Safety is no accident

March 15, 2024 2:18 AM

The IATA officially announced: “In a significant achievement, 2023 saw no fatal accidents or hull losses for jet aircraft, leading to a record-low fatality risk rate of 0.03 rate per million sectors.”

There was a single fatal accident of a turboprop aircraft, the crash of Yeti Airlies 691 in January 2023 with 72 fatalities.

posted by fairmettle (27 comments total)

There's a lot there to be positive about, especially if you're a regular flyer. But then they go and say this:

At this level of safety, on average a person would have to travel by air every day for 103,239 years to experience a fatal accident.

I'm not a statistician, but I don't think that's how it works.
posted by dg at 2:46 AM on March 15 [1 favorite]

Boeing says “Hold my door.”
posted by Horace Rumpole at 4:35 AM on March 15 [52 favorites]

Wow, this really snuck up on me. I remember the Yeti 691 crash as 2023 and it stands out for so many reasons, but for some reason forgot it was a turboprop. And sure enough the JAL Haneda runway incursion crash that involved both fatality from the prop plane and hull loss of the Airbus occurred on Jan 2nd 2024.

I'm on record as being fairly discouraged by apparent culture changes at Boeing over the past couple decades, but a lot of people worked hard and succeeded at making 2023 an overall success.
posted by midmarch snowman at 6:42 AM on March 15 [4 favorites]

I trust the planes, it's the damn passengers who scare me.

For the most part, pilots are skilled and dedicated; flight attendants will help me in the event of an accident. But fellow passengers are rude, unmasked, selfish, impatient, dropping roll-aboards on me, and all the other Deadly Sins at the same time.

My wife hates flying and when the news was showing the accident last week, one of my sons made a crack about this bering a ding on airline safety. I replied that akkkkkkshully the event demonstrated how redundant systems (thanks, engineers!) and good crew (thanks, pilots!) make us super safe even when something super unlikely happens to the hardware....which some stupid MBA probably compromised through cost-cutting measures.
posted by wenestvedt at 6:58 AM on March 15 [4 favorites]

I'm not a statistician, but I don't think that's how it works.

not sure why you think it isn't.

they did some rounding on the numbers to make them a little more mediagenic, but in 2023 (pdf) there was 1 fatal accident in 37.7 million flights for a rate of 0.0265 fatal accidents per million flights.

Assuming independence, the probability a given flight being involved in a fatal accident is 0.0000000265: every time a plane takes off and lands, there is a 0.0000000265 chance that someone is (or multiple someones are) going to die.

The distribution they use for modeling the probability of flight #T being involved in a fatal accident is the exponential distribution. If 0.0000000265 is interpreted as a hazard rate, the exponential distribution reflects the probability of a fatal accident on flight T (considering the possibility that a fatal accident already occurred sometime between times T=0 and T-1)

the mean of the exponential distribution (1/ 0.0000000265) is the average number of flights until the first fatal accident occurs. Assuming a single flight a day, every day, that works out to 103290 flights, which is a bit off from their 103239 years, but i'm guessing they used the non-rounded numbers to do the calculation.
posted by logicpunk at 7:05 AM on March 15 [19 favorites]

I think it was a crack at the unlikelihood of one person traveling by air every day for over 100,000 years.
posted by entropone at 7:59 AM on March 15 [5 favorites]

What’s happening with Boeing planes right now is certainly less disastrously awful than when the 737 MAX was crashing on a regular basis, but it’s not what I would call safe. You can only roll the dice on maintenance fuck ups and cost cutting culture so long.
posted by Artw at 8:00 AM on March 15 [2 favorites]

Several, possibly a majority, of my closest friends are air traffic controllers from Oakland Center, some of whom have worked other centers, worked towers, and a few who have done military ATC prior to their time with the FAA. Some have gone on to become trainers within the FAA or NATCA, or to help build the next-generation systems that controllers use when they're working the board.

It's really hard to express how goddamned serious these people take safety. They spend their entire careers in the safety industry, and are some of the highest-order professionals within that industry. I've spent hours in long, deep conversations with these folks, learning and listening, prying at their assumptions, digging into what they know and have discovered over decades.

Controllers, as much or more than manufacturers or industry players, are why incredible results like this occur.
posted by majick at 8:12 AM on March 15 [17 favorites]

You can only roll the dice on maintenance fuck ups and cost cutting culture so long.

Here's hoping that the current bad press is scaring the shit out of Boeing management. Because the solution is crystal clear--we know what a Boeing with a culture of safety looks like.
posted by mr_roboto at 8:42 AM on March 15 [5 favorites]

we know what a Boeing with a culture of safety looks like.

Not to keep doing the "well, I know a guy..." thing, but my uncle was a Boeing guy for a big part of his career up until the early 90s, and...

Yeah. Yeah, we do. Shit yeah, we do.
posted by majick at 8:49 AM on March 15 [5 favorites]

Except for John Barnett oh that was 2024.
posted by MonsieurPEB at 9:18 AM on March 15

majick, I am in awe of air-traffic controllers - please convey my deepest respect and admiration to your friends. Reporting on staffing shortages and shitty overwork practices for controllers this past year made me mad as hell. I hope their industry soon gets more of the support they deserve.
posted by rrrrrrrrrt at 10:25 AM on March 15 [1 favorite]

It's so important to recognize how much of this safety comes from redundancy, and incentives for people to err on the side of caution.

We have to always remember that while we see an incident-free flight as a sign of the system working, there are others who only see the money they "wasted" on redundancy when nothing even went wrong.

For many of them, being wildly profitable doesn't seem to mitigate that worldview. They see only what they don't have, not the embarrassment of riches they do. So we must make sure to always dangle real, personal consequences (criminal liability, complete financial ruin) over them, and show that we're willing to impose those consequences.

Not doing an amazing job on that part, and I think we'll be amazed that statistics like this existed in living memory as the rot comes to the surface over the next few decades.
posted by Riki tiki at 11:54 AM on March 15 [4 favorites]

closest friends are air traffic controllers from Oakland Center,
Hey, nice shout out for ZOA!
Former controller here....what your friends might also tell you that "stuff" happens every day. Planes touch on the "tarmac" ( a word I never ever used once in 34 years as an ATC...never) while being pushed around, planes lose wheels while taking off.....stuff happens. Every day.
What doesn't happen is the sort of thing happening to Boeing. The most amazing thing about the door flying off is...nobody died.
posted by pthomas745 at 2:56 PM on March 15 [3 favorites]

The most amazing thing about the door flying off is...nobody died.

Well...someone died, it just happened a bit later.
posted by Literaryhero at 5:17 PM on March 15

the unlikelihood of one person traveling by air every day for over 100,000 years

And yet it is well known that one may travel by boat for a million years.
posted by neuron at 6:27 PM on March 15

It has been said that one is safer flying on a US commercial airliner than sitting on one's sofa.
posted by neuron at 6:28 PM on March 15

Another factor is regulations. Commercial aircraft are subject to very strict maintenance schedules. There are parts that get replaced at specified times or conditions because the parts are crucial enough that they can’t be allowed to wait until the part fails to replace it. There are also scheduled maintenance services which basically entail disassembling the entire airliner and inspecting *everything* for problems, metal fatigue, anything that can go wrong. It’s expensive as hell. However, the safety statistics explain pretty clearly that it’s worth the time and expense.
posted by azpenguin at 8:28 PM on March 15 [3 favorites]

Well...someone died, it just happened a bit later.

Wait, what? I can't find anything to corroborate that.
posted by schmod at 7:06 PM on March 16

>>Well...someone died, it just happened a bit later.
>Wait, what? I can't find anything to corroborate that.
I'm assuming this is a reference to John Barnett, the Boeing whistleblower who was recently found dead after warning his family that if he ends up dead it wasn't suicide.
posted by mrgoldenbrown at 8:34 PM on March 16

Let's not forget the pilots.

Just last week I was on a flight into Newark that had to land under some nasty wind conditions. We circled for a good 45 minutes, almost touched down once... then the crew pulled the nose back up, probably due to the winds being too high to land safely. People around me were getting sick left and right because of the turbulence.

This is one of the few flights I've taken where I really wasn't sure we'd land safely at all. And I've flown a LOT. We circled for at least 10 more minutes, and NO ONE was talking. Guy seated next to me was visibly praying. I made a few du'aa (prayers) myself, and then the crew tried once more to land the plane. We bumped and juddered, but the plane landed without incident.

It's common on flights out of Spain and Portugal for passengers to applaud a good landing, but the applause this time was a bit more heartfelt than usual.
posted by rabia.elizabeth at 2:57 AM on March 17 [1 favorite]

Hey, nice shout out for ZOA! Former controller here....

The ZOA shout-outs never fricken end around here, I assure you. These are all post-PATCO guys, (obviously!) but not that much post-PATCO in the grand scheme of things, so they're all out now. The earliest of them out was a guy who used to do flow back in the ancient days of flow, out of Denver before he came to Fremont, but yeah, I get a lot of stories. A lot. A LOT of stories. Even stories from the ocean, which, it's like, what the hell even happens in the ocean? "Well, it turns out..."

Let me just say, though, that the level of commitment I see—particularly from people in that tranche of the profession—is just jaw-dropping. There's a reason my own daughter, who grew up in the shadow of these titans, is on the verge of putting her hand into this year's bid. All the old guys are practically screaming to get her to go in.

You want to keep seeing results like this, year after year? Well, we need controllers, and we need controllers badly. Tell your congresspersons.
posted by majick at 9:02 PM on March 17

'm not a statistician, but I don't think that's how it works.

not sure why you think it isn't. ...Assuming a single flight a day, every day, that works out to 103290 flights, which is a bit off from their 103239 years, ...

I think it was a crack at the unlikelihood of one person traveling by air every day for over 100,000 years.
It was more a crack at people misusing statistics to make convincing but wrong statements. Saying 'a person would have to travel by air every day for 103,239 years to experience a fatal accident' is wrong, because the likelihood of a fatal accident is the same (ie minuscule) the first time you fly as it is the 100,000th time you fly. Their statement says that you can fly every day for just 103,238 years and you won't experience a fatal accident ever. The statistics are good enough in this case that there's no need to torture them to make things seem better.
posted by dg at 2:57 PM on March 18

people are really bad at probabilities, tending to overestimate the frequency with which rare events occur and underestimate the frequency with which medium and high-probability events occur. There's a fun study showing how people interpret words that express degrees of uncertainty ('always', 'never', 'usually', 'sometimes' and the like) in terms of probabilities - even words that imply no uncertainty like 'never' are given a 'small' chance of occurring anyway. people suck at probabilities, especially really low probabilities

telling someone there's a 0.0000000265 probability of a fatal accident on each flight essentially invites a '. . . so it could happen . . .' Converting the probability to the average # of samples needed before observing a 'success' is a way to communicate the risks in a way people can understand better.

If you gave the average person a coin, told them there was a 0.0000000265 probability of flipping heads, and informed them that as soon as they flipped heads the first time they'd get one million american dollars, there is a good chance that person would get started flippin' immediately. if you told them it would take 100,000 years on average to get the money, you'd have fewer takers.

Their statement says that you can fly every day for just 103,238 years and you won't experience a fatal accident ever.

that is not what their statement says. their statement says 'on average a person would have to travel by air every day for 103,239 years to experience a fatal accident'. which, based on the hazard rate and their use of the exponential distribution, is true. Another way of saying it is that you would need to take 37.7 million flights to have a 50% chance of being involved in a fatal accident. but again, people are bad at probabilities and very large or small numbers, so they use the 103,239 to make the lack of risk as salient as possible.
posted by logicpunk at 5:24 AM on March 19

As we see people will just turn around and be all “okay then we will be differently bad at probabilities” if you do that.
posted by Artw at 10:43 AM on March 19

Maybe it's more of a language issue than a statistics one. Saying 'a person would have to travel by air every day for 103,239 years to experience a fatal accident' means that there is no chance of encountering a fatal accident without first traveling every day for 103,239 years. Adding in the 'on average' bit does change the meaning of the statement, but is likely to be brushed over by readers awed at the near-impossibility of anyone ever being killed in an air accident the statement presents. Maybe it's a technically correct statement through the eyes of a statistician, but it's a misleading statement in my eyes and, I believe, deliberately so. My original point was that the statistics are actually fantastic and there's no need to dress them up, but people who write reports like this can't help but try and make things look better than they are. When you try to dress up statistics that already look fantastic, you risk silly statements like the one quoted.
posted by dg at 4:35 PM on March 19

Let's not forget the pilots.

Pilot competence is a massive factor towards the gargantuan safety of commercial aviation. Despite the fact that the vast majority of the job, hours-wise, is more akin to driving a bus than it is to flying an airplane these people consistently deliver passenger safety in big and small ways over and over again. Every single day. Millions of small decisions and corrections are made by pilots daily, largely unseen, that make it safer to sit behind them than it is to walk down the jetway in order to board.

Crew competence in general is a big fuckin' deal. Most people don't keep at the forefront of their minds that cabin crew are trained passenger safety professionals first and foremost (other than a passing chuckle at the safety mime routine that begins their flight perhaps), and customer service second. Typically at the end of a flight I make a note to thank the FA working the deboarding door with a "thanks for keeping us safe up there." These folks are the last line of defense against loss of life in a wide variety of potential incidents.
posted by majick at 7:11 AM on March 20 [1 favorite]