Unit of distance is not the correct measure to use for safety because it skews heavily towards speed. The fastest means of transport will almost always look the safest. Consider if planes doubled their speed overnight (which they can, it's just a matter of gas/costs) and crashes remained the same - did they instantly become twice as safe? Definitely not, there is still the fixed risks of takeoff, landing, and being in the air for even a second, etc. The additional miles are almost free in terms of risk, but eat away a lot at the risk metric.
The better measure is time: deaths per hour traveled. And by that measure buses are actually safest, followed by rail, then air [1][2]. Cars are still 4x more dangerous, but that's much less stark than the common wisdom. And even that is misleading because with cars specifically, there is much more variability of drivers, car quality, maintenance, etc. If you are sober, responsible, educated, with a well maintained car, etc. your odds are significantly lower. Think drunk drivers who only kill themselves in accidents (this is a significant portion of car deaths).
While you raise some good points about the "fixed risk costs" of flying not scaling by distance, deaths per hour isn't a much better metric. The use case for most people is to travel a specific distance. e.g. "I need to get from A to B" so should I fly or drive there. It's usually not, "I need to move around for an hour" should I take the bus or a plane.
This kind of analysis assumes that the destination is independent of the mode of travel. We only consider going from NY to LA so often because plane travel makes it practical. But this doesn't map well to the general claim that "airplane is the safest way to travel". A per trip metric seems like a more accurate measure of this claim, but then bus is the safest.[1]
What you want is the distance but that's not the risk you're being exposed to. As soon as you enter a plane you are being exposed to risk, like a radioactive element, regardless if the plane travels fast, slow or even in circles. Only when you get off the plane does your exposure to this risk change. The distance you traveled in the meantime had no direct effect. The fact that cars or buses take much longer to arrive means you have to factor in the additional hours of exposure to their risks, so it still makes a material difference how far you need to go, but the more direct relationship is the amount of exposure to the radioactivity.
I disagree; I don't want to spend 10 hours traveling, I want to get from point A to point B. Unit of distance is what makes sense. Who cares if I have half the chance of dying per hour, if I have to spend ten times the hours subject to that risk?
I'm not sure why you think just doubling the speed would impact the crash/unit of distance ratio. If ten planes do a 600km trip and one crashes, that's 1 crash per 6000km regardless of the speed at which they're traveling.
You're right that fixed per-trip risks are diluted, but that's because plane trips are longer, not because they're faster. The right comparison, in my opinion, would be to draw the curves of death rate / trip length.
> Consider if planes doubled their speed overnight (which they can, it's just a matter of gas/costs)
No, they really cannot. Jetliners can't even increase their speed by 20%, because that puts them through the sound barrier, which they can't handle. Planes must specifically be designed to handle that, like the Concorde or fighter jets. All planes currently used for civilian aviation cannot handle it.
Are you suggesting that people take a bus from New York to Los Angeles, because though the chance of being injured or killed is higher, it is worth it because the trip takes so much longer?
What additional miles? The hypothetical was twice the speed, not twice the distance. If you double the distance and keep the same number of crashes, then of course planes are twice as safe. Why would you suggest otherwise?
The risk would surely increase with doubled speed, no? The Concorde was unsafe because of all the design constraints imposed by being able to go that fast.
Why do you think that the concode was unsafe? It had a single crash in over 25-years of operation, and that crash was due to runway FOD. That makes it probably the single safest passenger aircraft ever brought to service. In fact, it is probably one of the single safest forms of transportation ever.
The concord was involved in single accident resulting in death or injury, which was not due to a fault of its own.
> That makes it probably the single safest passenger aircraft ever brought to service. In fact, it is probably one of the single safest forms of transportation ever.
It had so few flights, yet one of them was a catastrophic loss that killed all on board. Just compare it with, e.g., the Boeing 747-400: the Concorde had one full loss in 90,000 flights. The 747-400 has had 0.5 full loss equivalents in 8,420,000 flights. The stats for many other planes are similar.
The Concorde really was quite dangerous. And you can't blame it as "no fault of its own"; it suffered tire blowouts at a rate 30X of other planes, which had previously caused punctured fuel tanks. A tire blowout caused the fatal crash.
Well put. I've always felt an intuitive unease about that "safest per unit traveled" metric but wasn't able to articulate the problem as clearly as you did.
The better measure is time: deaths per hour traveled. And by that measure buses are actually safest, followed by rail, then air [1][2]. Cars are still 4x more dangerous, but that's much less stark than the common wisdom. And even that is misleading because with cars specifically, there is much more variability of drivers, car quality, maintenance, etc. If you are sober, responsible, educated, with a well maintained car, etc. your odds are significantly lower. Think drunk drivers who only kill themselves in accidents (this is a significant portion of car deaths).
[1] https://en.wikipedia.org/wiki/Aviation_safety
[2] http://www.ijhssnet.com/journals/Vol_5_No_7_1_July_2015/15.p...