Comment by wdewind | Hacker News Reader

Comment by wdewind | Hacker News Reader

Parent Full thread

It's worth noting that, afaik, none of the reactors that have failed have been built using the "newest generation" of reactors, which are supposed to be much safer. This is a double edged sword, of course, because that also means they haven't been tested nearly as long. But I don't think the math to scaling nuclear power is as simple as we've fucked up 4/500 times, just extrapolate the numerator and denominator and you've got it. As we get more experienced with the technology we'll get better. And any conversation about the safety of nuclear needs to contain the relative risk of not using nuclear.

> More generally, the problem I see is that which was the real core message of the 1972 book Limits to Growth. There are limts to growth, and at exponential scales, those are reached rapidly even with increasing resources or pollution-absorption capacities by powers of 2, or 10, or 100. Those limits need to be embraced.

Embracing those limits requires, as far as I can tell, eugenic-style control over breeding. Personally I'd rather the human race burns itself out than go down that path.

I'm a fan of empirical over theoretical data and arguments. The 4:500 rate is what we've achieved, and, belabouring the point, all accidents have had massive sociological elements. I very strongly recommend reading Charles Perrow (Normal Accidents, but also other works).

Yes, something's got to control human breeding. Either humans or something other than human. Most likely viral or bacteriological, in my book.

I generally agree, but in the case of the environment we're not going to have clear data on what to do. How are you comparing the risk of building nuclear to the risk of not building nuclear? It's a very difficult thing to do.

That gets into some pretty profound discussions of just what "risk" is.

The current benchmark (e.g., used in the IPCC's own assessments of carbon-neutral energy options) is of lives per GW generation. I find this naive on numerous counts, e.g., my citation of the Banqiao disaster -- horrible if you were caught up in it (though also note: the bulk of the deaths werev from disease and starvation in the aftermath, the acute toll from the inundation was ~40-50k).

Virtually all deaths due to conventional processes -- physical, heat, trauma, even chemical -- are fairly ameliorable by direct precautions. Even in the case of a massive dam failure -- Banqiao, Johnstown, Vajont -- advance warning of a few minutes would have reduced fatalities tremendously, and with hours or days, all but entirely. Each of these disasters was precipitated by clear warning signs, though they were unheeded, for various reasons. In the Johnstown and Vajont cases, state of the art in engineering and geology was a major component. Perverse incentives to plant operators existed in all cases.

Looking at instances of explosions also suggests that warning, evacuation, and shelter could save virtually all lives, assuming a proper understanding of circumstances. Halifax, Canada, Galveston, TX, West, TX, Martinez, CA, Tianjin, China, and Lac Megantic, Canada all represented substantial loss of life which could have been avoided through both better advanced risk management and notice in advance of the precipitating disasters. Contrast the Pepcon disaster, among the largest non-nuclear blasts ever. Immediate deaths, though, were only two (another 370 or so injured), in large part due to the facility's isolation from any surrounding structures.

By contrast, long-term chronic exposures can be quite debilitating. Minimata, Japan, Love Canal, Hinkley, CA ("Erin Brokovich"), Bhopol, India, Agent Orange throughout Vietnam and SE Asia, and industrial contamination throughout China, will likely be with us for a long, long time. Flint, MI, can be added to that list.

Chemicals can generally be mitigated through binding and sequestration operations, nucleotides are more difficult. Long-term low-level chronic emissions, or perhaps worse: punctuated equilibria, concern me.

The bigger concern I have though is for systemic failure. Nuclear energy provisioning, particularly based on very sparse fuel sources (uranium from seawater, thorium from rare earths mining) creates a long-term dependence on a large and potentially fragile technological stack. By contrast, wind, solar, and hydroelectric power are rather simpler, and even geothermal is fairly straightforward.

"Complexity is the enemy of reliability" is a phrase first occurring in The Economist newspaper, January 18, 1958. It's been repeated, usually as "complexity is the enemy", many times since, though I suspect those citing it don't know the origins (my first encounter was through Eric S. Raymond's writings).

Creating a civilisation-supporting technological stack that isn't complex seems a strong goal. And quantifying systemic global risk part of the problem.

http://books.google.com/books?id=aDsiAQAAMAAJ&q=%22complexit...

https://www.reddit.com/r/dredmorbius/comments/24cxgc/program...

Thanks for a very well thought out reply.

> By contrast, wind, solar, and hydroelectric power are rather simpler, and even geothermal is fairly straightforward.

My understanding is scaling these technologies is where the complexity may lie, and that it would so extremely difficult to support the growing energy needs of the world's population using these technologies that they aren't really even an option. I'd like to be wrong. Am I?

I don't know.

This gets into a whole mess of considerations, and there's not a lot of agreement.

There's little doubt that there isn't a gross surplus of solar energy incident on Earth's surface. But that surplus is, by orders of magnitude, fairly thin relative to present human energy demands. By the time numerous factors are applied, you end up with infrastructure requirements which are truly prodigious.

I'm increasingly coming to the conclusion that the real lesson of Limits to Growth has largely been missed -- not what limits are, but that they exist, and even highly generous assumptions of abundance (or effluent absorption capabilities) are rapidly overwhelmed with even modest rates of growth. That is, the lesson is humans must embrace limits. There's considerable more math on how far in advance clear warning signs do or don't exist, though fundamentally the principles are simple.

This turns the typical statement of sustainability on its head. The question isn't "how do we provide energy, food, and other resources for the people we see coming", but "how many people can be supported, at what level of affluence, for what duration?" It's not energy supplies that are insufficient, it's humans which are overabundant.

Answers to "how many" vary, though I've seen published values ranging from 50 million to 50 billion -- three orders of magnitude. Someone's off by a lot.

Going by history, the range of 50 million - 2 billion corresponds to human population from early prehistory to about 1920. At the dawn of the Industrial Revolution, the value stood at about 500 million.

Levels at or above present populations would require a large number of complex problems be solved for a very long time.

The answer to your question may well be not that you're wrong, but that you're asking the wrong questions.

Put another way: there's nothing inherent to the Universe (or even this small speck of it) which gives humans any right to existence or level of affluence. Predicating reality on what we'd prefer is the definition of wishful thinking.

> none of the reactors that have failed have been built using the "newest generation" of reactors, which are supposed to be much safer.

At some point there has to be a compromise made between cost and safety. You could spend an extra $10 billion making a plant even more safe, but in doing so make the scheme unaffordable, and 20 years late. And even then the station could still have complex failure modes that are very difficult to model. Maybe that is exactly what happened to the nuclear industry since Chernobyl? We ended up with ever more complex and monolithic systems that require thousands of expensive physicists to do basic building and maintenance. Perhaps it would have been better to invest in small modular systems.

Reply on news.ycombinator.com