The right metric is traffic deaths per capita, because that means so-and-so many families are have lost a relative in a car crash. If you use traffic deaths per mile driven you begin to excuse excessive commutes when what you want to do lower fatalities.

If you spend more time driving, of course you're risking more.

If no-one left their house, there would be less casualties for many outdoor activities.

You have to account for per-mile.

Reducing the risk per mile driven would reduce deaths, but so would reducing miles driven. More public transportation would achieve that.

Higher-density housing wouldn't be the worst idea, especially higher-density housing near light railway when combined with public parks. The suburban paradises are plainly incompatible with public transport.

By that same logic we could avoid surgery deaths by not doing surgery.

But then you're ignoring the benefit.

The correct analogy: avoid surgery with preventative healthcare and lifestyle changes.

How do you call this paradox where mortality increases at a hospital because simple cases are diverted to outpatient clinic and only the severe cases are seen at the inpatient facility?

Adverse selection.

By that reasoning, traffic safety is increased by moving the bar from downtown to 30 minutes down the road when in fact it should be at the corner, so people can walk and leave the car at home. Good to have the misconception cleared up.

Sure, but no-one was arguing that.

Let's all start a fund to build bars closer?

Congrats on that break-through.

I disagree. Per Capita is the right metric. Per mile doesn't equalize everyone because we have to drive so much than everyone else to get around. When talking about social costs, per Capita makes the most sense.

You're comparing numbers in different units.

USA is in deaths per 100 million miles. The rest are in deaths per billion km. USA is 9.3 deaths per billion km if you do the conversion.