16 ms·

The Silent Anniversary: Fifteen Years Since Our Last Major Crash

272 points by maxcan 10y ago

maxcan 10y ago
What wasn't mentioned directly was that since 2009, US passenger airlines have had a perfect record of zero fatalities, which is mind blowing.
- mikeash 10y ago
  Especially mind blowing considering that not all that long ago, it was common to see hundreds of airline fatalities per year, despite much reduced traffic. If the 1960 level of safety carried through to today, we'd see over three thousand airline deaths each year.
  - imglorp 10y ago
    Furthermore, fatalities are still in the hundreds world-wide, often with the same equipment flown by US airlines. This number tells us that the FAA+NTSB, despite some other criticism, _usually_ has its act together where it counts the most. They iterate on identifying safety problems and force the airlines to adopt solutions.
    - npunt 10y ago
      US airlines often sell their older planes to foreign airlines, which may account for some of the increased issues those airlines have. Your point still stands though - culture and processes rule.
      - mjevans 10y ago
        You could argue that those planes should be forced to retire instead. A different perspective might be to require that every 10000 hours of service see a new mark painted on the hull. This way passengers could be informed of how much 'age' a plane has had and the regulations could also apply to any plane which /might/ fly in to the US (or Europe).
        cjrp 10y ago
        The hull's age only tells part of the story though; how well it's maintained, and the ability of the pilots flying it would need to be factored in as well.
        mikeash 10y ago
        Those are probably far more important. I'd be far more comfortable boarding a 30-year-old airliner with great maintenance and great pilots than boarding a brand-new 787 with a clueless maintenance crew and pilots who aren't quite sure which way to move the stick (edit: yoke) to make the houses get smaller. If you look at the causes of crash, it's usually pilot error or improper maintenance, occasionally design flaws, but not equipment age.
    - enjo 10y ago
      It brings to mind a flight just last week that crashed due to likely negligence (not carrying enough fuel). http://www.cnn.com/2016/11/29/americas/colombia-plane-crash-investigation/ http://www.cnn.com/2016/11/29/americas/colombia-plane-crash-...
      - imglorp 10y ago
        So it turns out not only was that guy negligent, but there was also another plane in the pattern who got priority because, at the time, it looked like that other one was low fuel also (maybe a false alarm). Here's a reputable source. https://fearoflanding.com/accidents/not-enough-fuel-the-disgusting-truth-about-lamia-flight-2933/ https://fearoflanding.com/accidents/not-enough-fuel-the-disg...
nkrisc 10y ago
I know perfectly well that a fear of flying is often an irrational, though very real, fear for many people. But it's exactly the irrational nature of it that boggles my mind: they should be terrified of driving, but they aren't. I'm way more comfortable flying than driving.
- TeMPOraL 10y ago
  It's weird how the brain works. I know very well that airliners are safe, and I'm aware just how much work goes every day to ensure they stay safe. And yet, every time I'm flying, I have this uncontrollable thought, "maybe this is the day we all crash and die". I always get a little anxious during takeoff and landing, I listen to all the noises and worry about them. "This doesn't make sense, the plane didn't usually make that sound". "Oh shit, the lights blinked, wiring must be damaged." Stuff like that. It usually takes me few minutes after liftoff to get those thoughts out of my head.
  - existencebox 10y ago
    At least for me there's a feeling of autonomy. If a plane crashes, there's very little that I as a non-superman can do to stop that from happening (with me inside it.) While driving, I can trust my own reflexes, awareness, and decision making to put myself in better or worse situations. Now this may still absolutely be a false perception, but this is where the feeling is rooted for me (from lengthy introspection on similar situations when I used to do trick-skiing and never really questioned my safety, but would be very hesitant to e.g. skydive). To some degree I think it's potentially also "failure magnitudes." Having been in a handful of very minimal car crashes, that calibrates my "something goes bad" prior to a not very impactful bar; whereas if something goes wrong on a plane, it can go VERY VERY wrong. Again, this is likely a false perception due to observations of outlier events in both cases and sparsity of signal, I'm just trying to question the origin of my feelings.
    - mortenjorck 10y ago
      There's this thing that will happen to me when I doze off as a passenger in a car: When I wake up, there's often a split second of panic between the visceral understanding that I'm in a moving vehicle and the slightly delayed recollection that I'm not the one responsible for driving it. Somehow, this turned into a problem for me when flying (even though I'm not a pilot). I wake up, I instantly know I'm moving at 500 miles per hour through the air, then I have a singular panic of being in a vehicle not under my control. This has only happened once – I haven't slept on a plane since.
blakesterz 10y ago
"2009: A Colgan Air (Continental Connection) Dash-8 Q400 crashed outside Buffalo. Fifty people were killed, including the occupant of a house struck by the plane." "2006: Forty-nine people perished when a Comair (Delta Connection) regional jet crashed after attempting takeoff from a too-short runway." I'm not sure, maybe "major" has an official cut off number, but 49 or 50 doesn't seem "minor". It's a long way from 260, but it's far from 0.
- gr3yh47 10y ago
  They're counting Major US airlines, not including regional carriers FTA: >To be clear, there have been a handful of tragedies involving regional carriers and freighters [...] It’s the majors that set the standard, and it’s only fair that we measure from there.
- el_benhameen 10y ago
  Colgan and Comair aren't counted in the metric because they're regional, not major/legacy, airlines. They codeshare with the majors, but they're different organizations. Tangential: Frontline had a great episode on regional airlines' safety issues. http://www.pbs.org/wgbh/pages/frontline/flyingcheap/ http://www.pbs.org/wgbh/pages/frontline/flyingcheap/
  - js2 10y ago
    Tangent on the tagent: Phillip Greenspun touches on regional airline safety in this post comparing US to foreign airline safety - http://philip.greenspun.com/flying/foreign-airline-safety http://philip.greenspun.com/flying/foreign-airline-safety If having a newbie in the right seat represents a challenge to safety, how do U.S. regional airlines manage? The U.S. regional airline is hiring pilots who are more experienced than those hired by foreign airlines (the U.S. regional F/O typically has been a flight instructor) but less experienced than those hired by a U.S. major airline. The answer is that U.S. regional airlines have an accident rate as much as 5-10 times that of U.S. major airlines. The culture is the same, but apparently there is no substitute for experience.
  - mannykannot 10y ago
slg 10y ago
I am surprised there wasn't more than a token mention of terrorism. People like to bemoan airline security and I think there is a valid argument regarding the effectiveness of the TSA, but this is also the safest stretch the US airline industry has had against terrorists and hijackers since the early days of the industry in the 40s and 50s.
- mikeash 10y ago
  I think the current level of airline security is overkill, but I'm always a little surprised to see people advocating for having none. In the late 60s and early 70s, US airliners were getting hijacked at a rate of once every week or two. Once basic security screening was introduced, it became far more rare. Making people take off their shoes and empty their water bottles is a bit silly, but basic checks for guns and bombs are a good idea.
  - cbr 10y ago
    In the late 60s and early 70s, US airliners were getting hijacked at a rate of once every week or two. https://en.wikipedia.org/wiki/List_of_aircraft_hijackings https://en.wikipedia.org/wiki/List_of_aircraft_hijackings is pretty dramatic
    - tunesmith 10y ago
      Hijacking was pretty different back then, which is why passengers were used to the idea of staying uninvolved and not escalating things. That's part of why 9/11 was just a (culturally) unimaginable shift, it wasn't really in the gestalt that a hijacker's main goal was other than to kill/die.
      - JshWright 10y ago
        Yeah, 9/11 could work exactly once (and even then, it only in three out of four concurrent attempts attempts)
  - throwaway2016a 10y ago
NickHoff 10y ago
The effect of culture should not be underestimated here. There really is a cooperative spirit among pilots, air traffic controllers, weather briefers, mechanics, the FAA, etc. Controllers are willing to help pilots with special requests if able, pilots file weather reports for other pilots if what they encounter aloft is different from what they expected, mechanics take pride in their work and are highly regarded by pilots. I generally take a dim view of regulators, who generally seem to have less expertise than the people they're regulating. That's mostly not the case here. I'm libertarian and a pilot - and I like the FAA. Nobody is out to nail you for a minor technical infraction, but they will yell at you for something that's unsafe but technically legal. This encourages people to report near misses, even if casts them in a bad light. The regulations themselves are generally reasonable, and frankly feel like they were written by pilots, controllers, and airport operators. The point is this - regulations matter and budgets matter, but if the culture is toxic, people will find a way around it anyway. When I see a problem elsewhere in society - investment banking in the 2000s for example - people propose heavy regulatory solutions. That may work, but wouldn't be so much of a battle if the culture were better. I haven't thought deeply about what "better" means in this context, nor about how to intentionally change a culture. I'd rather just fly.
- maxcan 10y ago
  > I'm libertarian and a pilot - and I like the FAA Me too. For every frustration about what they've done to the costs of things in GA, and those are valid, their accomplishments in general are incredible.
- THE_PUN_STOPS 10y ago
  I've often heard it said that FAA regulations are written in blood. Meaning, if a rule exists, it exists for a damn good reason. For anyone interested, NASA's ASRS [1] (Aviation Safety Reporting System) is what the parent commenter is referring to regarding reporting near misses. The idea is, anyone can submit an incident report with NASA, a neutral separate party from the FAA, which is placed in an anonymized public database and cannot be used to justify punishment against anyone involved in the incident. This promotes transparency and learning from the mistakes of yourself and others, instead of covering them up. Frequently, the person who caused an incident in the first place will promptly report it to ASRS, as the FAA looks favorably upon those who come forward to ASRS on their own and dimly upon those who do not. I think more industries could use this kind of transparency in reporting sketchy situations. [1]: https://asrs.arc.nasa.gov/overview/confidentiality.html https://asrs.arc.nasa.gov/overview/confidentiality.html
chinathrow 10y ago
RIP to the freight dogs though: "2013: A UPS flight crashed on approach to Birmingham, Alabama, killing the two pilots."
- chinathrow 10y ago
  Why the downvote? UPS is a major airline - even if they only transport cargo, humans still had to give their live. Also, "freight dogs" is the gentle term for the pilots in cargo planes.
js2 10y ago
In no way directly related to this post, but I'm always looking for an excuse to link to my two favorite airline stories. 1) The Turn - http://www.theatlantic.com/past/unbound/langew/turn.htm http://www.theatlantic.com/past/unbound/langew/turn.htm 2) The Crash of United Flight 232 talk by Capt. Al Haynes - http://clear-prop.org/aviation/haynes.html http://clear-prop.org/aviation/haynes.html
GFischer 10y ago
Plane crashes have been going down not only in the U.S., but worldwide, which is commendable. A bit unfortunate for this to be posted a few days after the Chapecoense crash shook the world (noted in the postscript), but that was a regional charter flight, not a major U.S. airline. Interesting infographic: http://www.ibtimes.com/how-many-planes-crash-every-year-how-many-people-die-plane-crashes-chart-1560554 http://www.ibtimes.com/how-many-planes-crash-every-year-how-... Edit: the Chapecoense crash shows exactly why the FAA's work is so commendable - disregard of proper procedure and safety measures led to that accident.
- gus_massa 10y ago
  About the Chapecoense crash. From [Portuguese] http://www1.folha.uol.com.br/esporte/2016/12/1838707-lamia-ja-havia-descumprido-regra-de-combustivel-em-voo-com-messi.shtml http://www1.folha.uol.com.br/esporte/2016/12/1838707-lamia-j... Autotranslation: https://translate.google.com/translate?sl=auto&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=http%3A%2F%2Fwww1.folha.uol.com.br%2Fesporte%2F2016%2F12%2F1838707-lamia-ja-havia-descumprido-regra-de-combustivel-em-voo-com-messi.shtml&edit-text=&act=url https://translate.google.com/translate?sl=auto&tl=en&js=y&pr... > On at least two occasions, the Bolivian company had flown beyond its declared autonomy. [4 hours and 22 minutes] > One between Cochabamba in Bolivia and Medellin in Colombia (4 hours and 27 minutes), and the second between Medellín and Santa Cruz de la Sierra in Bolivia (in 4 hours and 32 minutes).
  - wapz 10y ago
    Wouldn't the difference in 5-10 minutes not be the reason for the lack of fuel, though? I would guess a 4 hour flight would have enough fuel for at LEAST 5.5 hours if they couldn't land at an airport and had to be diverted to another one (right before landing).
- astrodust 10y ago
  Flying is so amazingly safe now that you're far, far more likely to be killed on a cab ride to the airport than on the plane itself.
estrabd 10y ago
9/11 doesn't count? Sounds like "fake news."
JHof 10y ago
Exploiting this aviation-related post to see if any aerospace folks will comment on my two-month old Ask - Ask HN: Why is the National Airspace System so dated? https://news.ycombinator.com/item?id=12651244 https://news.ycombinator.com/item?id=12651244 In regard to the safety of regionals vs. the major airlines that the article's author is writing about: I can see why the statistics are separate. Even though, according to the FAA, we're all under the "same standard", in practice, this isn't the case. As long as the paperwork looks the same, though, we're "safe". Also, within the industry, some pilots tend to think of themselves as shouldering the responsibility of safety, however, I sometimes think we're safe despite the pilots. Engineers of aircraft, airspace, arrival/departure/approach procedures, airports, ATC systems, etc. are silently keeping you safe. That the FAA is able to manage all of these elements is quite impressive. Still, its massive bureaucracy can be quite Vogon-like.
- rplst8 10y ago
  I think a lot of software engineers would be blown away at how strict software standards are for avionic systems. Like 100% modified condition/decision coverage for all software tests, strict. No pointer math and no dynamic memory allocation, strict. https://en.wikipedia.org/wiki/Modified_condition/decision_coverage https://en.wikipedia.org/wiki/Modified_condition/decision_co...
  - gurneyHaleck 10y ago
    Being a party to lackluster software development on a regular basis, I somewhat pine for projects this strict. As part of my daily routine, feigning surprise at the fact that, given a tree of decisions, I'm only permitted to code for the ones that make money, simply to find that "Quality Testing" reveals the 500 other permutations will inevitably result in thrown exceptions with hideous, glaring error messages. "WHOOPS."
    - rplst8 10y ago
      This sort of thing is exactly what some of the software development processes are all about. When I was in a CMMI training class 10 or more years ago, the instructor we had said basically that much. By instantiating a process, it takes the mental weight of knowing (or checking) if everything is correct. No one builds cars that way - why should we build anything that way. The processes should catch the inevitable human errors.
rl3 10y ago
This would be why I always tell people to fly major U.S. legacy carriers if possible, or their western equivalents (e.g. British Airways, Qantas). As a general rule, major carriers from countries that have a pool of predominantly native English speakers to hire from.
- jpatokal 10y ago
  Actually, only two of the world's safest airlines hail from the US: http://www.telegraph.co.uk/travel/news/The-worlds-safest-airlines-revealed/ http://www.telegraph.co.uk/travel/news/The-worlds-safest-air... And 6 of the top 10 are not from English speaking countries (ANA, Cathay, Emirates, Etihad, EVA, Finnair).
  - rl3 10y ago
    Actually, only two of the world's safest airlines hail from the US The list you just cited includes Alaska and Hawaiian on that list, for a total of four US carriers. My original statement was in context of major legacy carriers, so I could understand you excluding those—however, each of the airlines you cited has a comparable—if not smaller—fleet size to that of Alaska Airlines. Also, keep in mind that major US carriers have underwent major consolidation in recent years: US Airways and American Airlines United and Continental Northwest and Delta >And 6 of the top 10 are not from English speaking countries (ANA, Cathay, Emirates, Etihad, EVA, Finnair). If you total the fleet sizes of each of these, you end up with 907 aircraft. American Airlines alone operates 952 aircraft. Moreover, many of the airlines you listed tend to hire Western pilots formerly of major legacy carriers—captains in particular—in order to meet high PIC experience requirements. - It's also worth mentioning that AirlineRatings.com[0] (the source of the list you cited) currently assigns Malaysia Airlines a safety rating of 5/7, despite the fact they managed to lose two 777 hulls (MH370, MH17) only months apart—one of which was perfectly preventable (don't fly over active war zones?). The same site affords Asiana Airlines a similarly lofty rating of 6/7, despite a 777 hull loss (Asiana 214) due to its pilots being unable to perform a simple visual approach in ideal conditions at SFO. A truly stunning display of incompetence that belied a terrible training and safety culture at the airline. Of course, going by safety records alone, I'm sure both airlines looked great on paper about five years ago. [0] http://www.airlineratings.com/safety_rating_per_airline.php http://www.airlineratings.com/safety_rating_per_airline.php
nikcub 10y ago
There's likely no media coverage of the anniversary because only including major airlines is a very arbitrary cutting off point. The NTSB lists five major passenger air accidents since 2001: http://www.ntsb.gov/investigations/data/Pages/paxfatal.aspx http://www.ntsb.gov/investigations/data/Pages/paxfatal.aspx No agency breaks out "major" airlines from "regional" for the purpose of regulation or statistics, nor does any other nation. It is purely self-serving. What makes this post weirder is while cutting out regional airlines from the statistics, the causes he attributes with improving safety for the majors are the same things that regional airlines have also been doing.
- zardeh 10y ago
  Of those 5, 4 occurred in propeller craft. I think that it is relevant to include that kind of difference. Regional airlines and major airlines do work differently.
- erroneousfunk 10y ago
  He does mention that in the post, and, yes, I think the author would also agree that it's self-serving. However, I would also argue that this is a symbolic milestone of the increased safety of US commercial flights, and an interesting thing to reflect on. I wouldn't take it too literally as any sort of real statistical measure, but it's certainly worthy of a blog post.
sean_patel 10y ago
> Some will argue that it’s unfair to gerrymander statistics in this fashion, separating airlines into categories of convenience. However, it’s necessary. Regional carriers, for instance, can have substantially different cultural and operational environments than their legacy carrier affiliates. It’s the majors that set the standard, and it’s only fair that we measure from there. Yes. It is gerrymandering, and no you can't wash your hands of the regional carriers' crashes and deaths after you've stamped your 'trusted' name and brand on their flights. I am appalled at how they callously dismissed those plane crashes since 2001 like this. > To be clear, there have been a handful of tragedies involving regional carriers and freighters... in which dozens died. Really? 'dozens died' is ok, but 100s died is not? Are we cattle?
- rplst8 10y ago
  > Really? 'dozens died' is ok, but 100s died is not? Are we cattle? Yes. Some people are choosing to pay less for a little more risk. The market should allow for that in the fringes. The vast majority of flying happens on the major carriers, so having stricter standards for the majority of your flying passengers is the right way to approach safety.
throwaway729 10y ago
Does anyone know what this is referring to? > We’ve engineered away what used to be the most common causes of accident
- sowbug 10y ago
  I'm not enough of an experienced pilot to authoritatively answer, but here's a guess at some items. * Sterile cockpit: a rule that pilots avoid any distracting activities during critical parts of flight (mostly takeoff/landing, flying around busy areas). * Crew resource management: a shocking number of major commercial accidents have involved either confusion in the cockpit or one pilot who knew something was wrong but didn't feel he/she had the authority to say something. Tenerife is a very sad example. CRM is a set of procedures that specify how concerns are brought to the PIC (pilot in command) in a way that's unlikely to make things worse. Basically, it's an escalation procedure that is recognized on both speaking/listening sides as something that needs to be paid attention to. * Checklists: Not sure if this is new, but it is completely routine in flight to consult standardized checklists for everything, whether ordinary takeoff/flight/landing or emergency situations. It's thus a lot less likely you'll make a stupid mistake in flight like running out of fuel because you forgot to switch tanks. (Actually people still run out often, but the out-of-fuel checklist includes ensuring you've remembered to switch away from the empty tank, meaning people don't crash planes with half their fuel anymore.) There is no cultural shame in consulting checklists; in fact my (limited) experience is that most pilots would question the judgment of someone who didn't use checklists. This is a bigger deal than you might think because most aircraft incidents are fundamentally attributable to human error. * Better weather technology, both for prediction and real-time assessment. Flight 1549 (the one that landed in the water with no deaths and just a couple serious injuries) is a good example of the first three. You can listen to the cockpit recording to see them all in action. When the birds hit, the pilots were already 100% focused on the critical takeoff portion of the flight -- zero distractions that could have delayed reaction time. The copilot immediately got out the POH (plane's owner's manual containing tons of checklists) and started all the restart procedures while the captain reflexively put the plane in the maximum glide position to extend range. Then right as they're lined up for the river, Sullenberger asks the copilot "Got any ideas?" While this might sound like comical desperation, it's actually good CRM in action -- after they'd tried the checklists and were in a somewhat stable descent, the captain solicited feedback from the second-in-command, just to make sure they hadn't overlooked anything else. Anyway, I'm sure my list is far from complete. Someone more knowledgeable can continue it.
ahupp 10y ago
"... the first officer of flight 587, Sten Molin, who was at the controls, overreacted, rapidly and repeatedly moving the widebody jet’s rudder from side to side,... . The vigor of his inputs caused the entire tail to fracture and fall off." I was interested in learning more about this, and it turns out there's an alternative interpretation that argues he was flying according to best practices at the time rather than "overreacting": http://articles.latimes.com/2005/jan/11/opinion/oe-garrison11 http://articles.latimes.com/2005/jan/11/opinion/oe-garrison1...
- MaxfordAndSons 10y ago
  IIRC from watching the episode of "Air Crash Investigation" (aka "Mayday") on this flight, it was a pretty tragic confluence of factors: it was a best practice on other planes, but not on the A300, but American Airlines training manual had not been updated to reflect this change, nor did the simulator models account for this eventuality. Kinda surprised the author of this piece wrote it up as he did.
  - rrmm 10y ago
    If I'm remembering the details of the case correctly, it turns out he did overreact and that did cause the tail to fail: Certainly the tail did fail, and the pilot's input caused it. The important part and the part that has been noted above is the NTSB's (and the industry's) focus on not blaming the pilot but figuring out why the situation turned out the way it did, what factors allowed it to happen, and how things could be changed to insure the same chain of events doesn't keep occurring. (Again if I recall correctly), It was the confluence of a few different events compounding each other and resulting in the accident: One factor was the airline company sim training on recovery from upsets. Basically the pilot going through the training could come away learning entirely the wrong lesson. The simulation was hardwired so that the pilots' inputs were disregarded until the plane was in an upset and then the pilot took over. Pilots unfortunately seemed to learn that control inputs before the upset (which were ignored) affected the outcome, so they might randomly use large control inputs which were unrealistically ignored in the sim. Pilots also came out of training with an exaggerated expectation of wake turbulence because they were unable to control the plane during the early part of the upset. An additional problem is that simulators are based from data gathered during aircraft testing and represents the normal flight envelope of the craft. Using the simulator outside this regime can result in aircraft responses which do not simulate the actual aircraft responses. Another issue was the way the limits of the rudder pedals were handled between different revs of the A300. Like the steering wheel of a car the rudder pedals require less and less input to be effective as speed increases. To prevent overstress on the airframe, the mfr will somtimes shorten the distance you can physically move the rudder pedals. Another way to achieve this effect is to change the ratio of rudder pedal movement to actual rudder movement (at cruise speed, a given movement of the rudder pedal will result in a smaller rudder movement than at low speeds). The mfr changed from one scheme to the other between A300 models and additionally lightened the amount of force needed to move the pedals. All this added up to a pilot making the same rudder inputs and getting a vastly different rudder response depending on the revision of the A300 they were flying. Finally, I believe there was concern with pilots' understanding of the V_A speed (one of many V_* speeds that the mfr provides in the aircraft's operation manual). V_A is the maximum aircraft speed at which you can jam one of the flight controls to its stops and keep the plane in one piece. Most pilots the NTSB surveyed didn't realize that this only applied to a single maneuver/single control input. Putting the rudder pedal to the stops back and forth repeatedly, even under V_A would allow the energy to build and eventually cause the airframe to break up. Full NTSB report: http://www.ntsb.gov/investigations/AccidentReports/Reports/AAR0404.pdf http://www.ntsb.gov/investigations/AccidentReports/Reports/A... I think the NTSB reports are fantastic reading for anyone interested in the failure modes of people and machines as well as how to really approach and investigate difficult problems.
rplst8 10y ago
One word. DO-178B.
throwaway4891a 10y ago
Also: Air Crash Investigation / Mayday / Air Disasters is an interesting series if one is interested in real-world, root-cause engineering analysis.

272 points by maxcan 10y ago

What wasn't mentioned directly was that since 2009, US passenger airlines have had a perfect record of zero fatalities, which is mind blowing.

mikeash 10y ago
Especially mind blowing considering that not all that long ago, it was common to see hundreds of airline fatalities per year, despite much reduced traffic. If the 1960 level of safety carried through to today, we'd see over three thousand airline deaths each year.
- imglorp 10y ago
  Furthermore, fatalities are still in the hundreds world-wide, often with the same equipment flown by US airlines. This number tells us that the FAA+NTSB, despite some other criticism, _usually_ has its act together where it counts the most. They iterate on identifying safety problems and force the airlines to adopt solutions.
  - npunt 10y ago
    US airlines often sell their older planes to foreign airlines, which may account for some of the increased issues those airlines have. Your point still stands though - culture and processes rule.
    - mjevans 10y ago
      You could argue that those planes should be forced to retire instead. A different perspective might be to require that every 10000 hours of service see a new mark painted on the hull. This way passengers could be informed of how much 'age' a plane has had and the regulations could also apply to any plane which /might/ fly in to the US (or Europe).
      - cjrp 10y ago
        The hull's age only tells part of the story though; how well it's maintained, and the ability of the pilots flying it would need to be factored in as well.
        mikeash 10y ago
        Those are probably far more important. I'd be far more comfortable boarding a 30-year-old airliner with great maintenance and great pilots than boarding a brand-new 787 with a clueless maintenance crew and pilots who aren't quite sure which way to move the stick (edit: yoke) to make the houses get smaller. If you look at the causes of crash, it's usually pilot error or improper maintenance, occasionally design flaws, but not equipment age.
  - enjo 10y ago
    It brings to mind a flight just last week that crashed due to likely negligence (not carrying enough fuel). http://www.cnn.com/2016/11/29/americas/colombia-plane-crash-investigation/ http://www.cnn.com/2016/11/29/americas/colombia-plane-crash-...
    - imglorp 10y ago
      So it turns out not only was that guy negligent, but there was also another plane in the pattern who got priority because, at the time, it looked like that other one was low fuel also (maybe a false alarm). Here's a reputable source. https://fearoflanding.com/accidents/not-enough-fuel-the-disgusting-truth-about-lamia-flight-2933/ https://fearoflanding.com/accidents/not-enough-fuel-the-disg...

nkrisc 10y ago

I know perfectly well that a fear of flying is often an irrational, though very real, fear for many people. But it's exactly the irrational nature of it that boggles my mind: they should be terrified of driving, but they aren't. I'm way more comfortable flying than driving.

TeMPOraL 10y ago
It's weird how the brain works. I know very well that airliners are safe, and I'm aware just how much work goes every day to ensure they stay safe. And yet, every time I'm flying, I have this uncontrollable thought, "maybe this is the day we all crash and die". I always get a little anxious during takeoff and landing, I listen to all the noises and worry about them. "This doesn't make sense, the plane didn't usually make that sound". "Oh shit, the lights blinked, wiring must be damaged." Stuff like that. It usually takes me few minutes after liftoff to get those thoughts out of my head.
- existencebox 10y ago
  At least for me there's a feeling of autonomy. If a plane crashes, there's very little that I as a non-superman can do to stop that from happening (with me inside it.) While driving, I can trust my own reflexes, awareness, and decision making to put myself in better or worse situations. Now this may still absolutely be a false perception, but this is where the feeling is rooted for me (from lengthy introspection on similar situations when I used to do trick-skiing and never really questioned my safety, but would be very hesitant to e.g. skydive). To some degree I think it's potentially also "failure magnitudes." Having been in a handful of very minimal car crashes, that calibrates my "something goes bad" prior to a not very impactful bar; whereas if something goes wrong on a plane, it can go VERY VERY wrong. Again, this is likely a false perception due to observations of outlier events in both cases and sparsity of signal, I'm just trying to question the origin of my feelings.
  - mortenjorck 10y ago
    There's this thing that will happen to me when I doze off as a passenger in a car: When I wake up, there's often a split second of panic between the visceral understanding that I'm in a moving vehicle and the slightly delayed recollection that I'm not the one responsible for driving it. Somehow, this turned into a problem for me when flying (even though I'm not a pilot). I wake up, I instantly know I'm moving at 500 miles per hour through the air, then I have a singular panic of being in a vehicle not under my control. This has only happened once – I haven't slept on a plane since.

blakesterz 10y ago

"2009: A Colgan Air (Continental Connection) Dash-8 Q400 crashed outside Buffalo. Fifty people were killed, including the occupant of a house struck by the plane." "2006: Forty-nine people perished when a Comair (Delta Connection) regional jet crashed after attempting takeoff from a too-short runway." I'm not sure, maybe "major" has an official cut off number, but 49 or 50 doesn't seem "minor". It's a long way from 260, but it's far from 0.

gr3yh47 10y ago
They're counting Major US airlines, not including regional carriers FTA: >To be clear, there have been a handful of tragedies involving regional carriers and freighters [...] It’s the majors that set the standard, and it’s only fair that we measure from there.
el_benhameen 10y ago
Colgan and Comair aren't counted in the metric because they're regional, not major/legacy, airlines. They codeshare with the majors, but they're different organizations. Tangential: Frontline had a great episode on regional airlines' safety issues. http://www.pbs.org/wgbh/pages/frontline/flyingcheap/ http://www.pbs.org/wgbh/pages/frontline/flyingcheap/
- js2 10y ago
  Tangent on the tagent: Phillip Greenspun touches on regional airline safety in this post comparing US to foreign airline safety - http://philip.greenspun.com/flying/foreign-airline-safety http://philip.greenspun.com/flying/foreign-airline-safety If having a newbie in the right seat represents a challenge to safety, how do U.S. regional airlines manage? The U.S. regional airline is hiring pilots who are more experienced than those hired by foreign airlines (the U.S. regional F/O typically has been a flight instructor) but less experienced than those hired by a U.S. major airline. The answer is that U.S. regional airlines have an accident rate as much as 5-10 times that of U.S. major airlines. The culture is the same, but apparently there is no substitute for experience.
- mannykannot 10y ago

slg 10y ago

I am surprised there wasn't more than a token mention of terrorism. People like to bemoan airline security and I think there is a valid argument regarding the effectiveness of the TSA, but this is also the safest stretch the US airline industry has had against terrorists and hijackers since the early days of the industry in the 40s and 50s.

mikeash 10y ago
I think the current level of airline security is overkill, but I'm always a little surprised to see people advocating for having none. In the late 60s and early 70s, US airliners were getting hijacked at a rate of once every week or two. Once basic security screening was introduced, it became far more rare. Making people take off their shoes and empty their water bottles is a bit silly, but basic checks for guns and bombs are a good idea.
- cbr 10y ago
  In the late 60s and early 70s, US airliners were getting hijacked at a rate of once every week or two. https://en.wikipedia.org/wiki/List_of_aircraft_hijackings https://en.wikipedia.org/wiki/List_of_aircraft_hijackings is pretty dramatic
  - tunesmith 10y ago
    Hijacking was pretty different back then, which is why passengers were used to the idea of staying uninvolved and not escalating things. That's part of why 9/11 was just a (culturally) unimaginable shift, it wasn't really in the gestalt that a hijacker's main goal was other than to kill/die.
    - JshWright 10y ago
      Yeah, 9/11 could work exactly once (and even then, it only in three out of four concurrent attempts attempts)
- throwaway2016a 10y ago

NickHoff 10y ago

The effect of culture should not be underestimated here. There really is a cooperative spirit among pilots, air traffic controllers, weather briefers, mechanics, the FAA, etc. Controllers are willing to help pilots with special requests if able, pilots file weather reports for other pilots if what they encounter aloft is different from what they expected, mechanics take pride in their work and are highly regarded by pilots. I generally take a dim view of regulators, who generally seem to have less expertise than the people they're regulating. That's mostly not the case here. I'm libertarian and a pilot - and I like the FAA. Nobody is out to nail you for a minor technical infraction, but they will yell at you for something that's unsafe but technically legal. This encourages people to report near misses, even if casts them in a bad light. The regulations themselves are generally reasonable, and frankly feel like they were written by pilots, controllers, and airport operators. The point is this - regulations matter and budgets matter, but if the culture is toxic, people will find a way around it anyway. When I see a problem elsewhere in society - investment banking in the 2000s for example - people propose heavy regulatory solutions. That may work, but wouldn't be so much of a battle if the culture were better. I haven't thought deeply about what "better" means in this context, nor about how to intentionally change a culture. I'd rather just fly.

maxcan 10y ago
> I'm libertarian and a pilot - and I like the FAA Me too. For every frustration about what they've done to the costs of things in GA, and those are valid, their accomplishments in general are incredible.
THE_PUN_STOPS 10y ago
I've often heard it said that FAA regulations are written in blood. Meaning, if a rule exists, it exists for a damn good reason. For anyone interested, NASA's ASRS [1] (Aviation Safety Reporting System) is what the parent commenter is referring to regarding reporting near misses. The idea is, anyone can submit an incident report with NASA, a neutral separate party from the FAA, which is placed in an anonymized public database and cannot be used to justify punishment against anyone involved in the incident. This promotes transparency and learning from the mistakes of yourself and others, instead of covering them up. Frequently, the person who caused an incident in the first place will promptly report it to ASRS, as the FAA looks favorably upon those who come forward to ASRS on their own and dimly upon those who do not. I think more industries could use this kind of transparency in reporting sketchy situations. [1]: https://asrs.arc.nasa.gov/overview/confidentiality.html https://asrs.arc.nasa.gov/overview/confidentiality.html

chinathrow 10y ago

RIP to the freight dogs though: "2013: A UPS flight crashed on approach to Birmingham, Alabama, killing the two pilots."

chinathrow 10y ago
Why the downvote? UPS is a major airline - even if they only transport cargo, humans still had to give their live. Also, "freight dogs" is the gentle term for the pilots in cargo planes.

js2 10y ago

In no way directly related to this post, but I'm always looking for an excuse to link to my two favorite airline stories. 1) The Turn - http://www.theatlantic.com/past/unbound/langew/turn.htm http://www.theatlantic.com/past/unbound/langew/turn.htm 2) The Crash of United Flight 232 talk by Capt. Al Haynes - http://clear-prop.org/aviation/haynes.html http://clear-prop.org/aviation/haynes.html

GFischer 10y ago

Plane crashes have been going down not only in the U.S., but worldwide, which is commendable. A bit unfortunate for this to be posted a few days after the Chapecoense crash shook the world (noted in the postscript), but that was a regional charter flight, not a major U.S. airline. Interesting infographic: http://www.ibtimes.com/how-many-planes-crash-every-year-how-many-people-die-plane-crashes-chart-1560554 http://www.ibtimes.com/how-many-planes-crash-every-year-how-... Edit: the Chapecoense crash shows exactly why the FAA's work is so commendable - disregard of proper procedure and safety measures led to that accident.

gus_massa 10y ago
About the Chapecoense crash. From [Portuguese] http://www1.folha.uol.com.br/esporte/2016/12/1838707-lamia-ja-havia-descumprido-regra-de-combustivel-em-voo-com-messi.shtml http://www1.folha.uol.com.br/esporte/2016/12/1838707-lamia-j... Autotranslation: https://translate.google.com/translate?sl=auto&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=http%3A%2F%2Fwww1.folha.uol.com.br%2Fesporte%2F2016%2F12%2F1838707-lamia-ja-havia-descumprido-regra-de-combustivel-em-voo-com-messi.shtml&edit-text=&act=url https://translate.google.com/translate?sl=auto&tl=en&js=y&pr... > On at least two occasions, the Bolivian company had flown beyond its declared autonomy. [4 hours and 22 minutes] > One between Cochabamba in Bolivia and Medellin in Colombia (4 hours and 27 minutes), and the second between Medellín and Santa Cruz de la Sierra in Bolivia (in 4 hours and 32 minutes).
- wapz 10y ago
  Wouldn't the difference in 5-10 minutes not be the reason for the lack of fuel, though? I would guess a 4 hour flight would have enough fuel for at LEAST 5.5 hours if they couldn't land at an airport and had to be diverted to another one (right before landing).
astrodust 10y ago
Flying is so amazingly safe now that you're far, far more likely to be killed on a cab ride to the airport than on the plane itself.

estrabd 10y ago

9/11 doesn't count? Sounds like "fake news."

JHof 10y ago

Exploiting this aviation-related post to see if any aerospace folks will comment on my two-month old Ask - Ask HN: Why is the National Airspace System so dated? https://news.ycombinator.com/item?id=12651244 https://news.ycombinator.com/item?id=12651244 In regard to the safety of regionals vs. the major airlines that the article's author is writing about: I can see why the statistics are separate. Even though, according to the FAA, we're all under the "same standard", in practice, this isn't the case. As long as the paperwork looks the same, though, we're "safe". Also, within the industry, some pilots tend to think of themselves as shouldering the responsibility of safety, however, I sometimes think we're safe despite the pilots. Engineers of aircraft, airspace, arrival/departure/approach procedures, airports, ATC systems, etc. are silently keeping you safe. That the FAA is able to manage all of these elements is quite impressive. Still, its massive bureaucracy can be quite Vogon-like.

rplst8 10y ago
I think a lot of software engineers would be blown away at how strict software standards are for avionic systems. Like 100% modified condition/decision coverage for all software tests, strict. No pointer math and no dynamic memory allocation, strict. https://en.wikipedia.org/wiki/Modified_condition/decision_coverage https://en.wikipedia.org/wiki/Modified_condition/decision_co...
- gurneyHaleck 10y ago
  Being a party to lackluster software development on a regular basis, I somewhat pine for projects this strict. As part of my daily routine, feigning surprise at the fact that, given a tree of decisions, I'm only permitted to code for the ones that make money, simply to find that "Quality Testing" reveals the 500 other permutations will inevitably result in thrown exceptions with hideous, glaring error messages. "WHOOPS."
  - rplst8 10y ago
    This sort of thing is exactly what some of the software development processes are all about. When I was in a CMMI training class 10 or more years ago, the instructor we had said basically that much. By instantiating a process, it takes the mental weight of knowing (or checking) if everything is correct. No one builds cars that way - why should we build anything that way. The processes should catch the inevitable human errors.

rl3 10y ago

This would be why I always tell people to fly major U.S. legacy carriers if possible, or their western equivalents (e.g. British Airways, Qantas). As a general rule, major carriers from countries that have a pool of predominantly native English speakers to hire from.

jpatokal 10y ago
Actually, only two of the world's safest airlines hail from the US: http://www.telegraph.co.uk/travel/news/The-worlds-safest-airlines-revealed/ http://www.telegraph.co.uk/travel/news/The-worlds-safest-air... And 6 of the top 10 are not from English speaking countries (ANA, Cathay, Emirates, Etihad, EVA, Finnair).
- rl3 10y ago
  Actually, only two of the world's safest airlines hail from the US The list you just cited includes Alaska and Hawaiian on that list, for a total of four US carriers. My original statement was in context of major legacy carriers, so I could understand you excluding those—however, each of the airlines you cited has a comparable—if not smaller—fleet size to that of Alaska Airlines. Also, keep in mind that major US carriers have underwent major consolidation in recent years: US Airways and American Airlines United and Continental Northwest and Delta >And 6 of the top 10 are not from English speaking countries (ANA, Cathay, Emirates, Etihad, EVA, Finnair). If you total the fleet sizes of each of these, you end up with 907 aircraft. American Airlines alone operates 952 aircraft. Moreover, many of the airlines you listed tend to hire Western pilots formerly of major legacy carriers—captains in particular—in order to meet high PIC experience requirements. - It's also worth mentioning that AirlineRatings.com[0] (the source of the list you cited) currently assigns Malaysia Airlines a safety rating of 5/7, despite the fact they managed to lose two 777 hulls (MH370, MH17) only months apart—one of which was perfectly preventable (don't fly over active war zones?). The same site affords Asiana Airlines a similarly lofty rating of 6/7, despite a 777 hull loss (Asiana 214) due to its pilots being unable to perform a simple visual approach in ideal conditions at SFO. A truly stunning display of incompetence that belied a terrible training and safety culture at the airline. Of course, going by safety records alone, I'm sure both airlines looked great on paper about five years ago. [0] http://www.airlineratings.com/safety_rating_per_airline.php http://www.airlineratings.com/safety_rating_per_airline.php

nikcub 10y ago

There's likely no media coverage of the anniversary because only including major airlines is a very arbitrary cutting off point. The NTSB lists five major passenger air accidents since 2001: http://www.ntsb.gov/investigations/data/Pages/paxfatal.aspx http://www.ntsb.gov/investigations/data/Pages/paxfatal.aspx No agency breaks out "major" airlines from "regional" for the purpose of regulation or statistics, nor does any other nation. It is purely self-serving. What makes this post weirder is while cutting out regional airlines from the statistics, the causes he attributes with improving safety for the majors are the same things that regional airlines have also been doing.

zardeh 10y ago
Of those 5, 4 occurred in propeller craft. I think that it is relevant to include that kind of difference. Regional airlines and major airlines do work differently.
erroneousfunk 10y ago
He does mention that in the post, and, yes, I think the author would also agree that it's self-serving. However, I would also argue that this is a symbolic milestone of the increased safety of US commercial flights, and an interesting thing to reflect on. I wouldn't take it too literally as any sort of real statistical measure, but it's certainly worthy of a blog post.

sean_patel 10y ago

> Some will argue that it’s unfair to gerrymander statistics in this fashion, separating airlines into categories of convenience. However, it’s necessary. Regional carriers, for instance, can have substantially different cultural and operational environments than their legacy carrier affiliates. It’s the majors that set the standard, and it’s only fair that we measure from there. Yes. It is gerrymandering, and no you can't wash your hands of the regional carriers' crashes and deaths after you've stamped your 'trusted' name and brand on their flights. I am appalled at how they callously dismissed those plane crashes since 2001 like this. > To be clear, there have been a handful of tragedies involving regional carriers and freighters... in which dozens died. Really? 'dozens died' is ok, but 100s died is not? Are we cattle?

rplst8 10y ago
> Really? 'dozens died' is ok, but 100s died is not? Are we cattle? Yes. Some people are choosing to pay less for a little more risk. The market should allow for that in the fringes. The vast majority of flying happens on the major carriers, so having stricter standards for the majority of your flying passengers is the right way to approach safety.

throwaway729 10y ago

Does anyone know what this is referring to? > We’ve engineered away what used to be the most common causes of accident

sowbug 10y ago
I'm not enough of an experienced pilot to authoritatively answer, but here's a guess at some items. * Sterile cockpit: a rule that pilots avoid any distracting activities during critical parts of flight (mostly takeoff/landing, flying around busy areas). * Crew resource management: a shocking number of major commercial accidents have involved either confusion in the cockpit or one pilot who knew something was wrong but didn't feel he/she had the authority to say something. Tenerife is a very sad example. CRM is a set of procedures that specify how concerns are brought to the PIC (pilot in command) in a way that's unlikely to make things worse. Basically, it's an escalation procedure that is recognized on both speaking/listening sides as something that needs to be paid attention to. * Checklists: Not sure if this is new, but it is completely routine in flight to consult standardized checklists for everything, whether ordinary takeoff/flight/landing or emergency situations. It's thus a lot less likely you'll make a stupid mistake in flight like running out of fuel because you forgot to switch tanks. (Actually people still run out often, but the out-of-fuel checklist includes ensuring you've remembered to switch away from the empty tank, meaning people don't crash planes with half their fuel anymore.) There is no cultural shame in consulting checklists; in fact my (limited) experience is that most pilots would question the judgment of someone who didn't use checklists. This is a bigger deal than you might think because most aircraft incidents are fundamentally attributable to human error. * Better weather technology, both for prediction and real-time assessment. Flight 1549 (the one that landed in the water with no deaths and just a couple serious injuries) is a good example of the first three. You can listen to the cockpit recording to see them all in action. When the birds hit, the pilots were already 100% focused on the critical takeoff portion of the flight -- zero distractions that could have delayed reaction time. The copilot immediately got out the POH (plane's owner's manual containing tons of checklists) and started all the restart procedures while the captain reflexively put the plane in the maximum glide position to extend range. Then right as they're lined up for the river, Sullenberger asks the copilot "Got any ideas?" While this might sound like comical desperation, it's actually good CRM in action -- after they'd tried the checklists and were in a somewhat stable descent, the captain solicited feedback from the second-in-command, just to make sure they hadn't overlooked anything else. Anyway, I'm sure my list is far from complete. Someone more knowledgeable can continue it.

ahupp 10y ago

"... the first officer of flight 587, Sten Molin, who was at the controls, overreacted, rapidly and repeatedly moving the widebody jet’s rudder from side to side,... . The vigor of his inputs caused the entire tail to fracture and fall off." I was interested in learning more about this, and it turns out there's an alternative interpretation that argues he was flying according to best practices at the time rather than "overreacting": http://articles.latimes.com/2005/jan/11/opinion/oe-garrison11 http://articles.latimes.com/2005/jan/11/opinion/oe-garrison1...

MaxfordAndSons 10y ago
IIRC from watching the episode of "Air Crash Investigation" (aka "Mayday") on this flight, it was a pretty tragic confluence of factors: it was a best practice on other planes, but not on the A300, but American Airlines training manual had not been updated to reflect this change, nor did the simulator models account for this eventuality. Kinda surprised the author of this piece wrote it up as he did.
- rrmm 10y ago
  If I'm remembering the details of the case correctly, it turns out he did overreact and that did cause the tail to fail: Certainly the tail did fail, and the pilot's input caused it. The important part and the part that has been noted above is the NTSB's (and the industry's) focus on not blaming the pilot but figuring out why the situation turned out the way it did, what factors allowed it to happen, and how things could be changed to insure the same chain of events doesn't keep occurring. (Again if I recall correctly), It was the confluence of a few different events compounding each other and resulting in the accident: One factor was the airline company sim training on recovery from upsets. Basically the pilot going through the training could come away learning entirely the wrong lesson. The simulation was hardwired so that the pilots' inputs were disregarded until the plane was in an upset and then the pilot took over. Pilots unfortunately seemed to learn that control inputs before the upset (which were ignored) affected the outcome, so they might randomly use large control inputs which were unrealistically ignored in the sim. Pilots also came out of training with an exaggerated expectation of wake turbulence because they were unable to control the plane during the early part of the upset. An additional problem is that simulators are based from data gathered during aircraft testing and represents the normal flight envelope of the craft. Using the simulator outside this regime can result in aircraft responses which do not simulate the actual aircraft responses. Another issue was the way the limits of the rudder pedals were handled between different revs of the A300. Like the steering wheel of a car the rudder pedals require less and less input to be effective as speed increases. To prevent overstress on the airframe, the mfr will somtimes shorten the distance you can physically move the rudder pedals. Another way to achieve this effect is to change the ratio of rudder pedal movement to actual rudder movement (at cruise speed, a given movement of the rudder pedal will result in a smaller rudder movement than at low speeds). The mfr changed from one scheme to the other between A300 models and additionally lightened the amount of force needed to move the pedals. All this added up to a pilot making the same rudder inputs and getting a vastly different rudder response depending on the revision of the A300 they were flying. Finally, I believe there was concern with pilots' understanding of the V_A speed (one of many V_* speeds that the mfr provides in the aircraft's operation manual). V_A is the maximum aircraft speed at which you can jam one of the flight controls to its stops and keep the plane in one piece. Most pilots the NTSB surveyed didn't realize that this only applied to a single maneuver/single control input. Putting the rudder pedal to the stops back and forth repeatedly, even under V_A would allow the energy to build and eventually cause the airframe to break up. Full NTSB report: http://www.ntsb.gov/investigations/AccidentReports/Reports/AAR0404.pdf http://www.ntsb.gov/investigations/AccidentReports/Reports/A... I think the NTSB reports are fantastic reading for anyone interested in the failure modes of people and machines as well as how to really approach and investigate difficult problems.

rplst8 10y ago

One word. DO-178B.

throwaway4891a 10y ago

Also: Air Crash Investigation / Mayday / Air Disasters is an interesting series if one is interested in real-world, root-cause engineering analysis.