Hacker News Reader

Comment by andrewla | Hacker News Reader

> incredibly safe in comparison

I assume you mean this empirically, not based on the actual risk profile.

The cost, in lives, of a major nuclear containment failure is easily in the millions to tens of millions. The cost, in lives, of a major coal disaster is at worst maybe in the hundreds, even in principle.

So it comes down to a belief in whether we have properly contained the risk. My priors on this shifted tremendously after Fukushima.

View on HN

> The cost, in lives, of a major nuclear containment failure is easily in the millions to tens of millions.

> My priors on this shifted tremendously after Fukushima.

I'm curious why, despite Fukushima being pretty close to the worst-case scenario with modern technology, you feel these are mutually compatible feelings?

> despite Fukushima being pretty close to the worst-case scenario with modern technology

I mean, we all know what the worst-case scenarios actually look like because we have deliberately triggered containment failures for military purposes. Are you comfortable with the idea that nuclear reactors may melt down, but if we're lucky they will have minimal impact? Before Fukushima I feel like the message was unambiguously that meltdowns in modern technology are impossible. Now I do not trust that message.

I think there's a difference between the theoretical worst-case scenario and the realistic worst-case scenario. A one-two punch of the largest recorded earthquake on record followed by a tsunami feels pretty close to realistic worst-case scenario. We can always fret and wring our hands about plausible issues, but we also need to recognize the cost of inaction as well. The risk of catastrophic meltdowns costing millions of lives is so small it's hardly worth factoring into a cost-benefit equation at all.

> The cost, in lives, of a major nuclear containment failure is easily in the millions to tens of millions

The worst of the worst, Chernobyl, where there was a design issue, the design skimped on many safety features, and everyone involved was incompetent or ignorant, result in a few tens of direct deaths and few tens of thousands of premature deaths. Where are you getting millions from?

Specifically regarding Fukushima, less people died from the nuclear disaster caused by the tsunami than an oil tank fire caused by the same tsunami.

> Where are you getting millions from?

We have deliberately initiated containment failures for military purposes -- we know what this looks like.

After Chernobyl we were told that it was a design failure compounded by human failure, etc., etc., and that in a modern reactor, meltdown was simply not possible.

Then Fukushima melted down.

So my priors right now are that we can not place a good bound on the total odds of a meltdown. How confident are you that with a modern reactor a meltdown is impossible? How confident are you that in a meltdown scenario the impact will be small?

Can you provide a specific source for the "millions to tens of millions" figure?

> The cost, in lives, of a major coal disaster is at worst maybe in the hundreds, even in principle.

Sure, if you completely ignore what's coming out of coal plants' smokestacks day in and day out. Every day that coal plants exist and operate is a major coal disaster.

And the water! Storing ash in ponds and in the ground seemed like a great idea; and it's still not cleaned up.

The cost could be that much - but we've had 2 bad incidents, both considered "worst case", without anything near that situation. That is only 2 data points, and things could be worse - but that is the prior we have.

How to weight a .00(0000)1% chance at tens of millions of affected, versus the guarantee of millions of life-years made worse by the actual coal plants? That is a philosophical plan, but the likelihood * badness is one approach. We also appear to have good control over the worst case situation with a plant - it looks to me like a bounded problem space with modern nukes.

Do you have any evidence for these claims?

The claim for the millions to tens of millions number? I mean, we have deliberately initiated containment failures for military purposes, so we know what they can look like.

Do they look like Fukushima or Chernobyl?

To redirect back to my original point -- this is an empirical argument.

Russian Roulette is not safe because nobody got shot in the first two rounds. The turkey is not safe because he was well treated through October.

Containment failures of nuclear reactors have a very large upper bound of damage.

> I mean, we have deliberately initiated containment failures for military purposes, so we know what they can look like.

This is an empirical argument.

Is the upper bound of damage higher for nuclear reactors or fossil fuels?

Man, you've got more patience than I do for someone dancing around the question. Especially when the dancer thinks that the worst case for a modern nuclear plant is Hiroshima circa 1945, but won't actually come right out and say it (for reasons I suspect are obvious).

They will post a source for the millions dead figure soon, I'm sure of it.

Reply on news.ycombinator.com