Noah Smith is annoyed by nuclear power enthusiasts who he thinks should be enthusiastic about solar power instead.
And I think there is something to his analysis. In certain circles, being pro-nuclear is a way to be climate-aware and pro-science while also signaling a kind of masculine tough-mindedness and hippie-punching attitude. But the scientific and technical achievements of the solar industry over the past few decades have been staggering, and those of the battery industry have been only slightly less so.
That being said, I think the real problem here is both proponents and critics treating “pro-nuclear” as an identity rather than a policy agenda. It’s more useful to think in terms of concrete policy ideas. For example, if a state is considering a regulatory mandate about what kinds of electricity its utilities buy, that could be a renewables mandate (which wouldn’t include nuclear) or it could be a zero-carbon mandate (which would). On this, I am strongly pro-nuclear — don’t turn down carbon-free electricity. If the nuclear hype guys are wrong, nothing will come of it. If they’re right, we’ll be glad the regulations accommodated that. Similarly, it’s just insanity to prematurely retire existing nuclear power plants. The problem with these light water reactors is they are insanely expensive to build. Having built them, we should use them.
Last and most importantly, we don’t need to yell at each other on the internet about whether we’re on the verge of a technical breakthrough in long-term electricity storage or a technical breakthrough in molten salt reactors. What we need to do is recognize that the existing regulatory framework for nuclear reactors is extremely hostile to breakthroughs.
With batteries, mercifully, you can build it if you invent a better one. Nuclear, for good reason, is subject to stricter regulatory oversight. But that oversight currently spells out the safe way to build a nuclear reactor. That means that if you think you have a different, better way, there’s basically no way to put it into commercial service. Congress recognized this problem and passed a law to try to change it a couple of years ago, but the Nuclear Regulatory Commission isn’t really implementing it. We should change that. Maybe it will unleash a brave new world of atomic power, or maybe it won’t, but we at least ought to try.
The limited case for nuclear
There’s a specter haunting the internet whose only interest in nuclear power is to deride environmentalists and/or suggest that there’s no reason to have any kind of energy policy other than pointing out that the United States (and most other countries) made serious policy mistakes in the 1970s. This is the nuclear bro, and it’s where I think Smith is right. You can’t go back in time and fix nuclear policy errors of a previous generation, and right now we can meet a large and growing share of our electricity needs by deploying more and more solar. Solar is cheap and getting cheaper, and while batteries are not cheap, they are definitely getting cheaper.
I’m very pro-solar. I have solar panels on my house, and I am in fact more pro-solar than conventional liberals. I want to stamp out the anti-solar NIMBYs who elevate saving random trees or historic preservation over generating more solar power.
But I also think these debates tend to play out in a weird way as if we have a fixed pool of subsidies to dole out that different potential energy sources are in zero-sum competition for. That’s not really how it works. If you want to subsidize carbon-free electricity production, you can just do that in a mode-agnostic way. What nuclear really needs is specific regulatory changes that would give advanced reactor designs a chance to prove themselves. If it works out, the benefits could be very large simply because energy is really good, and if you want more and more energy it’s useful to have sources that don’t take up as much space as the endless proliferation of utility-scale solar projects.
If the regulatory changes still leave us in a world where small modular reactors never pan out, then that’s bad news for the world but there’s no harm done. We should make the changes that would be helpful without fear or prejudice and without too much of the annoying game of “political pundits make guesses about future technology.” There was a time when most people thought fracking was a pipe dream, but coal gasification would be crucial to the future of energy. Advanced nuclear might or might not pan out, but we should have a regulatory environment that gives it a try.
Deterministic vs. agnostic regulation
To say something nice about the FDA for once, the way they regulate proposed Covid-19 medications is by asking you to prove that the medication is effective at improving outcomes among Covid-19 patients and prove that it’s safe. Actually demonstrating that to the FDA’s satisfaction is an arduous process, and there are a lot of questions one can raise about the specifics of how they do their work. But it’s a nice flexible standard. I’m sure the people who work there have a lot of opinions about which lines of research are the promising ones, but pharmaceutical companies can try whatever they want as long as they provide evidence.
Automobiles are regulated in a different way.
You can’t just build a car and say “this car is safe to drive,” and then prove that experimentally. The rules are deterministic. Each vehicle needs to have certain precise characteristics that the regulator believes enhance safety in a cost-effective way. One consequence of this is that the rules are actually totally different in the European Union. The Toyotas that Toyota sells in Sweden aren’t the same as the Toyotas that Toyota sells in the United States, even when it’s the “same model” of car. And the Japanese Toyotas are different again.
There’s no dramatic difference in aggregate safety performance; there are just a lot of little differences. A while back as part of the now-defunct Trans-Atlantic Trade and Investment Partnership (T-TIP) talks, there was a discussion of trying to harmonize these regulations, which led to a fun NPR segment spelling out some of the details — in America we have bigger windshield wipers, for example.
The auto rules are a little funny and bureaucratic, but in this particular context, I think there is some upside to deterministic regulation. You can walk into a car rental place, get a make and model of a car you’ve never driven before, and get yourself up to speed very quickly on how to drive it safely. There are probably lots of perfectly good ways a car could work, but having them all work very similarly gives us big interoperability advantages even if it does stymie some innovation.
But applying the same logic to the nuclear industry creates problems.
People want to build new kinds of reactors
I am not a scientist, but apparently there are lots of different ways that you can make a stable nuclear fission reaction. Decades ago, the United States settled on a specific kind of nuclear reactor known as a light water reactor. The civilian versions of these are really big and they generate a ton of electricity. They need to be built on-site, and because they are so large, the cooling apparatus also needs to be really big. The size and scale create a lot of potential points of failure, so you need a lot of redundant systems and backup.
But these days folks are amped up about a family of different reactor designs that they call “small modular reactors.”
These are much smaller (and cheaper) than a large light-water reactor, and they don’t make as much electricity. The hope is that you could build them in a factory and then plop them down wherever, greatly reducing construction costs.
Small reactors would also have applications that existing light-water designs are unsuitable for. You could locate one in a factory and co-generate electricity and heat, for example. Or they could provide power on islands or remote communities that are currently reliant on oil power. There’s my dream of nuclear maritime shipping. If you want the details on this, ask a physicist or an engineer. It’s also my experience that there are lots of little sub-communities within this space, each with its own ideas and objections to other people’s ideas, and I’m not here to vouch for anyone’s actual or hypothetical technology.
What I’m here to do is say that the proponents of these designs are severely hampered by the deterministic regulatory process. A big part of the value proposition is the idea that these smaller reactor designs can be operated safely with a less elaborate apparatus. Indeed, their proponents say small modular reactors will be safer than existing light-water designs (which are already extremely safe), even without nearly as many backups and redundancies.
Is this really true? The Union of Concerned Scientists says it is not, but they seem to me to be in the minority. I think one tell on the UCS report is that they conclude by saying that Congress should focus more on improving the safety of existing light-water reactor models — but these are extremely safe already; the problem is the exorbitant cost of building them. It signals to me that these doubts are coming from a place of anti-nuclear ideology rather than a sober look at the tradeoffs.
But, again, I am not a nuclear engineer. The policy problem here is that given deterministic regulation, there’s no real way to introduce a disruptive new reactor design and prove that it’s safe. To get a commercial license, you need to do everything in just such a way so big change is impossible.
Congress already tried to address this
In a good example of “quiet climate policy,” back in the 2018-19 lame-duck session, Congress passed the Nuclear Energy Innovation and Modernization Act which had a bipartisan group of Senate co-sponsors (Cory Booker! John Barrasso!) and passed the House 361-10.
The bill had a few provisions, but for our purposes the crucial issue, as Sonal Patel wrote for Power at the time, is that it directed the Nuclear Regulatory Commission to “complete rulemaking to establish a ‘technology-inclusive, regulatory framework’ that encourages greater technological innovation for the advanced nuclear reactor program.”
We are not yet at the deadline for NRC action on this mandate, but as Adam Stein explains, the preliminary rule basically fails to meet Congress’ objectives. They are proposing a new licensing regime (good) but creating extra safety criteria. Again, it is good for nuclear power plants to be safe. But as with the UCR recommendation to make light-water reactors even safer, there is a forest/trees issue here. Energy is really useful, and burning fossil fuels to get the energy is not “safe.” We shouldn’t start licensing new designs that are palpably worse than old ones, but existing reactors are perfectly safe — there’s no need to set the bar higher and a real need to at least explore the possibility of a big new energy source.
Making bureaucracies move is challenging. The institutional origin story of the NRC is that people got concerned that the old Atomic Energy Commission was too pro-nuclear, and they wanted a regulatory structure that would pump the brakes. The brakes were successfully pumped, with the result that greenhouse gas emissions and particulate pollution got much worse over the course of the 1980s, 1990s, and 2000s than they otherwise could have. And the institutional culture of the NRC is one in which they feel they will be yelled at if anything goes wrong in nuclear power, but neither fossil fuel interests nor the environmental movement will mind if no new nuclear is built.
Leadership from the White House and the various appointees to the NRC can make a difference here. But so can leadership from Congress. NEIMA passed overwhelmingly, but it was also a sufficiently low-key piece of legislation that I don’t think necessarily sent a super-clear message. Regulators need to hear in public statements from elected officials and hearings that they want to see nuclear given a real shot.
A bigger reform
An even larger reform, which as best I can tell is considered so pie in the sky that industry people won’t even ask for it, is that nuclear power be subjected to the same cost-benefit scrutiny as environmental regulation is in general. It’s just that the considerations would be flipped.
Fossil fuel power plants create pollution. Environmental Protection Agency rules limit the amount of pollution that they are allowed to make. And when the EPA proposes a rule, it’s supposed to show that the benefits of the pollution reduction (in terms of climate, but also particulates, sulfur, etc.) are worth it compared to the costs. Air pollution, as I’ve written, is really bad. But it’s not so bad that we’d be better off without any electricity. It’s a regulatory balance of considerations, and that’s a sensible way to think about things.
Nuclear safety should be seen in the same light. If you permit a new reactor, you are imposing some safety risk (bad!), but you are also giving the world a bunch of electricity (good!) that creates no air pollution (very good!).
The right permitting standard should take those benefits into account. The fact that we don’t call it a “safety” issue when a natural gas plant or an internal combustion engine pumps noxious fumes into the sky doesn’t change the fact that it’s a risk to human health and well-being.
In the future, if we have so much solar and wind power that the practical impact of licensing new reactors is to compete with renewables, then it would make sense to raise the safety bar. But that is not currently the situation; there is plenty of room for multiple zero-carbon fuel sources to grow, especially because we also aspire to electrify our vehicle fleet. For now, a reactor that was 10 times riskier per terawatt-hour than existing light-water designs would still be much safer than a gas power plant and dramatically safer than coal or oil. Again, to be clear, when I floated this talking point with some people who work in the industry they hated it — they maintain their designs are safer than current light water and don’t want any indication that anyone anywhere would consider any compromises on safety.
So that’s great and I won’t suggest it either (wink). But the fact remains that we really ought to integrate our regulatory strategies across domains with the risk of nuclear plants measured against the very real benefits of tapping new scalable forms of zero-carbon energy. It’s worth saying that we already apply a different regulatory standard to military and research reactors than to commercial ones, and there are no apparent problems with the alternate regulatory regime. MIT has a nuclear reactor sitting in the middle of Cambridge, Massachusetts, and nobody minds. The United States has never had a nuclear accident on a submarine, even the ones that sank. All of the actual problems in this industry relate to costs, and the folks who claim they can bring those costs down with new approaches deserve a shot.
I study energy and nuclear stuff as a historian--power generation in my MA work, nuclear applications in medicine for my PhD--and I see a lot to like in this post. I also currently do public health work, so this is right in my wheelhouse. And I more or less agree with the basic premise.
I think the nuclear debate, like all energy debates, gets badly derailed by the unwillingness of participants to engage with the full complexity of the cost structure. When you say "nuclear is safe," or "nuclear is clean," or "nuclear is cost effective," or whatever, the meaning of those claims is subject to manipulation almost beyond recognition.
Here's a simple example: How are you amortizing the cost of waste storage? How long will the waste storage and protection costs need to be borne? Waste storage was supposed to be a cost borne by taxpayers in the US (in order to stimulate the industry), but that doesn't make it go away, any more than the cost of carbon dioxide emissions goes away just because a coal plant operator isn't paying that cost. And if you cost out the active maintenance of storing anything for a thousand years, that cost is going to be pretty high.
Likewise, it is totally fair to say that nuclear power has a good safety record relative to coal! But if you do a comprehensive analysis of safety issues, the balance between nuclear and, say, geothermal enabled with deep horizontal drilling, nuclear starts looking pretty bad. That's because you have to add in what might be described as a "catastrophic risk premium," i.e. the very low chance of a highly catastrophic event. And despite the relatively low number of nuclear power reactors in the world (relative to fossil fuel plants), they suffer mildly catastrophic breakdowns at the rate of roughly one per twenty or so years. There has only been one Chernobyl level event so far, but in subsequent analysis of the Fukushima and Three-Mile Island events, it turned out that they were more dangerous than initially recognized. TMI, in particular, was much closer to having an explosion and / or suffering a pressure vessel breach by meltdown, with an accompanying massive radioactive material release, than realized. No one knew how bad the meltdown and loss of core geometry was until they finally opened it up in the cleanup, years after the event. The good news is that they didn't become Chernobyl-level catastrophes! But that's not how risk works. If you see almost-catastrophes on the regular, you can more or less predict that sooner or later you will get the full enchilada. And the fact that these events keep happening for, essentially, different reasons every time (human error, equipment breakdown, natural disaster) tells you that the underlying system is prone to catastrophic, chain-reactive failure.
Moreover, you can't think about the safety of nuclear without giving thought to the safety issues around waste storage--I tend to think that the danger of someone using waste to make a "dirty" radiological weapon is small, but it is not zero, and good old fashioned water or soil contamination is definitely a potential health hazard--and around proliferation. Modern reactors don't produce weapons grade material...unless you want them to. And remember that producing weapons-grade material is, at this point, a 75-year-old technology. It isn't that hard. One of the big bottlenecks for potential proliferators is access to the unenriched nuclear material; that access will get naturally easier as the industry grows, for all the basic reasons of supply chains.
Speaking of which, mining of nuclear materials is itself a highly fraught enterprise with a dubious health record, major safety and environmental concerns, and a nasty legacy. And remember those reactors that Matt mentioned on ships that sank? Good news! They haven't produced major releases! Bad news! It takes a long time for modern ships to break down--they are well constructed out of tough, resilient alloys. But "long time" and "forever" are not the same thing. "Long time" and "1000 years" are not even the same thing.
None of this means you can't do nuclear. These are all problems that have solutions, and you could level similar types of concerns at every energy technology. Producing batteries requires mining activities with many of the same concerns. Both PV cells and batteries have serious environmental costs around production and disposal. A lot of solar tech precursors are currently made with slave labor. I can tell you a great story about why space-based solar was (and maybe still is) potentially a really, really bad idea. Fossil fuels are a huge disaster all the way around.
But I am personally pretty meh on nuclear for the simple reason that I described above: it seems to be prone to catastrophic, chain-reactive failure. The simple reason for my assessment is that fission-based technologies are, at a deep, fundamental level, quasi-uncontrollable. They involve forces that are hard for humans to parse. Two examples come to mind:
1) Chernobyl blew up in part because of the design of the control rods, which could not be inserted fast enough to overcome the fact that they initially could cause a spike in fission rates by displacing neutron-moderating water. But "not fast enough" was around 1.3 ft/s--it took less than 30 seconds to scram the reactor. It's not like the rods were being screwed in or something. Stuff just happens really, really fast in a nuclear reaction (which is why you can use it to make a bomb). Humans struggle to grok that.
2) Fukushima was a problem in part because scramming the reactor (shutting down the reaction) is not enough. Fission reactions get so hot that the pile in a big reactor can melt itself down for literally days after you scram the reactor. And if the fuel melts together, it will start fissioning again (this is problem with loss of core geometry), because remember that sufficiently radioactive isotopes will do their thing with nothing more than simple proximity. There have occasionally been "natural" nuclear reactors in the wild for precisely this reason.
So with nuclear you always have the fundamental problem that enriched fuels want to do The Bad Thing on their own, and your whole fancy reactor design is about getting them close, but not too close, but fissioning, but not too fast--it's a goldilocks engineering problem, and goldilocks engineering is actually really hard, because you can screw up in either direction. And in this case, if you screw up, The Bad Thing can happen at a speed too great for humans to effectively parse and respond, given the limitations of our senses and cognition.
So, TL:DR - All energy production systems have big problems related to the complexity of what you are trying to do. But given that we are now living through that disaster with fossil fuels, I would argue that we should be much, much more risk-averse in our next step, and I tend to be unconvinced that fission-based nuclear meets that criteria because of both the accumulation of little dangers around the technology and the catastrophic tail risk that it represents.
I work in the energy field, and all I want to comment on is this.
"You can walk into a car rental place, get a make and model of a car you’ve never driven before, and get yourself up to speed very quickly on how to drive it safely."
Sure, maybe you can figure out the basics.... but all the little details of how to pop the gas cap or trunk, or attach Bluetooth you definitely can't figure those out very quickly. As someone who rents probably 30 cars a year.... It seems like I have to google features pretty regularly.
Now energy. This post is no fun because it makes sense. Hell it makes sense to people on all sides of the political guiding line, except maybe for the hardcore anti-nuke greens.
As someone who works on gas turbines for a big energy company, renewables don't scare me. They aren't going to put me out of a job in the next two decades... but Nukes... nukes would put me out of a job quickly. Well not technically, I could work on steam turbines as well (I do sometimes), but they suck to work on.
I'm not sure why, but the US has lost some of its ability to innovate in the last few decades. Ok, I am absolutely sure why... stupid regulations.
The same sort of inflexibility that prevents energy innovation is the same reason our CDC and FDA suck and we don't have rapid covid testing.