That's one way to calculate it. Other times, people measure life in "quality-adjusted years". So a coal plant that kills a young person every 20 years is worse

That's one way to calculate it. Other times, people measure life in "quality-adjusted years". So a coal plant that kills a young person every 20 years is worse than one that kills 78-year-olds regularly. This is a more rigorous method, measuring quality*years instead of deaths, which is why you almost never see it in activist claims. It's too rigorous to be manipulated. Other times, people just look at the immediate cause of death (e.g. lung cancer) and extrapolate to the cause of that cause (e.g. air pollution). But this imports a lot of assumptions and is easy to game. For example, I've seen cases where hundreds of thousands of deaths have been attributed to a given cause, but it turns out that those people were dying just a few years short of average life expectancy, and would've died for other reasons very soon regardless - quite different from the image of people in their prime dying needlessly. It's good to question these figures. As they are currently presented, we don't know the methodology, which makes them meaningless.