare doesn't mean it's not systemic.

> It's systemic because the whole point of procedures and policies is to avoid the failure of any one part of the system.

I can see that point and agree with it. (Well, I'd say the point is to prevent the failure of any one part of the system from producing unacceptable outcomes rather than preventing the failure of any one part of the system, but that's a minor quibble.)

I think my more significant disagreement with the original post is more with the description in that post that it is systemic and little to do with the individual. I'll agree that the fact that the system allowed the individual problem to produce a catastrophic undesired outcome is a systemic problem that (provided a reasonable correction is available) ought to be addressed, but that doesn't mean that the problem had little to do with the individual(s) involved, the fact that it is an outlier means that you aren't dealing with the normal behavior of the system, and that if you are going to address this kind of outlier event, you need to understand the contribution of special circumstances (of which, again, the contributing features of the individuals involved are quite likely relevant components) that produce the outlier event. You can't effectively address the systemic issue if you view it as unrelated to the individuals.

> Well, I'd say the point is to prevent the failure of any one part of the system from producing unacceptable outcomes rather than preventing the failure of any one part of the system

That's effectively what I was trying to say. I didn't get all the way there, though, and you've stated it quite nicely.

I guess part of the reason that people are suggesting this is indicative of systemic problems is that there's very little in the way of compounding factors. Yes the glide path indicator was non-operational, but that didn't cause 20 crashes that day. It had been out for quite some time (http://www.latimes.com/local/lanow/la-me-ln-equipment-out-of...). Many other planes landed safely. From what I understand (http://www.weather.com/news/san-francisco-plane-crash-weathe...) the weather was fairly clear so that didn't play in.

A lot of the analysis I've read seems to indicate that this was basically a rookie-esque mistake and that there's no possible way such a thing should have happened. That there should have been 20 checks to make sure that this guy wasn't authorized to fly the plane, but that for whatever reason, none of them were acted upon. At that point it's a systemic problem.

I would be looking at this very differently if there were a bunch of bad circumstances beyond his control that converged in ways completely unforseeable and as a result of that there was a crash. But from what I can tell, this was a very nice day to be flying around SFO.

EDIT: I just realized that I got really off-topic. Whoops!

I don't think you're off topic at all.

> Also, the korean culturalism developed when the overly formal/seniority centric cockpit of Korean Airlines was identified as a key contributor to a poor safety record with some korean airlines.

identified by whom? with what credibility? The whole point of the article is refuting that hypothesis. You've done nothing but restate that thought, without adding any more weight to it.

Identified by both the (US) National Transportation Safety board and Korean Air themselves, both of whom have considerably more credibility than Gladwell or Korean bloggers who get their accident stats wrong when trying to put him right (two passengers died on Flight 902, not "nearly half")

Here's a article about how bad the situation was in the late 1990s, published nearly a decade before Gladwell got in on the act. http://news.bbc.co.uk/1/hi/world/asia-pacific/575544.stm

It's generally believed that South Korean airlines have addressed a lot of those problems through changes to the organisational culture and better training, but after a crash which initially looks like a significant error by the flight crew, it's not wholly unreasonable to ask if the pattern is re-emerging.

Some citations regarding cockpit culture issues - I recall reading about them before Gladwell's book. Gladwell just put his own (somewhat hyperbolic) twist on them.

[1] http://www.nytimes.com/2002/03/26/business/new-standards-mea...

Here is an excellent article on the safety issues with Korean Airlines in the 1990s - much of it, btw, conflicting with the OP's description.

http://online.wsj.com/article/SB933090613281869060-search.ht...

The vast majority of golf players make bad shots most of the time, and, are a function of individual performance.

Sure, but if you took an untrained, unskilled person and had him flying a trans-Pacific jet, you'd see more crashes. This isn't done, for obvious reasons. A terrible golfer who shoots a 173 just annoys people who have to wait behind him. With an unskilled pilot, the stakes are higher and great efforts are made to prevent one from flying a passenger jet.

Good golfers, on the other hand, don't make terrible shots most of the time (although horrible shots are more common than plane crashes).

Also, the korean culturalism developed when the overly formal/seniority centric cockpit of Korean Airlines was identified as a key contributor to a poor safety record with some korean airlines.

OP's argument is that this isn't specifically a Korean or East Asian problem, and I agree. Most people are-- except when they perceive a life-threatening crisis (and that point of perception may be too late)-- deferential to authority. It's not limited to one culture.

Wow, talk about missing the point.