This strikes me a little like the rational economic model of criminals, i.e. they weigh the cost of getting caught * likelihood against benefit of the crime. I

This strikes me a little like the rational economic model of criminals, i.e. they weigh the cost of getting caught * likelihood against benefit of the crime. In reality, having data people are biased towards action. But I agree, a holistic interpretation - including the population level "bad decisions" - is the ultimately correct approach. The caution I would urge is not to allow "no true scotsman" arguments, ie "well it's a good thing as long as people are careful". Many people aren't going to be, which is part of the calculus.