I do not buy this paper's conclusion. First problem I see is that they use Road Fatalities as proxy metric for overall goodness. If someone's driving drunk, a

I do not buy this paper's conclusion. First problem I see is that they use Road Fatalities as proxy metric for overall goodness. If someone's driving drunk, and they run their truck into your living room at 2am - yeah maybe no one died, but a lot of damage and stress was caused. Not to mention just destruction in the form of minor accidents. I simply cannot buy the conclusion - "Yeah it's a good thing we have a few drunk drivers". Call me anti science if you must