The problem with those decisions is that you can't get out of them. Killing them isn't your will. It's not the same thing.

Suppose you make cars. You know that 5% of your customers will get into a serious accident, and of that 5%, if you add $1000 to the cost of each car, you can save 1% of them. So $1000/car to save one in 2000 people; two million dollars per person. If you don't do it you know that you sell 60 million cars and so 30,000 people will die. Do you do it? What if it costs $10,000/car? $50,000/car?

At some point you have to say no or a car would cost thirty trillion dollars. And if some company decides that a human life is worth 1.5 million dollars but you later decide that it should be worth 2.1 million dollars, is that really the level of mistake that deserves execution?

All you really have to do is let them know ahead of time how much money it will actually cost them in jury awards, and let the victims sue the coal companies. If you set the value right they'll either figure out how to kill fewer people, or shut down operations.

The analytic scenarios like that have never been the problem because if the company is already doing the math and you don't like the outcome then all you have to do is change the numbers. The problem has always been what to do about businesses that kill people without expecting to.

There is a difference between giving someone poison for no reason, or purely out of malice, vs. doing it because there is some countervailing advantage that justifies the risk. And sometimes the advantage is just that it costs less than an alternative.

People know that an old Ford Mustang isn't as safe as a new Mercedes S-Class, but it's also less expensive.

Thus once again illustrating the advantages of regulation for sellers, not just buyers.