4 ms·
Strong agree with this, although worth noting that a crude back-of-the-envelope calculation suggests that the 737 Max has a rate of fatal accidents per mile tha
by jessriedel 3y ago
Strong agree with this, although worth noting that a crude back-of-the-envelope calculation suggests that the 737 Max has a rate of fatal accidents per mile that is only safer than driving a car by a factor of ~10. (See below.) Since cars are one of the most dangerous forms of transportation, it's certainly not crazy that people would want to avoid an airplane model that's only 10x safer than a car. Bicycles are only ~3x as safe as cars, and people reasonably use the danger of biking as a reason to not commute that way.
(In comparison, air travel is usually safer than cars by a factor of 100 or more. It's hard to even estimate how safe it is because there are so few commercial airliner crashes.)
There's also an inherent difficulty in trying to attach a risk to a single model of airplane, rather than broad activities like driving cars or bicycles, due to lack of data and data inhomogeneity. It seems reasonable to me to put some Bayesian probability on the 737 Max being significantly more dangerous than it naively appears (say, by another factor of 3 or 5) due to Poissonian noise and factors we can't observe.
---
> 1,300 aircraft have been built since the [737 Max] first started flying in 2017, with two deadly crashes. I don't know how many miles those have accumulated, but presumably it's of order 4k miles per day per aircraft, and maybe [1.5] years ([500] days) of flying to date per aircraft on average [in light of delivery dates and extensive groundings], giving a very rough estimate of a [~2] billion miles? So maybe a deadly crash per billion miles, in comparison to a bit over one deadly crash per 100M miles for cars.
https://news.ycombinator.com/item?id=38895004 https://news.ycombinator.com/item?id=38895004
https://news.ycombinator.com/item?id=38895077 https://news.ycombinator.com/item?id=38895077
- gruez 3y ago> 1,300 aircraft have been built since the [737 Max] first started flying in 2017, with two deadly crashes How valid is this calculation for today? Presumably they fixed the issues with the two deadly crashes, so those arguably shouldn't be included in future forecasts.
- jessriedel 3y agoIf you think the problem with the 737 Max has been fixed, none of this is relevant. The crude estimate I gave is if you adopt the reasonable theory that the production/operation of the 737 Max is bad more generally, with the cause of the prior crashes being just one of the flaws discovered. An illustrative example: suppose there was a plane model with 100 flaws that each cause a crash every 100B miles, and after each crash that particular flaw is fixed completely. The "correct" fatality rate would still be ~once per billion miles even after several crashes, and would only approach the standard airplane background rate asymptotically after dozens of crashes.
- StreetChief 3y agoDo these calculations assume all miles flown by ANY max variant are the same?
- jessriedel 3y agoSort of. It averages over them. If you want to dig up some data and make a more detailed estimate, please do so.
- StreetChief 3y agoNope I was hoping to point out what is to me a failure of logic, that since previous versions of the plane flew safely, newly manufactured max 8 and 9s somehow have trillions of miles of safe flight. Should we combine all ford mustang car miles driven since the 80s to calculate the safety of the newest version?! Obviously the analogy doesn't hold completely, but the versions of the 737 are different. If I put a jet engine on a ford mustang, is it still a ford mustang? It's the same car, but I mounted a bigger engine!
- jessriedel 3y agoOh mean, should we look at the 2021 model year Ford Mustang to estimate the risk of 2024 model? What about different 2024 cars that were manufactured in winter vs summer? I'd say yes, but you're welcome to do what you want.
- StreetChief 3y agoi'm shocked you would say yes to that, but consider all the 737 miles the same