Hacker News new | past | comments | ask | show | jobs | submit login
Aviation safety: Transport comparisons (wikipedia.org)
45 points by tomerbd on Nov 16, 2019 | hide | past | favorite | 62 comments

It kinda bothers me how much safer cars are than walking and cycling, given that most of these deaths are probably caused by cars. ('Caused' not necessarily meaning the accident was the drivers fault, but the death being due to the car being big and made from metal vs. humans being soft and squishy.)

Roads are basically "killing people here is ok as long as you follow few simple rules" zones. It's insane.

If I told people "I'm going to shoot my gun in the city from time to time - if you don't want to die please step away when I light this red laser sight" and then did exactly that and blamed people for not avoiding it - that wouldn't fly. But cars are so convenient it's allowed.

What you're describing is basically what happens in France during hunting season in the countryside. Luckily, the most frequent victims are the hunters themselves, but sometimes random hikers too. "But we had put up a sign that there was a hunting session in the forest!" seems to be a valid defense most of the time.

"What happens is, men put on bright red hats, and they go out in the woods, and they shoot at each other. There's no penalty for this, and it's a lot of fun. Once in a while someone even shoots a deer!"


Sounds pretty similar to US deer hunting season then, except beer wasn't mentioned anywhere.

There's something special about the recklessness of deer hunters specifically, as other game seasons don't seem to have this problem.

Calling it convenience is minimising the importance. Economic value is also important. This is most obvious when talking about trucks, but people being able to move large distances rapidly has a positive benefit to society.

The value to society of trucks is undebatable but I'm not so convinced for the personal car.

If we're talking about people moving large distances rapidly, there's solutions that don't involve transforming our entire living habitat to death zones.

These solutions dont apply to rural or small-town environments, but do apply to cities. And it just so happens that cities, by virtue of their population density, maximize the negative externalities of cars (less open space and more people means the pollution is more concentrated and affects more people. the noise of every honk is annoying more nearby residents, etc)

In conclusion, cars aren't necessary in cities but are awful for the residents, even for those who choose to never touch a vehicle.

> In conclusion, cars aren't necessary in cities but are awful for the residents, even for those who choose to never touch a vehicle.

Disagree, for many they are a necessary evil. You can live without but you have to contort your lifestyle.

> for many they are a necessary evil

Don't get me wrong, on an individual level I strongly agree that cars are necessary for probably 90%+ of the united states for a reasonable standard of living. That's kind of the issue: we've built a spread out, suburbanized environment where the only people who can live without a car, are the people who live really close to the downtown part of the city. These people, in an ironic twist, have to breath the fumes and hear the sounds of the hundreds of thousands of cars belonging to those outside city center who drive through every day.

My argument is just that it doesn't have to be that way.

I think cars have pretty much shaped our world in a way where they became a necessary "evil".

Simplified thought: the local grocery dies because everyone can easily drive 10km to the supermegamarket, which makes driving there the only choice.

It would be interesting to see how our world would have evolved without cars. There sure would be a lot missing, but also a lot of things simply done in a different way. Maybe this kind of SF already exists? Cyclepunk anyone?

Have you heard of gun ranges? There are so many examples of situations where you risk death if you don't follow rules, your argument isn't persuasive at all.

Except gun ranges are not just liberally sprinkled throughout cities, making it difficult to leave the house without having to move through one.

Gun ranges are private property not public space.

Can't find a reference now... but I read a while ago that you can count the risks of cycling differently, if you want.

If you cycle to work, you get perhaps an extra hour's phyiscal activity compared to driving, and you get some risk of being squished to death by a car. An extra hour's moderate physical activity is good for you, being squished to death is absolutely awful for you, and the story I read said one of the German health insurers did a study and found that the former effect is the bigger for most of the German population. (Or put differently, most Germans don't exercise enough.)

Also would be interesting to know the scale of air pollution fatalities & similar side-effects

Commercial aviation is super-safe. General aviation is pretty dangerous. The statistics don't make much sense without those broken apart.

The table that's linked is also in the UK for the decade between 1990-2000, so about twenty years ago.

I'd wager that in first world countries, aviation has gotten safer since the 90s as well. The last decade has seen exactly one commercial aviation fatality in the US during the engine failure on a Southwest flight last year, while there are about ten deaths per day on average in the US for cars.

In comparison, Japan has had zero fatalities in 55 years of operation for the Shinkansen (super express trains).

I don't really see the sense for breaking this down by nation, beyond having nation-scale car statistics to compare against... I mean, the Boeing max fuckup could well be regarded as "made in the US" no?

There are different levels of maturity of airline operations and plane reliability between countries, and this is recognized by both the EU and the US. They both have different lists of airlines (for the EU) and countries (for the US) that are not allowed to operate in their respective airspaces without a wet lease (a lease from another company where that company assumes maintenance operations and provides the personnel to operate the plane).

It also doesn't really make sense to put regional carriers from small countries (eg. Gabon or Kyrgyzstan) in the same bucket as strictly regulated large carriers that follow the ICAO standards.

Does anyone know why that's the case? Is it just that piston engines (and single-engine planes) are less reliable than commercial planes with redundant jet engines?

The engines aren't the big difference. It's the operating procedures in commercial aviation that are much different. If I step into a single engine Cessna in my private time, I'm allowed to fly with almost no equipment, very little reserve fuel, no plan for an alternate airport, in shitty weather all by myself.

Now if I step into a plane with passengers (even if it's the same small Cessna), suddenly I'm no longer allowed to fly single pilot IFR operations without autopilot, need much higher fuel reserves, need to plan an alternate, check weather requirements and have done a 6-monthly checkflight and training. And for most airplanes I would also need a co-pilot, so there are two people monitoring everything.

Most small airplane crashes are not due to engine or equipment failure, mostly it's pilot error, and that's where all the extra procedures in commercial aviation really help.

Engines are a big one. But the reasons are innumerable. Commercial aviation is always scheduled, so you have a very definite takeoff time, location, and destination. This cuts down the variables encountered along the way greatly and allows for precise route planning taking weather and traffic into account at all times. Commercial flights are also 100% IFR even in perfect weather, which means their every maneuver is coordinated with ATC. ATP pilots are also literally 100x the experience of your average GA pilot with a couple hundred hours. Not only that but they are in regular practice, not just someone hopping in the left seat for potentially the first time in years. They're also held to a much higher medical clearance level by the FAA, so you have far fewer medical incidents inflight. GA also includes training flights too, which contribute a significant amount to the accident numbers.

Equipment used is part of it, but smaller than you’d imagine.

Well maintained single engine aircraft (in the certified category) are generally reliable. The majority of incidents in these planes come down to avoidable pilot error. Looking at the forces behind pilots making mistakes will provide a lot of insight. General aviation is not forgiving to mistakes and tired, under-trained, overconfident or briefly otherwise human behavior can (and does routinely) cause issues. These factors are more carefully compensated for on the commercial side.

Of course, the counter argument here is the missing safety features that should exist to compensate for mistakes.

One more reason, not mentioned above, is ability to fly in weather. CA airplanes have proper de-icing for wings, are big, and fly above the clouds.

A lot of GA -- certainly anything under around $200k -- is a fair-weather activity. Planes can't really fly in the winter -- ice builds up on wings. Strong winds throw small airplanes around. There isn't instrumentation to land with limited visibility from fog, severe snow, or severe rain. Etc.

Weather can change unpredictably. If you've flown for a weekend getaway Friday, the weather turns over the weekend, and you have an important business meeting Monday morning, do you risk flying back? Do you miss your meetings? Do you leave your airplane there and fly back commercial, with thousands of dollars in hanger+rental fees and days of hassle to get your airplane back?

It's a value call. Different people have different risk tolerances based on personality, life situation, and culture.

A lot of this applies to cars too actually -- if everyone drove conservatively, they'd be pretty safe. At least where I live, virtually everyone speeds. People cut other cars off. Etc.

Redundancy plays a part, but it is not a major one. Major differences include the amount of scrutiny and government supervision in terms of inspections, certifications, regulations, maintenance, traceability, medical checks etc. that commercial aviation requires and the difference in technology and support between a modern commercial airliner and the average general aviation plane.

Also, it got significantly safer by time.

Deaths per KM is 60x lower when flying rather than driving a car. Even for short haul flights, flying is considerably less deadly than driving the same distance.

You can’t replace flying to Hawaii with driving there. Further, short flights which may be directly comparable are vastly more dangerous per mile making the overall average meaningless.

Thus it turns out if your going on a relatively short trip ~300km or less taking a bus at ~0.04 deaths per billion passenger km is safer.

That’s one way to look at it, but i think the general feeling people have regarding air travel ( aka scare) is better explained by the « death per journey » statistic (in addition to the obvious fact that a plane .. flies).

Not to mention the fact that once you remove obviously dangerous car behaviors ( such as dui or driving with a wreck), chances of accident probably decline a lot.

Wouldn't it be more reasonable to measure per travel occasion instead? When thinking about car travel, the type of road also matters, as does the rest of the surroundings. True for airplane travel as well, but only arrival/departure seems to matter most there.

I think per travel time would be better. "Trips" in air and automobiles tend to be of very different average durations.

Car risks grow linear with length of trip while plane risks are concentrated especially at takeoff and landing. If you only count deaths/trip, it would discount much relevant information.

I too think we should reinvent the discipline of statistical analysis in this thread.

For example, richer people fly, should we adjust for that?

I love these dim headlines

I'm surprised to see buses listed as being much safer per journey than trains. I guess that's just because "per journey" is a rather meaningless metric. Train journeys must be longer than bus journeys on average. However, even when you compare bus vs train on a death per billion km base, the bus is much safer (0.4 vs 0.6). Is it because on average buses travel slower than trains? How accurate are these numbers?

I think a lot of these don't suffer direct comparisons well. eg, the greatest risk of accidents to a bus, comes from other passenger vehicles where the bus is more likely to win that particular dispute. The greatest risk of accidents to a train, comes from the trains own inertia, where the train is much more likely to lose. The greatest risk of accidents to an aircraft is the ground, in which the ground has an undisputed record.

So I think it's safe to assume busses would be more likely to have accidents, but less likely to have fatalities.

I think it actually gets worse if you try to normalise the results. There is greater risk in a car journey from LAX to JKF, than there is for an air journey along the same route. However, if I tried to commute by air to work (a 2km commute), the risk per km would rocket, as take-offs and landings are the most dangerous part of each journey, so with an atypical number of take-offs per km, I'd get an atypical risk per km.

The wiki article does go on to point out the difference in usage of these statistics. per-journey does make more sense for air insurance, as the difference in risk between long-haul and short-haul isn't really significant - the take-off and landing are still the riskiest phases, there's just a larger period of boredom between them for long-haul. So 10 short flights vs 1 10x longer flight might normalise to the same hours & kms, but 10 short flights means 10 landings etc (10x the chances to miscalculate fuel, 10x the entries into & out of congested airspace, etc)

But comparing per-journey between a car and a plane is rather meaningless, as they are on average very different journeys. commuting in a car isn't two phases of high risk separated by one phase of near-immunity.

indeed, per-journey is entirely meaningless, per-hour is the correct measure of risk.

A journey seems like a meaningful unit of work here. Sometimes the start or end of a journey is more dangerous. The decision to even start journey would look at the per journey risk perhaps. Think of space launches, as an extreme example. Risk per distance or per hour, if you consider orbiting for a while, could be low but risk per journey quite high.

The point is the different modes have entirely different profiles.

You don't fly to the supermarket, or walk to an exotic holiday location.

There is a Wikipedia tangent article called Micromorts which is "unit of risk defined as one-in-a-million chance of death." from a particular activity.

Travel section https://en.wikipedia.org/wiki/Micromort

This is the kind of chat that gives statistics a bad name. How does it make any sense to compare deaths per journey on foot to skydiving? What even defines a journey on foot? Why would you include the number of kilometers for the space shuttle, which is dominated by the arbitrary safe time spent in orbit, but not for skydiving? It also makes air travel look more dangerous by combining commercial and non-commercial air travel, while making cars look safer by separating out motorcycles. Some rigor would make this more honest.

Considering most daily (commuting, grocery shopping) trips are either very short or slow due to traffic, I fail to see how this fact is interesting at all.

I could imagine people have a small background process thinking « i’m taking my car to go to the office, how much chances do i have to make it safe to the end » every time they take their car. And same goes for air trips: « i’m taking the plane to SF, am i going to die ? »

Distance travelled probably is a second thought compared to the concept of actually having a trip.

In a way your are comparing apples to oranges unless you can choose between either mode of transport for the same trip. Trying to directly compare car and plane journeys doesn't really make sense. Cars vs buses vs suburban trains, or planes vs high-speed trains are better comparisons.

For example, comparing the risk of flying or taking the Shinkansen between Tokyo and Osaka is a fair comparison. But you would probably not choose to drive that journey.

Interestingly around 50% of car accidents occur within 5 miles of the individuals home which is barely double the length of an airport runway!

Number of trips may be a better risk measure when comparing plane flights as most crashes occur on take off or landing. The actual distance traveled is probably irrelevant.

Yes, because air travel have less trips but quite longer. Title is as fair as me saying that space shuttle is safer than walking per billion km.

The pedal-cycle, and many of the walking deaths should perhaps be included in the 'car' deaths, since that is usually the cause.

Telling quote from that article:

> Aviation industry insurers base their calculations on the deaths per journey statistic while the aviation industry itself generally uses the deaths per kilometre statistic in press releases.

From the article: "The first two statistics are computed for typical travels for respective forms of transport, so they cannot be used directly to compare risks related to different forms of transport in a particular travel 'from A to B'."

Why post this? It's not a secret.

What is more worrying is the emissions of air traffic, which is enormous.

If I remember well, a single seat in an air plane is equivalent to a single car, per unit of distance. So it's like accelerated travel, meaning if you move a person 1000 km by air, it's equivalent to riding that person 1000km with a single car. It's the same emissions per distance, except it's done between 8 and 10 times faster (not considering takeoff).

Imagine replacing a single airplane of 100 passengers with 100 cars who don't share passengers.

Harassment & intimidation of pedestrians & cyclists per billion car journeys is substantially more than per billion air journeys.

The fact the discussion is about a table on Wikipedia, an encyclopedia, means we are at a stupid level.

Which fits why we'd care about 'journeys' over anything meaningful.

But if we pretend there's something to recover here, lets look at the source

"The risks of travel Archived September 7, 2001, at the Wayback Machine. The site cites the source as an October 2000 article by Roger Ford in the magazine Modern Railways and based on a DETR survey.

Data would be at least 20 years old, from a random website allegedly from a Railways magazine allegedly from a DETR survey.

Sure, if you count 5 minute drives to get cigarettes

Exactly, the per-hour rate is obviously the most meaningful, and then the per-km rate which in some sense includes benefit/risk.

Per-trip is nonsense, since the the different modes entail completely different trip lengths and distances.

If you're trying to figure out whether it's safer do drive somewhere than to fly, the only metric that is relevant is deaths per unit of distance.

That’s because when a car crashes it’s not typically travelling at 500mph with 250 people on board.

Phrased another way, given the figures of 40 for cars and 117 for planes, unless everyone on a flight all travelled to the airport together in no more than 3 cars there’s a higher risk of death on the way to the airport than on the flight itself.

It's not that simple. Car crashes are much more common than plane crashes, and per kilometre, car travel is much more deadly than air.

Not sure if this is a meaningful metric. Much better to compare the deaths per traveled distance.

Yet the research is all about self driving cars.

How is this even remotely interesting? Car journeys are far more frequent and shorter than airplane journeys. It doesn't mean anything.

The difference is control. For me[0] the probability of death in an airplane is higher than the chance of death by car - whether I'm a pedestrian or driving.

[0] Of course I could be wrong -- but I really don't think so

The problem here is that everybody thinks that they are a better than average driver.

There is a lot still outside your control when you are driving or walking, namely other people driving their cars. You can be the best driver in the world, it's not going to stop people behaving like idiots.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact