Hacker News Reader

Comment by fighterpilot | Hacker News Reader

> This is basically the the same logical errer again

It's not necessarily a logical error. Extreme value theory should be the default tool for understanding tail risks (e.g. earthquake magnitude), unless we have good theoretical/logical reasons to depart from an understanding that's mostly informed by historical data.

Nuclear weapons are where I agree with your perspective. The tail risk is significantly larger than what is present in historical data, and attempts to understand this risk by looking at historical data are flawed.

Nuclear energy, like earthquake magnitude, is where I disagree with your perspective, unless you can present a coherent rationale about why nuclear power tail risk is drastically underestimated by the historical data, especially after taking into consideration modern plant designs which are safer than Chernobyl + the possibility of building them in the desert with nobody nearby.

View on HN

Hey first of all thanks for the constructive replies, much appreciated!

"Attempts to understand this risk by looking at historical data are flawed" is a good summary of my main point, you got it. Basically I think we can't ignore the possibility of large incidents just because they haven't happened yet.

We can extrapolate from known small incidents, some of which are listed on wikipedia:

https://en.wikipedia.org/wiki/List_of_civilian_radiation_acc...

Some standouts:

https://en.wikipedia.org/wiki/Goi%C3%A2nia_accident

https://en.wikipedia.org/wiki/1996_San_Juan_de_Dios_radiothe...

Other incidents mentioned without separate stories include radioactive waste finding its way into road pavement, apartment building materials (hundreds of contaminated flats, nobody found out for years).

Incidents such as these have killed many people over several decades in many countries, and that's just through sheer negligence. What if someone adds malice to the mix?

That's actually my main concern about nuclear power, not the safety of the reactors.

But since you mentioned building them in a desert in order to be safe I feel the need to mention my perspective there too. First of all, Chernobyl had 4 reactors and only one exploded, thanks to heroic effort. (Meaning it could at the very least have been 4x worse than it was).

The event was actually discovered in the west by sensors in Sweden, which is very far away. Fallout from Chernobyl made certain grazing animals surpass legal radiation levels for over 30 years (and counting) in areas as remote as Lapland and Bavaria.

To put the distances in perspective if you are American, it means that a similar event as Chernobyl taking place in Florida would make pigs from Texas and Caribou from Ontario be unsuitable for human consumption, for many decades.

So distance does not really help. In order to be safe by putting power plants in a desert, the USA would basically need to put it's reactors in the Sahara.

The safest place for a reactor would be inside a very stable and peaceful country with a very high level of mental health in the entire population. Maybe Switzerland?

I'm not saying we shouldn't even consider it under any circumstance, but if there's any other option (which I believe) then that option is probably preferable.

Let's take a step back. What about molten salt reactors? I feel like it is a mistake to think future nuclear reactors have the same safety profile as old ones when thinking about tail risks.

Better reactor designs that can’t basically self-combust through accidents or operational errors are an improvement of course, but I’m not aware of any design that eliminates the need to produce relatively large quantities of waste. That in itself poses a large risk.

Ok but what is the tail risk here. Can you outline a series of causal steps where a molten salt reactor will cause more than 100,000 to 500,000 deaths? If you're just saying that perhaps maybe there will be 2,000 deaths, well that's still an incredible safety profile compared to the number of deaths caused by other energy sources.

Sure - Creating the fuel for molten salt reactors requires a series of steps where highly radioactive materials will have to be handled, stored, transported and controlled by humans.

As shown with the examples given earlier (which are just a few of many), such situations will inevitably provide opportunity for the materials to be lost or stolen. (IAEA has documented 18 cases of stolen radioactive elements from various sites that handle radioactive materials).

One of the elements needed for molten salt reactor fuels is plutonium. 500 grams of plutonium is estimated to be enough to kill up to 2 million people by conservative estimates.

Presumably you will need more than 500 grams regularly to run a salt reactor, so any number of malicious methods can kill 100 000 to 500 000 people, if someone gets their hands on even small amounts of plutonium.

I think that when people reach the conclusion that the benefits of nuclear outweigh the downsides it is usually because they haven't really considered the real downsides fully. The benefits are huge, to be sure, and tempting.

Now that being said - if a reactor design appears that would significantly /reduce/ the amount of dangerous materials that have already been created, and not require any more of it to be created, I think it would be a good idea to build such a reactor to get rid of the waste we already have.

Speaking of molten salt, I find it more interesting as a concept for energy storage. Solar power can superheat salt and use it to genereate energy even at night. A solar power plant in Spain has generated electricity continously (day and night) for over a month using this technology.

https://cleantechnica.com/2013/10/14/worlds-largest-solar-th...

I believe that we have an abundance of very promising - sometimes already proven - technologies that will let us generate stable electricity from intermittent, clean sources with comparatively insignificant risk to human lives.

Reply on news.ycombinator.com