I tried to find some kind of data on this and the estimate is that coal power kills ~13,200 people each year in the US as of 2010. That's about a third of the number of people killed each year in car accidents and a sixth or so of Diabetes deaths.
But a well-engineered coal plant perfectly run in the land of Theory (next to the well-run and well-engineered nuclear plants) sequesters its CO2 and its coal is mined by robots, right?
If nuclear is so safe, why is it impossible to get private insurance for nuclear power plants?
http://en.wikipedia.org/wiki/Economics_of_new_nuclear_power_... (nuclear plants only have a tiny proportion of their liabilities covered by private insurance, they require state guarantees or no-one would ever build one).
No-one knows how bad the downside may be so it's impossible to gauge the risk. If a coal power plant is hit by a tornado, an earthquake, a tsunami, and a wildfire all on the same day the maximum downside is no more coal power plant.
Nuclear plants kill a lot of people from direct/indirect effects too. Uranium doesn't grow on trees and jump into reactors by itself. It's funny how the pro-nuclear crowd likes to wave away (a) legacy plants, (b) geopolitical issues (e.g. uranium mined unsafely in third world countries), (c) proliferation, (d) spent fuel storage, (e) transportation, and so on and so forth, and then tally up lung cancer, people run over by dump trucks at mines, and so on for coal.
"Yes, the damage done by the coal plant is more diffuse. But it is not less than the damage done by Fukushima to the ocean and to eastern japan."
I simply don't know how much damage Fukushima has done and will do in the future. Do you? Wait until we find out what happens to the spent fuel at Fukushima and whether it successfully rolls the dice with aftershocks.
I'll share your assessment of the risks of plant damage -- in the case of coal, it's normal operations which are a risk, in the case of nuclear, it's abnormal operations which pose the greatest risk. We're starting to get sufficient baseline data to have a sense of nuclear's actual risks (though some, such as long-term waste storage still haven't been resolved).
The biggest risks, as I see it, with nuclear are actually organizational. Tepco, Hanford, and Chernobyl are all cases in point: managerial incompetence or malfeasance have resulted in hugely damaging events or consequences. A large part of this managerial threat can be attributed to the very high concentration of value that a nuclear plant poses. At roughly 18 GWh annual production rom a typical plant, that's $1-2 billion in annual retail electrical revenues -- a pretty substantial turnover.
I'm not disagreeing with you about the lala land coal plant btw, just comparing apples to apples. I assume the well run perfectly engineered nuclear reactor might not be as cheap or efficient as the badly run stupidly engineered reactor.