[andrewla]

> Even with some modest ongoing investments, we'd have designs that are FAR safer than systems currently in operation.

Why would you think that? The problem with nuclear is that we are playing with significant tail risks, which are the hardest to engineer around. Catastrophic failures can have tremendous consequences, and in a sufficiently complex design there will be possibilities of catastrophic failures. While there are dead-simple reactor designs (like radiothermal), the vast majority of approaches work by layering complexity.

After Fukushima I became much more lukewarm on the subject of nuclear power, not so much because it was a huge Chernobyl-like disaster, but because I had repeatedly been told that a modern (post-Chernobyl) reactor was simply incapable of these failures modes. I had even, ignorantly, parroted back almost those exact words when discussing nuclear with people.

[overlordalex]

> I had repeatedly been told that a modern (post-Chernobyl) reactor was simply incapable of these failures modes

Fukushima was commissioned in 1971. Chernobyl happened in 1986. I would not call the design a "post-Chernobyl" reactor.

I can recommend the section on the wiki page for further reading: https://en.wikipedia.org/wiki/Fukushima_Daiichi_Nuclear_Powe...

> I had even, ignorantly, parroted back

I think this is probably the root cause ;)

[andrewla]

You are correct. My sense of history is very flawed. Even the construction of Chernobyl post-dates Fukushima, and in fact most still operating reactors.

Still, Fukushima has forced me to adjust my priors in estimating what the level or risk of a nuclear disaster is; before that I was a strong supporter of nuclear because I thought that the risk was essentially zero.