Hacker News new | past | comments | ask | show | jobs | submit login

This is framed misleadingly as a risk calculation, but the math is unfairly generous because planes go a long way and carry a lot of people. It is not a valid calculation for "how likely am I to die if I get on a plane".

Imagine if each plane carried a million people, and traveled a million miles in a single flight, and 1 plane in a dozen vaporized all its passengers. By the logic of the post, you would crunch the same safety number - "passenger miles between accidents" - but we would not call such a mode of transport safe because you have a 1-in-a-dozen chance of death.

If you are going to sum up passenger miles, rather than aircraft miles, you need to compare that against passenger-accidents, rather than aircraft-accidents. Each aircraft crash kills about 100 people at a time, so that "2 light years" figure is two orders of magnitude too generous, right from the start.

There are other arguments to be made about "passenger-miles" as an entire concept (air miles are not always fungible with other modes of transport, and a plane which explodes on takeoff 1% of the time is not safer if you fly it farther each flight) but I'll stop here.




Just look at likelihood you are going to die each mile traveled. Using that method, passenger planes are 750 times safer than cars.


I was taking issue specifically with the calculation in the article, not making an unrelated comparative analysis of planes vs cars.

That being said, such stats as yours do not tell the whole story. The likelihood of dying while driving across the Atlantic Ocean approaches 100%...


Oh, and fun fact: an Apollo moon mission racked up nearly 3 million passenger miles per flight and did not suffer a single fatality. Even if the astronauts on Apollo 13 had not survived, and the whole program cancelled right then and there, by fatalities-per-passenger-mile a Saturn V to the moon would still be far "safer" than driving, which averages one death every quarter of a million miles.

I think this demonstrates two important flaws with the "passenger miles" concept: 1, miles are not always fungible between modes of transport. 2, intuitively we care more about the risk per trip, rather than the risk per mile.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

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

Search: