[nathantotten]

From Nate Silver: “This article is full of terrible statistical logic. Yes, it's hard to exactly quantify the risk of a highly deadly nuclear accident. By nature, tail events are rare. There is intrinsically some guesswork. But empirically the chances are very low.

The author also blatantly cheats by counting Fukushima as 3 separate accidents, as though they occurred independently from one another and didn't have a common cause like oh I dunno a magnitude 9 earthquake.“

https://twitter.com/natesilver538/status/1558530091860336640...

[mchusma]

This article is so flawed in so many ways. The most obvious issue is failing to show the actual safety statistics! I mean, at least start with the obvious starting point: that nuclear energy's record is that it is one of the safest forms of energy. https://www.statista.com/statistics/494425/death-rate-worldw...

There could be a disaster tomorrow 1000x worse than all previous nuclear energy disasters combined and it would still have been net-safer than coal (and this is excluding climate change effects, if you choose to include those.

I mean, if you want to then say "we got really lucky over the last 60+ years" I guess you can do that. In 1970, sure there were a lot more unknown risks. That was 52 years ago.

[arcticbull]

It's funny that hydro gets left off these charts, and when it's put on, it's super contentious. The worst hydro accident in history, the Bangqiao Dam failure killed ~200,000 people in one fell swoop. Wiped settlements off the face of the earth. [1]

Brown coal kills 100 people per TWh generated, coal on average about 25. [2]

Chernobyl killed 4000 (31 immediately, the rest were computed over the full course of time including forward looking estimates and counting the people who committed suicide because they feared they were 'contaminated'), Fukushima killed 0, Three Mile Island killed 0.

The US generates about 960TWh from coal per year, or 24,000 deaths. The US' coal consumption alone is equivalent to 6 Chernobyl's per year.

[1] https://en.wikipedia.org/wiki/1975_Banqiao_Dam_failure

[2] https://ourworldindata.org/safest-sources-of-energy

[bombcar]

People are way more afraid of "secret silent invisible" killers than they are of "massive wall of water that wipes out everything in its path."

We're not wired correctly when dealing with rare occurrences.

[AstralStorm]

It so happens that hydrocarbon and coal power plant exhaust is a relatively invisible silent killer.

What people are afraid of is what they're told to be afraid of.

Maybe if we linked oil to fuel bombs or other explosives we'd get more headway.

[BrandoElFollito]

In general, we are not wired correctly to deal with concepts outside of our immediate neighborhood.

You can say with confidence that there are 3 people in a room. Or about 20. Or roughly 100-200-300. Or a lot. A stadium filled with 10,000 people is not different from one filled with 100,000 people - for someone who sees them for the first time.

A million times bigger does not mean anything. What is a hair x 1M? No idea.

I am an ex-physicist and I leaned to just look at the numbers and compare them when needed. 10^-18 is fine for something because I learned that but I cannot imagine it. Same for 10^23.

This is also the reason why homeopathy does not sound bad to people when it is written 100 CH on the bottle. 100 looks good. It is a 10^-100 dilution ratio.

[bee_rider]

Unless I'm misreading the chart, it seems to compare quite well to coal and oil, and in line with natural gas. Why would that be contentious?

The only quibble I'd have is that dams serve multiple purposes -- they could help prevent flooding and help out with agriculture. So in some sense a dam could be helping to save lives (apart from the obvious benefits of... having electricity). This seems like a unique perk.

[arcticbull]

The same is true of any bridge past a cliff edge.

[bee_rider]

Err, I'm not sure I follow sorry, is this an expression?

Edit: Oh -- I mean unique among power sources, not unique in general. Clearly other things exist to save people.

[duped]

One should be skeptical of the Chernobyl numbers. However even if it was 100x worse it's still a net win for humanity to invest in nukes.

[natmaka]

About Banqiao: https://news.ycombinator.com/item?id=24301436

[arcticbull]

I agree but one could say the exact same thing about the Soviet Union running tests at 2 in the morning at Chernobyl. So if you're willing to discount one it's only fair to discount the other no?

[natmaka]

The Banqiao dam suffered from major flaws then from decades of a fair amount of diverse very adverse conditions and absence of proper maintenance... and it kept up. Then it took a typhoon to finally destroy it. The grotesquely bad handling continued during the crisis which followed.

A few hours of improper use were sufficient to trigger a disaster at the Chernobyl's reactor, then the authorities' reaction (evacuation, liquidators...), albeit imperfect, was way better than at Banqiao.

[sofixa]

Chernobyl was of a flawed design with a very serious bug which was known (but classified), and it took a terrible very poorly coordinated drill to cause it to actually meltdown.

A more accurate comparison would be Fukushima, where the design was wrong (backup generators in the basement, in a flood prone zone) that survived a 9 on the Richter scale earthquake and was only damaged by the resulting tsunami (but only because the operator had ignored all the warnings about the placement and protection of backup power).

[arcticbull]

Give me a few hours and I guarantee you I can do just as much damage with a dam and I can with a nuclear reactor.

[ZeroGravitas]

That is a terrible link to use as evidence. You think discussion in 2022 about what the 'safest forms of energy' is should be based on a table from 2012 that doesn't include grid scale solar PV?

Lets look at some slightly more up to date numbers:

https://ourworldindata.org/safest-sources-of-energy

See, nuclear does fine. Basically drawing with the cheaper forms of energy this article is arguing for that are being rolled out in ever greater numbers around the world.

[arcticbull]

Also zero people died in the immediate aftermath of Fukushima, and one employee sadly eventually died years later of a cancer that's being attributed to it after receiving an annual dose of radiation in the incident. Counting it as anything other than an indication of how safe these plants have gotten is pretty disingenuous.

The data is clear, it's the safest form of energy in deaths per TWh generated. [1]

Anyways while we fitter around and argue, China is building 150 new reactors in the next 15 years, as much as the entire world has built in the last 35. To go with their massive solar deployment. Now that's an energy grid getting cleaned up neatly. [2]

[1] https://ourworldindata.org/grapher/death-rates-from-energy-p...

[2] https://www.bloomberg.com/news/features/2021-11-02/china-cli...

[akamaka]

It’s really ridiculous to say that nuclear is the safest form of energy, when deaths at Chernobyl and Fukushima were prevented only by enormous cleanup efforts that will probably easily exceed $100 billion each. That doesn’t include the lost economic value of the exclusion zone, higher energy prices, etc.

Meanwhile, there has been a worldwide effort to spend billions of dollars upgrading existing reactors with enhanced safety equipment after Fukushima, which suggests that many nuclear operators had not been accurately calculating risk factors up to that point.

The more accurate statement would be to say that humans are capable of overcoming the inherent danger of nuclear energy, with enough money, will, and sacrifice.

[boomboomsubban]

"It's ridiculous to say nuclear is the safest form of energy when it costs so much money" is a ridiculous argument.

The cost of nuclear may be a fair argument against it, and should definitely be considered when determing our energy policy. That doesn't mean it somehow kills more people.

Every form of energy has an inherent danger that humans can use money, will, and sacrifice to overcome.

[akamaka]

If we’ve spent more money evacuating populations from irradiated areas than on training wind turbine technicians to use harnesses properly, does that mean nuclear energy is safer than wind power?

[boomboomsubban]

Yes? As even after that training, fewer people have died due to nuclear energy. Cost and safety are two separate points, both need to be considered but you can't just say something is unsafe because it's costly.

Imagine someone says "you have to either pay $10 to play Russian roulette with a five chamber gun or $100 to play with a six chamber gun." Which is safer?

[himinlomax]

> It’s really ridiculous to say that nuclear is the safest form of energy, when deaths at Chernobyl and Fukushima were prevented only by enormous cleanup efforts

Do you not understand how superlatives work?

It's the safest form of energy. It's not entirely safe, alright. But pointing that out in the absolute is completely pointless anyway. We need/want energy, there's a price for it. In terms of safety, nuclear is the best. Superlative. It's better than all alternatives.

[Teknoman117]

It's almost nutty to me that the number of workers who've fallen to their deaths from wind turbines per GWh is more than the number of people killed by nuclear power accidents and maintenance per GWh.

[wizofaus]

Doesn't surprise me at all really. And it doesn't even necessarily mean it's irrational to prefer living near a wind turbine farm than a nuke plant, given most people aren't turbine engineers. I also imagine it's a statistic that will change considerably with time as we build out wind farms, especially if doing so can largely be automated. At any rate, we need both (given the current state of affairs). I doubt the choice to build one or other is going to come down to deaths-per-GWh.

[arcticbull]

The number will go up as more wind is deployed. It has to be maintained, serviced and replaced on a periodic basis.

But yes, we need both.

[cplusplusfellow]

> how safe these plants have gotten

Consider that Fukushima was designed in the 1960s and that virtually no reactors with designs that date more recently than that have been built, and we have an absolute quorum that this is the safest form of base load energy you can procure.

It’s saddening to me the green movement ruined our chances of clean energy and averted climate crisis in my lifetime.

[wizofaus]

I've never understood why there has been such a strong anti-nuke sentiment among many environmental groups, but you're really stretching it to say that's what's "ruined our chances". The blatant sponsorship of AGW denialism and buying out of politicians by vested interests deserves the bulk of the credit on that front. That and weak and backward looking political leadership in general.

[cplusplusfellow]

AGW advocacy would enjoy a significant improvement of reception if every solution other than me suffering a lower quality of life wasn’t neatly boxed out of our set of choices.

Meanwhile the politicians and celebrities who wish I would eat crickets (literally) fly on private jets to spend time on yachts that burn 500 gallons of diesel a day, and complain that my work truck doesn’t run on lithium and cobalt batteries powered by solar panel.

[ZeroGravitas]

I was expecting something much less throrough than the article turned out to be based on this comment.

It's mostly the official evidence of people covering up incovenient safety facts in the article that I would say qualify as "terrible".

[atwood22]

The empirical changes are not very low. Assuming each plant has a uniform probability of failing once every 10,000 years, that comes out to an accident once every ~50 years in one of the ~500 nuclear power plants on Earth.

[AstralStorm]

That is ignoring the scale of an incident.

We had a major radiological incident in a Polish coal plant recently. That not counting a whole power plant block failing, contaminating water and emitting tons of fumes.

It still has not been fixed as the replacement block turned out to not meet new emission standards.

[atwood22]

Three mile island, Chernobyl, and Fukushima were all extremely severe events. Also, the timing of these events does seem to support the estimated probability of a severe nuclear event somewhere in the world every 30-50 years.