Hacker News Reader

Accidents per driver is the only metric that matters, because the one tells you how risky a certain class of drivers is. Accidents per passenger mile is not even meaningful, because you do not decrease the risk by by making more miles.

View on HN

Ah, except it does not tell you that. Consider this extreme example: if Honda drivers on average drive 200 miles a week and 10% of them get into an accident each year, while Lamborghini drivers on average only take their cars out for a single 10-mile spin once a year but also 10% of them get into an accident each year, it would a poor conclusion to say that these are equivalently risky driver classes.

> Accidents per passenger mile is not even meaningful, because you do not decrease the risk by by making more miles.

Er, right, it is meaningful because you obviously increase the risk by driving more miles. Very easy to own a car and have 0 accident risk: never drive it.

Reply on news.ycombinator.com