[Pyramus]

That's the issue with "safe" - it's difficult to capture in a single factor like number of accidents and for this factor, the distribution of the metric is very different.

Here's a different metric that entails both: Market price to ensure a nuclear power plant against damages. Why are they so high as to make nuclear power creation uneconomical? A market inefficiency?

It's very hard to explain away. So my point is: It's not black and white unfortunately, and it's both a technological and societal effort to make nuclear power safe.

[SECProto]

> Market price to ensure a nuclear power plant against damages. Why are they so high as to make nuclear power creation uneconomical? A market inefficiency?

Yes. The nonrenewable energies (coal, natgas, oil) in my source are constantly emitting carbon dioxide, an unpriced negative externality. This cost is entirely socialized - insurers don't pay for it, producers and users don't pay for it - it's largely being shifted to future residents of planet earth, with a small fraction of the cost starting to be felt now, and paid by government (emergency relief). Has both high cost and high (100%) risk. Insurers don't take on any of this. Compare that with nuclear, where the negative impact is risk of meltdown - high cost but low risk - and insurers are the ones taking it on.

[Pyramus]

Very unlikely. In this case 'the market' is not WallStreetBets, it's experts around the world who do nothing else but assess and estimate the risks.

There are two explanations I can think of:

> Compare that with nuclear, where the negative impact is risk of meltdown - high cost but low risk - and insurers are the ones taking it on.

This may be one reason. The risk of a 'meltdown' is actually so high cost but very low risk, that the corresponding distribution has no first moment. The 'expected cost' is infinite. I've heard folks use it as a real-world example of a Cauchy-type distribution, for more info see [1].

Another explanation might be that there is some sort of recency bias. But that would mean the cost was estimated correctly for the past - which is not what you would want to hear either.

You make a good point that externalities for fossil fuels are not priced in either, but that doesn't explain the phenomenon here. And I'm not arguing for or against the use of nuclear power - societies have already agreed to socialise the cost. I'm pointing out it's not as black and white ("nuclear power is and has always been safe") as some HN commenters may think.

[1] https://en.wikipedia.org/wiki/Cauchy_distribution

[adgjlsfhk1]

no it's not. coal gives millions of people cancer, but it does so in ways that are hard to prove in court on an individual level.

[Pyramus]

That is correct - if we were to price in "cancer" for coal maybe indeed coal may not be economic anymore.

But that's unrelated to the insurance cost of nuclear power. Either 'the market' (i.e. the people most knowledgeable to pricing the risk) are off or HN commentators don't see the whole picture?

I'm just pointing out nuclear power is no blanket solution to climate change (it may well be part of the solution), and that there are non-negligible risks that we have to consider as societies (and different societies already are dealing with the risk differently). For some reason HN tends to be a bit defensive when that is pointed out.