A more complete description with the math and the code is available on the GKG website and below for people who want to replicate the results. I am deeply indebted to Daniel Murray for working through the math, and then patiently taking me through it.

PDF of Rate SNT

2.01MB ∙ PDF file

Download

Here's the PDF. It has the fully panoply of math symbols, and an appendix with the proofs and codes.

Download

Figure 1. Simplest SNT on Okuma low outdoors dose rate profile, first 50 days only. The SNT mortality curve is shown in red.

Figure 2. Rate SNT on Okuma low outdoors dose rate profile, first 50 days only. Read the undone damage on the far right. The damage is measured in equivalent sieverts.

The continuous version of SNT

As soon as we acknowledge our bodies’ remarkable ability to repair DNA damage, we must carefully distinguish between damage and harm. Harm is cancer incidence or mortality. Damage is some form of DNA disruption. From this perspective, an increment of sieverts, a dose rate over a short period of time, does two things:

1) Increases the subject’s chance of cancer by creating a viable mutation that may lead to cancer.

2) Increases the damage to the subject’s DNA by direct effects and producing Reactive Oxygen Species. This increased damage (e.g. more double strand breaks) increases the chance that the next increment of dose leads to a viable mutation.

Throughout this piece, repair means ``safely dispose of as far as cancer is concerned’‘. This can be done by correctly rejoining a double strand break (DSB) in the DNA, or by triggering the cell’s death, or by simply giving up and failing to rejoin the broken ends. A cell’s whose DNA is not rejoined can’t replicate, and can’t cause cancer. We call the damage that has been safely disposed of done damage.1 The remaining damage is undone.

Simplest SNT is an all or none model. A repair period of a day means, if two dose pulses are incurred in the same day, they are treated as a single dose and added together; and then SNT’s S-shaped acute dose response curve is applied to that sum to estimate the harm, the cancer mortality. But if the two pulses are a day or more apart, then they are handled separately. The mortality for each dose is computed, and then those two mortalities are added together to determine the harm.

Suppose each pulse is 5 mSv. If the two pulses are on the same day, we consult the sigmoid acute response curve at 10 mSv and find the cancer mortality is 3.2e-5 (3.2 chances in 100,000). But if the two doses occur on different days, we go to the response curve at 5 mSv, find that the mortality is 7.1e-6 (7.1 chances in a million), and the total mortality is 7.1e-6 + (1 - 7.1e-6)* 7.1e-6 = 1.4e-5 (1.4 chances in 100,000) or about 40% of the combined dose.2

If two dose pulses are received in the same day, then there is no repair. If two doses are separated by a day or more, then the repair is complete. The first dose pulse has no impact on the second pulse. Simplest SNT does a far, far better job of replicating actual cancer data than LNT, when a dose is spread more or less evenly over a protracted period; but the all or nothing repair process is obviously artificial.

Daniel Murray, a radiation researcher in Melbourne and math whiz, has developed a continuous SNT. In Danny’s rate version of SNT, the dose is received continuously and repaired (aka done) continuously. The repair period is replaced by a constant repair rate. The repair rate is proportional to the amount of undone damage. This leads to exponential repair. Mathematically, Danny’s model can take any repair process that depends only on the current undone damage. But for regulatory purposes, we choose to use the simplest repair model that is consistent with the biology. To paraphrase Einstein, we want the simplest model possible but no simpler.

Danny’s rate SNT depends on a remarkable result that he came up with. For any harm function, that combines period harms by multiplying together the probability of no harm up to period i and the probability of harm in period i, the incremental harm in period i conditional on having survived up to that point is h(u_i + d_i) - h(u_i) where d_i is the damage from the dose in i, u_i is the undone damage at time i. We call this the Add Subtract Rule. Add the period dose to the existing damage, compute the harm resulting from that sum, and then subtract the harm from just the existing damage.

The Add Subtract Rule results from accepting two simple rules about incremental harm. Call h_i(u, d) the incremental harm in period i. If the dose in that period is d and the undone damage in the period is u.

A) Acute consistency: h_i(0, d) = ci(d), where ci(d) is the mortality from an acute dose of d. If there is no residual damage, the incremental harm is the acute response to the dose.

B) Subdivision consistency: h_i(u,a) + h_i(u+a,b) = h_i(u,a+b). There is no difference between starting at u and receiving dose a and then immediately receiving another dose b, and starting at u and receiving a dose of a+b.

The proof is simplicity itself. Start with no damage, get hit with dose u and then get hit with dose d. Using subdivision consistency: h(0,u) + h(u,d) = h(0,u+d). Using acute consistency, this is ci(u) + h(u,d) = ci(u+d) or h(u,d) = ci(u+d) - ci(u), which is the Add Subtract Rule. This is the first time I have inflicted a mathematical proof on the congregation; but Danny’s result is so important (and the proof is so simple) that it is worth the pain. The Add Subtract Rule’s key requirement is that the damage can be expressed in units of the dose. But that’s exactly what the gray to sievert conversion is supposed to do. Despite the same units, a sievert is not a measure of absorbed energy per kg tissue. It’s a measure of DNA damage severity.3

In the case of radiation induced cancer, the damage we are most concerned about is double strand breaks(DSB’s) in our DNA. We have considerable evidence that the number of DSB’s is linear in absorbed energy per kg tissue. Linear in grays. But for heavy particles, the DSB’s are much more localized. The damage is far more clumped. Clumped damage is worse than evenly spread damage, probably because closely spaced DSB’s are much more likely for a given amount of grays. Two closely spaced DSB’s can be viably misrepaired by rejoining the wrong ends.4 Some of those mutations will lead to cancer. The RBE is our best attempt so far to try to measure this.5 This is the thinking behind tracking DNA damage in equivalent sieverts.

The RBE is largely based on the number of dicentrics that a gray of each type of radiation produces. Normally, during replication, the chromosome takes the form of a lopsided X. The two identical halves are connected at a single point. A dicentric is a chromosome with two crossing points. Dicentrics were chosen mostly because they are a DNA aberration that could be counted with an optical microscope. (Importantly, dicentrics are damage, not cancer. The RBE measures damage, not cancer. LNTers assume cancer is proportional to damage. So they conflate the two.) Radiation induced dicentrics are caused by closely spaced DSB’s. The radiation protection people may have stumbled on a useful measure of clumpiness.

According to the Add Subtract Rule, a subject could have an immense amount of undone damage; but, if the dose in period i is zero, then the mortality from that dose is zero. But if the acute response curve is curved upward, the mortality from the same period dose will be larger if the undone damage is large. If the acute response curve is linear, then we are back to LNT.

The Add Subtract Rule makes rate SNT work. The math required is fairly advanced. What I’ve been calling “incremental harm” is the DRP’s hazard function, the probability density of cancer from the DRP at time t, given no mortal cancer up to this point. To put it bluntly, it’s the death rate at time t. We must multiply all these conditional probabilities together, and then accumulate the hazard function over the entire period from 0 to t. We then move on to the next larger t and repeat, to get the DRP’s mortality curve. In so doing, the Add Subtract Rule difference translates into the slope of the hazard function times the dose rate as the time interval goes to zero. Amazingly Danny can implement all this in a few lines of code.

Danny’s also figured out that the repair rate that is most consistent with a 1 day repair period in simplest SNT is 1.93 per day or a mean time to repair of 0.517 days or a half-life of 0.358 days (8.6 hours).6 This is consistent with the Berkeley data which showed that the repair process was essentially complete in a day, whether or not the DSB had been repaired.\cite{bissell-2011}

Figures 1 and 2 compare rate SNT with simplest SNT on the Okuma low DRP with a location factor of 1.0 (eg the ambient outdoor dose rate). The mortality curves are almost the same. The big difference is under continuous SNT we must keep track of the undone damage.

An important concept for this type of model is the equilibrium damage. For any combination of damage rate and mean repair time, there is a level of damage, where the damage process and the repair process balance. Damage is being repaired just as rapidly as new damage occurs, and the undone damage remains unchanged. It turns out this level is d_r * tau where d_r is the damage rate and tau is the mean repair time. For a mean time to repair of 0.517 days, the equilibrium damage is a little over half the damage/dose rate numerically. If the current damage is below the equilibrium level for the current dose rate, the damage will increase. If it is above, the damage will decrease.

Figure 3 shows how rate SNT responds to a profile in which the dose rate is 100 mSv/day for 1.2 days, then 500 mSv/day for 0.4 days, then 100 mSv/day for 0.4 days.

Figure 3. Rate SNT for a dose rate profile of 100 mSv/day for 1.2 days, followed by 500 mSv/day for 0.4 days and then 100 mSv/day 0.4 days.

In the first segment, the undone damage is below the equilibrium damage for the 100 mSv/day rate. The undone damage climbs slowly. By 1.2 days we are almost at equilibrium damage; but then the dose rate jumps up. In the 500 mSv/day segment, the undone damage rises rapidly. The mortality curve is curved upward, since the undone damage is increasing. But the damage curve is curved downward because we are getting closer to the equilibrium damage for that dose rate. When the dose rate drops at 1.6 days, the damage curve abruptly turns downward. It’s hard to see, but the mortality is curved downward because the damage is decreasing, but rather slowly.

For this DRP, both LNT and both SNT’s come up with roughly the same mortality, with SNT a little higher than LNT. This should not be surprising. This DRP is a large total dose (360 mSv) delivered over a very short period, most of it in 10 hours. This is the kind of dose rate profile that LNT does reasonably well on.7 But this DRP in no way represents the kind of DRP that would be incurred by the public in a big release, where a similar sized total dose would be spread over years and decades.

Suppose we spread the same total dose over 200 days, Figure 4. LNT which denies DNA repair says the mortality is unchanged. But the SNT mortality is down by a factor of 80, and nearly 60 times lower than the LNT mortality. The two SNT’s produce virtually the same mortality. Notice the undone damage reaches equilibrium in a few days. As long as the undone damage is at equilibrium, the mortality curve is a straight line, the same increment of mortality in each period. In most real world release profiles, the repair rate will be faster than the rate of change of the DRP, except possibly during plume passage.

Figure 4. Rate SNT for a dose rate profile of 1 mSv/day for 120 days, followed by 5 mSv/day for 40 days and then 1 mSv/day for 40 days.

Figure 5 shows a real world dose rate profile. At Fukushima, the control room personnel endured three 8 hour shifts in a control room sandwiched between two badly breached reactors, Figure 6. The dominate exposure was inhaled Iodine-131. Since iodine concentrates in the tiny thyroid gland, multiplying the dose rates to that organ by about a factor of 1000, the main concern is thyroid cancer. These people were carefully measured with full body counters, so we have a good idea of the total internal dose, and can work that backwards to a fairly good estimate of the dose rate profile. The biggest total internal dose was around 600 mSv. Figure 5 shows the GKG estimate of that dose rate profile.

Figure 5. Fukushima control room dose rate profile with total dose of 600 mSv.}

The DRP consist of three big jumps representing the three 8 hours shifts in which the I-131 was inhaled, followed by a fairly rapid decline thanks to I-131’s halflife of 8 days. The peak dose rate is about 50 mSv/day. Almost all the mortality is incurred in the first 20 days. The rate SNT mortality is nearly the same as simplest SNT’s. Both numbers are about 5 times lower than the LNT mortality. As of late 2019, no excess thyroid cancer had been detected in the workers.\cite{kitamura-2022}[p 5] But the cohort with this large a dose is very small. Most of the 5000 workers being tracked received far less damaging DRP’s.

Figure 6. Fukushima. Control room circled was manned throughout the release.

This is about as jagged a dose rate profile as will be incurred by the public in a nuclear power plant release (and the dose rates are much larger). This example tells us that simplest SNT will result in practically the same results as rate SNT in just about all real world releases.

Properties

Rate SNT has a number of attractive properties. The first four it shares with simplest SNT.

How rapidly the dose is received is critically important.

Both SNT’s are consistent with the fact that 1000 mSv received in a minute or two creates an entirely different amount of harm than a 1000 mSv spread evenly over 50 years (0.055 mSv/d). LNT says the two harms are the same. According to SNT, a 1000 mSv acute pulse results in an increased cancer mortality of 0.064, a bit more than LNT. But the increased cancer mortality associated with 0.055 mSv in a day for 50 years is an increased cancer mortality of 7.0e-6, 7 chances in a million, which would be undetectable. SNT models both the harm associated with a large dose received over a short period, and the lack of detectable harm when the same dose is received over an extended period. Any model that cannot do this must be rejected.

Under both SNT’s, dilution is an effective countermeasure

This is true, even if in doing so, we increase the exposed population proportionally. If we are able to dilute from a single person acute dose of 50 mSv’s down to an acute dose of 1 mSv, at the cost of increasing the exposed population by a factor of 50, the collective mortality goes from 1.19e-3 to 50 * 2.42e-7 = 1.21e-5, reducing the mortality by nearly a factor of 100. If a necessary task requires significant radiation exposure, spread it over a large number of people. LNT says it does not make any difference.

There is a cumulative effect

Sometimes LNTers will argue for LNT on the grounds that there is a cumulative effect, as if non-LNTers are denying that obvious fact. Of course, there is a cumulative effect. But in the case of SNT, we accumulate harms, not doses.

There is collective harm

Under both SNT’s not only is there a cumulative effect; but, if we make the ethical judgement that all lives are equal, then we can add the individual mortalities, to produce a metric of the total harm associated with a release, as we just did in the second last paragraph. In dealing with cancer, Lost Life Expectancy, which can be derived from the mortality, is a more meaningful measure of harm than mortality. The ethical judgement required is all life days are equal.

The entire dose rate profile counts

For rate SNT, the whole dose rate profile counts. The peaks in a DRP may be more important than the valleys; but they are not all important. In particular, the spacing between peaks is important. A larger spacing means less residual damage. Simplest SNT (and LNT) claims whether two pulses are a day apart or 10 days apart makes no difference.

Dose rate increments are not independent.

LNT and simplest SNT claims what happens before day n has no impact on the harm from the dose in day n. This assumption is at best an approximation, and in theory can be a bad approximation. Continuous SNT keeps track of the undone damage. A small dose has more impact if it follows a big dose than if it precedes it. 100 mSv over a day followed by 700 mSv over a day has a mortality of 0.068. Reversing the order increases the mortality to 0.071. But even in this artificial DRP, the effect is surprisingly small. and, if the DRP varies slowly relative to the repair rate, the effect becomes very small.

Importantly, rate SNT shows us how well simplest SNT works on the dose rate profiles of interest in nuclear power regulation, despite ignoring this effect. The answer is pretty darn well.

The Policy Issue

The congregation is faced with an important policy issue. Do we exchange simplest SNT with rate SNT as our proposed replacement for LNT? The policy implications will be virtually the same. The mortality results and the radiation exposure compensation will be nearly the same.

Mathematically, rate SNT is far more attractive. There is no need to defend the independent period assumption, which will be attacked even though LNT makes the same assumption. The results are likely to be more accurate in the few cases where the results do differ significantly. A repair rate is something that can be measured empirically in a way that repair period can’t.

But because of this, rate SNT can be attacked on the grounds that the DNA repair rate is not constant. The actual repair rate is a function of the DRP. Small priming pulses can stimulate the repair rate. We have lots of evidence that, at very large dose rates, the DNA repair rate falls, possibly because the repair process itself is damaged. Weirdly this can reduce cancer rates, probably because most cancer starts with a viable misrepair. Less repairs, less misrepairs. Do we want to get into this morass, rather than just saying both LNT and SNT make the same untrue assumption; but SNT’s results are far closer to reality on spread out dose rate profiles?

Rate SNT is harder to explain and understand. Few politicians will be able to comprehend the math. People are unlikely to accept a radiation harm model they don’t understand. Why not just use rate SNT’s results to show that, despite the all-or-nothing assumption, simplest SNT still does a good job of estimating the harm?

I have waffled back and forth on this for a couple of weeks. The GKG has decided to do the democratic thing and rely on mob wisdom. Here’s the poll.

The poll will be open for a week. Supporters of either option can make their case in the comments section. You might want to wait a day or two to see those arguments before voting.

Figure 1. NRC Headquarters. Occupancy about 34% (Federal employees, June 2024.) I wouldn’t want to work in this building either. It’s worse on the inside.

In the NRC’s world view, making people poorer is unimportant. Global warming is not their business. Harm from alternate sources of electricity is not a consideration. Evacuation and exile costs can be disregarded. The only thing that counts is radiation harm, based on a harm model that is biological nonsense. It is a preposterously blinkered outlook; and it is costing humanity dearly.[Infallible Preacher]

I’m always reluctant to point out individual examples of the auto-genocidal nature of the NRC’s dictates, because it implies that we can ‘change’ the NRC incrementally. It this or that rule is nuts, just fix it; and in the end the NRC will be aligned with societal welfare. Nothing can be further from the truth. The problem is an all powerful bureaucracy whose incentives by design are anti-societal. And the people who gave that bureaucracy that autocratic power and those anti-societal incentives is us. The problem is us. We must change; and we must replace the current system.

But occasionally I stumble on a regulation that is so blatantly anti-societal that perhaps we can use it as an example for why the whole system must be replaced, not just a few individual regulations improved. The Design Certification deep freeze is such a rule.

Design Certification goes back to 1989. Before that every individual plant seeking a license had to start from scratch, even if the plant was essentially a copy of an already licensed plant. This meant the NRC was reviewing and reapproving the same design multiple times.

Under Design Certification, a vendor could work through the Design Certification process once, and then any plant could “reference” that Design Certificate in its actual license application. On the surface, obvious good sense. Below the surface, it’s not so clear. Vogtle 3/4 used the already Design Certified AP1000. How did that go?

Originally a Design Certification was good for 15 years, and then it had to be renewed. In September 2025, the NRC extended the Certification period to 40 years, claiming this would reduce regulatory burden and save $55 million over a 65 year analysis period.

Yesterday I learned something I did not know. A vendor can only change his Design Certification “not less than 12 nor more that 36 months before the expiration of the initial 40 year period”.[10 CFR 52.57(a)] The design must be frozen for 37 years. What possibly could be the intent of this limitation? Does the NRC expect that the initial design to be so perfect, that the vendor will learn that there is nothing to improve after building the first one? Does the NRC expect technology to remain static for 37 years?

Of course not. Even the jailhouse layers at the NRC, who call themselves engineers, are not that lobotomized. They knew that the requirement would force the vendors to apply for a brand new certification every time they wanted to improve their design.

There is only one possible explanation: bureaucratic security. The NRC was worried that Design Certification would so reduce the “regulatory burden” that their jobs would be endangered. The Vogtle 3/4 experience indicates that the bureaucracy has other ways of protecting itself; but who knows when we might need another layer of protection. We might call this defense in depth.

Last April, Westinghouse submitted a request to the NRC asking for a “renewal” of the AP1000 certification incorporating the design changes already approved during the construction and licensing of Vogtle Units 3 and 4. A “renewal” is the only legal way of changing the Design Certification.

On July 2nd, the Commission issued a one time exemption to Westinghouse allowing them to apply for a renewal of the AP1000 Certification more than 36 months before the 40 year expiry, saying “the exemption is authorised by law, will not present an undue risk to the public health and safety, and is consistent with the common defence and security”. Duh!

So we impose a regulation whose exemption does not present an undue risk to the public health and safety, and then grant exemptions on the basis of political expediency. I don’t know about you, but I’m really tired of this nonsense.

On July 1, the NRC released its proposed rule changes in response to Trump’s May,2025 Executive Order requiring the NRC to “reconsider” LNT. A month ago I predicted what the document would say and would not say. Here’s my score card. My prophesies are in italic.

The response will not say that LNT is biological nonsense. It will not acknowledge that LNT is the no DNA repair theory whose foundational premise is contradicted by well established, indisputable biology. It will not point out that the radium dial painters and the Taipei apartment dwellers clearly falsify LNT.

This was a lay up. Instead we are told

the NRC recognizes that there is insufficient evidence to refute the use of the LNT model[page 24];

and given this gibberish:

The NRC considers its current standards for protection against radiation to be science-based to the extent that adequate methods and analyses have been applied in the works that have been referenced in the development of the NRC’s regulations and guidance.[p 21]

My second prediction was.

The response will embrace the Muller Trap and accept the false LNT or no threshold dichotomy.

Another layup. The NRC is quite explicit:

To date, no threshold has been established in this model, meaning that as the dose decreases to zero, the corresponding risk follows proportionally to zero.[p 24]

Emphasis mine. Proportionally is the key word in this non-sequitur.

It will offer the deeply flawed INWORKS study as support for LNT, and ignore Life Span Study (LSS) of the bomb survivors on which the Japanese and American taxpayers have spent nearly a billion dollars over 70 years, despite the fact that until recently the LSS study was the NRC’s Goldstandard. What changed was the Radiation Effects Research Foundation admitted their LSS data is non-linear. It will point out that the LSS exposures were acute dose rate profiles despite the fact that LNT claims there is no difference between acute and chronic profiles. It will also ignore the Nuclear Shipyard Workers Study by far the best controlled of the nuclear worker studies.

Only half-right on this one. The NRC does ignore the Life Span and Shipyard Studies; but it also makes no mention of INWORKS. In fact, no data at all is offered to support its claim that LNT is “science-based”. The document carefully stays away from both DNA biology and radiation harm data. That’s how weak the case for LNT is.

It will not say what counts is not the cumulative dose, not the average dose rate, but the entire dose rate profile. In fact, it will never use that term. Instead it will repeatedly talk about the “dose”, implicitly accepting LNT. There will be no mention of the repair time. What recommendations it comes up with will be based on the biologically meaningless and medically dangerous period of a year.

Mostly right. “dose” as in total dose is used over and over again, which is an implicit acceptance of LNT. All the limits are based on a year. Repair time is never mentioned. The word “repair” shows up exactly twice in 180 pages, once only to make the point that DNA repair does not necessarily imply a threshold:

At present, there is no scientific consensus establishing a threshold below which stochastic health effects do not occur. As described in SECY-12-0064, as it relates to stochastic effects, ‘It is unlikely there might be a threshold level of exposure below which biological response does not occur. Such a threshold could only occur if DNA repair processes were totally effective in that dose range or if a single radiation track were unable to produce an effect. The cellular processes such as apoptosis and cellular differentiation that can protect against later phases of tumorigenesis are judged to be efficient but can be bypassed; there is no reason to believe that those defenses act differently on spontaneous and radiation-induced tumors or have specific dose dependencies.’[p 31]

But the proposal does not just “implicitly accept” LNT. The NRC reaffirms LNT. Trump and the DOGERS be damned.

Stated another way, the NRC proposes to continue to use the linear dose response model, as it continues to be the most appropriate model upon which to base a radiation protection framework;[p 27]

Accordingly, the NRC reaffirms its position that the linear dose response model is the most appropriate available consensus model for formulating radiation protection standards and planning radiation protection programs.[p 32]

The document then repeatedly uses LNT to compute the cancer incidence associated with its proposed limits. It also uses LNT to defend dropping the current public dose rate limit outside the plant:

Second, the NRC is proposing to delete the short-term dose rate limit in the unrestricted area of 0.002 rem in any hour [0.02 mSv/h] in § 20.1301(b) because it serves no safety purpose that is not already achieved by the public dose limit of 100 mrem per year. This is because a member of the public could receive the full 100 mrem annual public dose instantaneously and this fact in and of itself would have minimal safety impact.[p 68]1

Finally, the NRC uses LNT to allow Planned Special Exposures, in which a plant worker can “apply” his four previous years “under-exposures” (50 mSv - actual annual exposure) plus this year’s 50 mSv to obtain the limit for a single exposure. This could result in an acute dose of 250 mSv. Page 61 cites LNT correctly pointing out that under LNT there is no difference between 250 mSv received all at once and 250 mSv spread evenly over five years. This change is a full throated endorsement of LNT; and a little too risky for my taste. The SNT cancer incidence for 250 mSv acute is nearly 4%.2

It will not offer a well defined replacement for LNT. Therefore, the NRC has no way of converting an acceptable risk to dose rate limits. Despite this, it will call for some relaxation of the current limits, probably 50 mSv/y for nuclear workers and 5 mSv/y for the public, without offering a compelling argument for these numbers.

Wrong. The NRC’s proposed replacement for LNT is LNT. LNT is well defined. The 50 mSv/y is slightly relaxed by the Planned Special Exposure rule. There are some changes to the radiotherapy caretaker limits, which I did not know existed. The public limit is 1 mSv/y (not 5). It is not changed, but the emissions limit is relaxed from 0.1 mSv/y to 0.25 mSv/y. LNT is used to defend this change.

It will make a few handwaving generalizations about not requiring ALARA. But since it offers no replacement for LNT, there will be no way to do the benefit-cost analysis needed to actually tell the regulator where to stop. Whether or not ALARA is dropped in name, regulator incentives will remain unchanged and so will the regulator’s behavior.

Wrong, except for the last sentence. Under the proposal, the term “ALARA” is banned from all NRC regulations. But then the NRC goes to great pains to explain that the regulatory philosophy hasn’t changed. ALARA is partially replaced by a graded approach to exposures below the limits. The idea is we impose regulation below the limits to ensure that the limits are not breached. And just so everybody knows that nothing has really changed:

The graded approach to dose management would rely on a series of threshold doses below the regulatory dose limits and dose management actions to be taken at each threshold dose. These threshold doses and dose management actions would generally correspond to existing requirements in 10 CFR part 20 or other regulations and, therefore, should already be incorporated within the radiation protection programs of existing licensees to a degree ‘commensurate with the scope and extent of licensed activities’ as is currently required by § 20.110[p 48]

Again emphasis mine.

Cost-benefit analysis is now allowed as an argument against being forced to further reduce exposure; but the licensee has to show that it will cost more than $5200 to reduce exposure by one person-rem (10 mSv). In other words, the regulator can impose any requirement that costs less than $5200 per person-rem. Further the burden of proof is on the licensee to show that the cost is more then $5200. Cost-benefit analyses like these require all sorts of assumptions. Who decides whether the licensee’s assumptions are reasonable? The regulator.

There will be no discussion of the deadly downside of LNT, the unnecessary evacuations and exiles, the millions of lives shortened by fossil pollution, and the failure to raise billions out of poverty.

This was a slam dunk. There was a grudging admission that in the past some of our implementation of ALARA might have been overly conservative. Here’s as close as the NRC can come to an apology:

Regulatory experience, though, has demonstrated that the primary issue is not the standards themselves, but their implementation. For example, those standards that go beyond what is necessary for reasonable assurance of adequate protection and that can be interpreted as requiring continuous dose reductions even to very low doses of radiation have often been applied without a reasonable stopping point. In these instances, the radiation protection standards have led to outcomes that are overly conservative. As a practical matter, such over conservatism can significantly undervalue the benefits of activities when compared to their risks, sometimes at great economic costs or stifling of innovation.[p 23]

I guess you can call immediately killing 50 feeble old people, shortening millions of lives, and impoverishing billions economic costs.

The response will end up neither endorsing nor repudiating LNT, but call for more study, preserving the status quo.

Wrong. The NRC’s replacement for LNT is LNT. The proposal is a strong endorsement of LNT as a basis of regulation. The NRC does make one valid point. The document doesn’t really try to defend LNT on biological or empirical grounds. Their strongest argument for staying with LNT is there is no alternative. They make this point several times:

The NRC recognizes that there are limitations to the accuracy of the LNT model at very low doses, however, the NRC has also not identified a suitable alternative model.[p 32]

However, the NRC finds that no consensus-supported, regulation-ready alternative model to the LNT model exists at this time.[p 24]

At least some of the drafters/reviewers were aware of the Sigmoid No Threshold (SNT) model. I sent a copy of SNT: A Plug IN Replacement for LNT to Ms. Kenny who is the NRC lead for this document, and received a polite acknowledgement. SNT is “regulation-ready”. But the NRC has the “no consensus-supported” part right.

The radiation protection profession is split into two groups: LNTers and thresholders. The entire profession accepts the false dichotomy that it’s either LNT or a non-zero threshold. The LNTers reject SNT because it’s not linear. Much worse SNT comes up with results that say that a big release is not a big deal radiation-wise. SNT would put most of the radiation protection people, including the NRC, out of business.

The thresholders reject SNT because it accepts the no threshold doctrine. But they don’t offer a well-defined alternative, where well defined means a computer program for converting any dose rate profile into a prediction of cancer incidence. As a result there is no consensus-supported alternative. The NRC is quite content to use this fact to continue to base regulation on a model that is biological nonsense. LNT perpetuates the Intolerable Harm Lie, and that’s what the bureaucrats need to continue to govern nuclear power and us.

Bottom line: LNT’s auto-genocidal reign continues. A release will be just as deadly on paper if the proposed rules are imposed as if they are not. The regulators’ incentives remain unchanged whether to not they are allowed to use an acronym. ALARA is a symptom not a cause. It was just a codification of what the AEC was already doing with its autocratic power. ALARA was not officially adopted until 1974. The most massive increase in US nuclear costs took place between 1967 and 1973.

To expect a regulator to change regulation in a way that puts him out of a job is imbecilic optimism. This document is a clever attempt to blame LNT’s tragically wrong implications on ALARA; and thereby allow LNT to survive. LNT must be expunged from the outside. In this case, it is as simple as an Executive Order.

Figure 1. The BEIR Committee’s Prior and Posterior on the LNT slope. The blue spike says, after seeing the Karunagappally data, the Committee is nearly certain the LNT slope is somewhere between -0.11 0.126 and -0.13 0.146 per Sv.

Although it is 20 years old, BEIR VII is the most recent report by the National Academy of Sciences Committee on the Biological Effects of Ionizing Radiation.\cite{beir-2006} The report starts out with a version of The Dumbest Graph and maintains that static view of a dynamic process throughout. The report is over 400 pages long. TheUS shipyard workers, the Taipei apartment dwellers, Evans studies of the Radiothor drinkers, and most importantly, the the radium dial painters are never even mentioned. The Committee carefully avoids populations that have have received large doses spread over protracted periods, which is a test that any radiation harm model must pass. But in its Statement of Task, the Committee defines its job:

The primary objective of the study is to develop the best possible risk assessment for exposure to low dose, low linear energy transfer(LET) radiation in human subjects. In order to do this, the committee will (1) conduct a comprehensive review of all relevant epidemiologic data related to the risk of exposure to low dose, low-LET radiation;

...

In performing the above tasks, the committee should consider all relevant data, even if obtained from high radiation exposures or at high dose rates. \cite{beir-2006}[p vii][Emphasis mine.]

The Committee had to put that last part in because their preferred LNT model was almost entirely based on the atom bomb survivor data, which was large to enormous doses in a very brief period.1 But it makes their omission of the dial painters and the other large dose populations from a self-proclaimed “comprehensive review” scientific misconduct.

The unsurprising bottom line associated with this willingly blinkered approach is:

The Committee concludes that current scientific evidence is consistent with the hypothesis that there is a linear no-threshold dose-response relationship between exposure to ionizing radiation and the development of cancer in humans.\cite{beir-2006}[p 15]

I am not going to rehash how horribly LNT predicts cancer in situations where people have been exposed to very large doses spread over protracted periods. Nor am I going to waste any more time on why it goes so wrong. It’s too obvious. LNT denies our amazing ability to repair DNA damage, which is biological nonsense.

Rather I’m going to take a new tack. The authors of BEIR VII are Bayesians. How do I know that? On page 279 they gives us their best guess of the LNT slope (0.0970/Sv for men and 0.1410/Sv for women). But they admit these numbers might not be right. They also give us a Confidence Interval on these estimates. and they label that interval Subjective 95% CI. In the text they say ``The subjective confidence intervals reflect uncertainty due to sampling variability, transport, and DDREF.” The key word here is subjective. This tells me these confidence intervals were based on the Committee’s judgement, not the product of some statistical test. The Committee is prepared to quantify their subjective feelings about uncertainty, and call it a Confidence Interval. That’s as Bayesian as you can get.

Those 95% Confidence Intervals are quite wide. For men, it’s 0.0490 to 0.1920, a range larger than the mean. For women it’s 0.0740 to 0.2690, even wider. The Committee is honest about the fact that they are quite unsure about the LNT slope.

The BEIR Committee and I are on the same page philosophically. We are both Bayesians, although I’m a little more hesitant to use informative priors than the Committee. But the key part of being a Bayesian is willingness to update your probabilities as new information comes in.

BEIR VII was published in 2006. In 2009, Nair published the results of the 15 year long study of the Karunagappally cohort. Karunagappally is a district in southwest India. The villages right on the beach sit on sands that are very high in thorium. These people experience dose rates up to 0.2 mSv/d their entire lives. Villages farther inland have much lower dose rates. Dr. Nair and his group carefully tracked some 70,000 people for 15 years. Figure 2 summarizes the results.

Figure 2. The results of the Karunagappally Study

As Bayesians, the BEIR Committee needs to update their probability density on the LNT slope to account for the new data. This is a completely mechanical process depending only on their old density, the data, and Bayes Rule. Being the helpful soul I am, I will do it for them.

I assume the Committee’s original density on the combined male/female LNT slope is a normal with a mean at the average of their male and female estimates and a standard deviation of 1/4 their 95% Confidence Interval. I also assume that the process generating Dr. Nair’s data is Normal with an unknown slope and an unknown variance. Notice here I’m assuming LNT is correct, which I’m sure the Committee agrees with. We just don’t know the slope of the straight line.

I concocted a probability density matching the Committee’s belief about the LNT slope prior to see the Karunagappally data, and updated that prior density according to the data and Bayes Rule to get the Committee’s posterior density after seeing the data. This updating process is purely mechanical and non-controversial. I won’t bore you with the details.2

Figure 1 show the results. The red line is the Committee’s density on the LNT slope, before seeing the Karunagappally data. It is centered on 0.11/Sv, that is a 1 Sv dose results in a cancer incidence of 0.11. The old standard deviation is about 0.04/Sv, which shows the Committee was pretty unsure of how accurate their best estimate was.

The blue line is the Committee’s new density on the LNT slope. It is very tightly grouped around -0.12. 70,000 people for an average of 10.5 years is a very large sample. So the data overwhelms the Committee’s wishywashy prior feelings. If you are a committed LNTer, you will be happy to know that the kind of dose rate profiles (DRP) incurred by the Karunagappallians decrease your chance of getting cancer.

I’m not an LNTer. I know its biological nonsense. Nor am I willing to accept Figure 1 at face value. I suspect that something else explains the lower cancer incidence of the higher dose rate profile people. Maybe living next to the sea is beneficial in itself. Maybe the people living next to the sea are wealthier. They have the option of fishing and thorium mining. But if you are an LNTers and a Bayesian as the BEIR VII Committee clearly is, and you attribute the cancer incidence to radiation, Figure 1 is conclusive. You must become a hormetian LNTer.

A portion of mankind is awash in electricity. For these lucky humans, darkness has been banished. They live in homes and work in offices where 70F is too hot in the summer and too cold in the winter. All the hot water they can use is available with a flick of the wrist. All the ice they can consume is available on demand. Food can be stored pretty much indefinitely. Machines wash and dry their clothes and clean their dishes. Electric powered robots are taking over manual task after manual task, freeing up people to do all kinds of mischief. Electricity powered server farms store all the world’s knowledge, millions of terabytes of trivial data, and very little wisdom. Electricity is the foundation of their economies and their wealth.

For these people, it is hard to imagine what life without electricity is like. Currently mankind consumes electricity at a rate of about 3,000 gigawatts(GW). But the distribution is horribly uneven as Figure 1 shows. The USA consumes 1,500 watts per person. The Scandinavian countries considerably more. But most of Latin America is below 250 W. Most of South Asia below 125 W. Much of Africa below 25 W. The national averages mask wide disparities. Nearly a billion humans have no access to electricity at all.

Figure 1. Per human power consumption by region. Width of each box is regional population. Area is regional power consumed.

Things are not getting better. After considerable progress in the first two decades of this century, extreme poverty (income less than 3 international dollars per day) is no longer declining. It has stagnated at around one billion humans. The World Bank now expects extreme poverty to start rising. That means more women cooking over wood and dung fires in confined spaces. That means more kids won’t have the light they need to learn the skills that are essential to breaking the bonds of poverty.

Much of this poverty is in Africa. There are at least 6 African countries where most of the people attempt to live on less than 3 $/day. The numbers are not dropping.

Bright bubbles that tell a tragic story

Figure 4. Life expectancy versus GDP/head. Each circle is a country. The circles hide large intra-country disparities. In the USA, a wealthy country, the difference in years of life expectancy between the top 1% in income and the lowest 1% is 14.6 (men) and 10.1 (women).

Globally the worst health hazard of all is being poor. Nothing else is close. The difference in life expectancy between the poorest and wealthiest humans is measured in decades, Figure 4. Wealth requires electricity. A rough rule of thumb for poor economies is every kWh produced is worth six dollars in GDP, Figure 5.

Figure 5. GDP versus 2024 electricity consumption

Another way to look at this is via the United Nation’s Human Development Index(HDI). The HDI is a somewhat arbitrary combination of longevity, wealth, and education. Figure 6 plots the HDI against electricity consumption. Unfortunately, the HDI tends to squash countries together. Looking at the HDI, you might get the impression that the quality of life in Chad is maybe half that of Norway. The table inset into Figure 1 tells a different story. The top 15 countries consume something like 500 times as much electricity per person as the bottom 15. But Figures 4 and 6 do make a couple of points:

1) For high electricity consumers, the curve is rather flat. These people could get by with a little less electricity without a major impact on their life expectancy.

2) But for the low consumers, the curve is quite steep.

For these people, a little more electricity is literally a matter of life and death. Consume less does not work for them.

3) The size of the circles represents population. An awful lot of the planet’s humans are in the steep part of the curve.

Figure 6. Human Development Index versus per capita electricity consumption.

If mankind is to prosper, then clean, dependable electricity must be available to all. And we must provide this power without polluting the air we breathe, without poisoning the land we live on, and without impacting the climate we depend on.

But here’s the important part, Figures 4, 5, and 6 tell us that electricity must be cheap. Electricity that makes people poorer is a health hazard. 15 cent/kWh (Vogtle 3/4) and 18 cent/kWh (Hinkley Point C) electricity is making people poorer. If you are not cheaper than coal (4-5 cent/kWh), you are a health hazard. Unless you are cheaper than coal, don’t bother.

Why take any radiation risk at all?

About a year after Three Mile Island, the utility, Metropolitan Edison (MetEd) needed to release some krypton-85 from the wrecked TMI2 containment, so that workers could start cleaning up the mess inside. Krypton-85 is primarily an electron emitter with a 10.7 year half-life. Electron emitters have to be ingested or inhaled to go any real damage. Krypton is a noble gas. It does not react with anything. If you breathe it in, you will breathe it out. Thus, the concentration in the lung is the same as the concentration in the atmosphere. The dose coefficient, the ratio between dose rate and krypton concentration, is a tiny 0.0000000000019 mSv/d per Bq/m3. Kr-85 is regularly released by nuclear plants. For this release, no member of the public would get a dose of 0.01 mSv. The LNT Lost Life Expectancy from 0.01 mSv acute is 3 minutes; the SNT LLE is 3.5 milliseconds.

MetEd held a townhall to explain to the locals what they were doing and why. Two of the attendees were the Gormans, who lived about 1.5 miles from the plant. Beverly Gorman and her husband had two children who at the time were teenagers. The Q&A session resulted in this exchange, between the Gormans and MetEd VP Robert Arnold.

Mr. Gorman:

I want a straight answer. How much radiation were we exposed to the first two days after the accident?

Arnold says the estimate is 83 millirems (0.83 mSv) and then attempts to compare that to background levels.1

Mrs. Gorman jumps in:

I dont want any flowery speeches. Can you tell me my children have suffered no physical damage?

Arnold:

We would have to say that your children did receive some radiation from the accident at Three Mile Island. Any radiation could result in some cell damage, whether it is from a school building, the ground where they live. The judgement has to be made if the cell damage is at levels that are acceptable. I don’t know anybody who could decide.

Mrs. Gorman:

You are proposing to expose them to more. I have to live here. We don’t have the money to move away. I don’t think they should be subject to more.

Afterwards Mrs. Gorman was still very upset. Here she is talking to a reporter.

I didn’t get a good answer. I want somebody up there to tell me my children did not suffer physically from the TMI accident. Yes or no. That’s all I want to know. All I hear is rems and millirems. I just want a straight yes or no. It’s always a a long drawn out answer. If I can’t be concerned about my children’s health, then I can let them eat jelly beans for breakfast, and stop worrying about what they eat and how long they sleep. It’s all for nothing if they are sacrificed to nuclear energy.

Here’s what I think Arnold should have said.

Mrs. Gorman, the answer is no. Not only will there be no measurable harm, your children are not being sacrificed to nuclear energy. If TMI did not exist, its dependable electricity would need to be replaced by fossil fuel. That fossil fuel plant will cause far more harm than nuclear. Not producing that electricity would cause still more harm. Nuclear is the cheapest source of electricity. Nuclear is making your kids wealthier than they otherwise would be. And wealth means health. Nuclear means your children will live longer than they otherwise would.

This argument requires that nuclear be cheaper than fossil. If this is not the case, Arnold can’t use this argument, and I do not know what to tell Mrs. Gorman.

Ban and MAD

In the 1950’s and 1960’s, everyone was petrified of the Bomb, and with good reason. The Bomb had been employed just a few years back with horrific effect.1 Current bombs were 100 times more powerful. The US and Russia were locked in an existential struggle, which periodically flared into proxy wars and eyeball to eyeball confrontations. Both sides had thousands of nuclear armed missiles locked and loaded.

There were two schools of thought about how to handle this perilous situation.

a) Ban the Bomb.
b) Mutually Assured Destruction (MAD)

For the Banners, the first step was to stop nuclear weapons testing. They were led by the Rockefeller Foundation whose management felt personally responsible for the Bomb. In their support of theoretical physics in the 1930’s, they had funded almost all the Manhattan Project greats. Worse, they had single handedly funded the cyclotron at Berkeley. This proved critical to the creation of the Bomb. The Foundation was determined to make amends.\cite{fosdick-1945, rf-1956}

The tool they chose was the health hazards of radioactive fallout. Their problem was the dose rates associated with test fallout were a small fraction of the natural background radiation which all of us experience each day, Figure 1.

Figure 1. UK Fallout dose rates. The average background dose rate in Cornwall is 7.8 mSv/y.

The Rockefeller Foundation and its allies decided to argue that radiation produced genetic damage and that damage is unrepairable. Damage just keeps building up. Therefore, the harm is proportional to the total dose, regardless of how rapidly or slowly that dose is incurred. This is like saying taking 365 tablets of aspirin at once is the same as taking 1 tablet per day for 365 days. This no repair hypothesis is called the Linear No Threshold model or LNT.

If LNT is correct, then the Banners could aggregate the tiny increases in dose rate due to test fallout over hemispherical populations and over decades to argue that Bomb testing was invisibly killing millions of people worldwide. The Foundation expertly (and unscrupulously) promoted LNT with all its resources.

The MADers argued that banning the Bomb was politically impossible. Their solution was to make the Bomb so big and so scary that it would never be used. It was also in their interest to see LNT adopted. Their stance boiled down to: radiation is as extremely harmful as LNT claims; but that just means we need to make sure the Bomb is never used by making the weapon too horrible to use. Both sides needed to bash radiation, and bash it hard.

The Atomic Energy Commission(AEC) was in a bind. In 1954, Congress had given the AEC autocratic control over both nuclear weapons and nuclear power. The AEC had to both implement Mutually Assured Destruction, and promote and regulate nuclear power. Their quandary was captured by Karl Darrow, an important member of the Manhattan Project, who wrote to a colleague: “I take it that there are two main objects. One is to please the public with the prospect of beneficial uses of atomic power, and the other is to scare it out of its boots by threatening it with new weapons.”

The Foundation and it allies were successful.2 In 1959, the National Council on Radiation Protection, many of whose members were funded by the Foundation, recommended that radiation protection standards be based on LNT. In the same year, the AEC quietly adopted LNT. LNT became the law of the land.

LNT is biological nonsense

There was one problem. LNT is wrong. Our bodies do have the ability to repair radiation damage. The damage that we should be and are most concerned about is damage to our DNA. DNA is the part of the cell that contains the information we need to function. If that information is altered, a cell can go rogue and turn into cancer. Radiation damage can do just that.

But so can our own bodies. Humans are oxygen breathers. We combine oxygen with other chemicals to produce the energy we need to function. This is called metabolism. Our oxygen based metabolism allows us to do some pretty amazing things. But as a by product, we are constantly ripping up our own DNA. It turns out that our own internal metabolism produces over 25,000 times as many Double Strand Breaks as average background radiation. Double Strand Breaks are the kind of DNA damage that can lead to cancer.

Nature had to come up with a solution to this carnage. Otherwise we would not be here. Her solution was a panoply of usually very effective repair processes. Google ``DNA Repair’‘. Go ahead do it. You will get around 130 million hits. DNA repair has been studied in mind numbing detail. Three Nobel prizes have been awarded in the field of DNA repair. LNT’s foundational premise that radiation harm is unrepairable is flat wrong. This is indisputable.

Figure 2. Super Simplified Diagram of DNA Repair Mechanisms

Radiation is a bit like aspirin. If we get a whole bunch of radiation at once, our DNA repair processes are overwhelmed, and we see an increase in cancer incidence. However if the radiation dose rate is not too high, our repair processes can keep up with the damage. In that case, we can detect no increase in cancer, even when that dose rate is incurred for decades.

But why take any radiation risk at all? That’s an easy one. Humanity must have cheap nuclear power to flourish. That cheap electricity can make all humanity wealthier. The worst health hazard of all is being poor. Nothing else is close, Figure 3. Wealthier is healthier. By making nuclear power far more expensive than it should be, LNT is killing people.

Thanks to hundreds of millions of dollars of research, we have a pretty good idea what the dose rate is at which we start seeing an increase in cancer. Because our DNA repair processes must handle internal damage rates that are many thousands of times above background radiation damage rates, that dose rate is well above the dose rate that almost any member of the public will receive in even a release as large as Fukushima. In fact, we have detected no increase in cancer at Fukushima.\cite{unscear-2021}[p 106] Even at Chernobyl, the perfect experiment, the thousands of deaths confidently predicted by LNT have failed to materialize.\cite{leung-2019, zupunski-2021} We have detected no radiation induced harm to the public, other than thyroid cancer due to children drinking badly contaminated milk.3 This could easily have been prevented by confiscating the milk, as was done at Fukushima. In short, thanks to the remarkable DNA repair processes a providential Nature has bestowed on us, even a very large release will produce little or no detectable radiation harm to the public. The planet would be a far better place, wealther and healthier, with cheap nuclear and an occasional large release, then without both.

While LNT’s unrepairable predictions are roughly correct when a large dose is received all at once, LNT over-estimates the cancer incidence by factors of a 1000 or more when the same dose is spread over months and years, as it is in a nuclear power plant release. This has caused the radiophobia that both the Banners and the MADers intended. As a result we have turned control of nuclear power over to an autocratic, all powerful regulator who has been told a big release is an intolerable catastrophe. His overriding priority is preventing a release. The result is the current does-cost of nuclear power in the West is five or more what it did and can cost.

Nuclear’s Did and Can Cost

Figure 4. The 2500 megawatt Oconee plant in South Carolina. These three reactors were built for just over 350 million dollars between 1967 and 1974. That’s $1141 per kilowatt in 2024 dollars. They took about 6 years to build. Oconee can produce reliable, on-demand, zero pollution, very low CO2 electricity at less than 3 cents/kWh in today’s money. Oconee’s average capacity factor over the last 5 years was 98.2%. Oconee and its cooling pond Lake Keowee have turned a depressed part of western South Carolina into a second home and tourist magnet.

In the 1960’s, fission’s 100,000 to 1 advantage in energy density allowed nascent nuclear power, which was just starting down a steep learning curve, to go head to head with coal. In 1965, GE was able to show TVA that it would produce electricity for less than 3.7 mills per kilowatt hour.\cite{bupp-1978}[page 90] That’s about 3.2 cents per kWh in 2024 dollars. Komanoff, no friend of nuclear power, estimated nuclear’s 1971 CAPEX at 366 1979 dollars per kW, and coal without scrubbers at $346/kW.\cite{komanoff-1981}[p 20] Nuclear’s fuel cost advantage tipped the cost of electricity in favor of nuclear. In 1970, Paul Ehrlich, a determined foe of nuclear on Malthusian grounds, complained ``Contrary to widely held belief, nuclear power is not now `dirt cheap’. ... At best, both [nuclear and coal] produce power for approximately 4-5 mills per kilowatt-hour.”\cite{ehrlich-1970}[p 57] In the late 1960’s nuclear power was as cheap as coal, when coal was as cheap as it ever was in real terms.

Figure 5. Overnight (no construction period interest) Cost, USA Nuclear Power Plants, adapted from \cite{lovering-2016}

This did not last long. Figure 5 shows the Overnight Cost of USA nuke plants built between 1950 and 1980 in 2024 dollars. The date shown is the date that construction started. Almost all the plants started between 1964 and 1968 came in at under $2000/kW in 2024 dollars. Six of those plants came in at under $1000/kW. See text in Figure 4. Five of them are still operating, and are licensed to the mid-2050’s. These early plants have been producing some of the cheapest electricity ever generated for more than 50 years, harming exactly zero members of the public in the process.

But by the mid-1970’s, the same size plants had an overnight cost of up to $10,000 2024 dollars and more with a very large scatter. Overnight costs are only a portion of the CAPEX. At the same time as overnight costs were skyrocketing, build times were doubling and tripling. Plants that took 4 years to build in the late 1960’s were taking 10 or more years in the late 1970’s. The 1970’s was a period of inflation and high interest rates. Construction period interest expense skyrocketed.

Nuclear quickly went from dirt cheap to prohibitively expensive. Few nuclear plants were ordered in the US after 1974, and all of those were canceled. This was five years before Three Mile Island, and in a period when public support for nuclear was more than 70%.

What happened? Through the 1950’s and early 1960’s, the real cost of oil was declining, reaching its all time low in 1967. The Majors were buying oil in the Mideast for a penny a liter. This forced coal prices down, and real coal price was also at an all time low in 1967, Figure 6.

Figure 7. Fossil prices and USA nuclear overnight CAPEX. The oil price index is a little misleading. It’s an average of prices at the loading ports. The landed price jumped earlier due to the tanker market boom in 1967. You can see that in the US coal prices.

But in June of 1967 the Six Day War closed the Suez Canal. The extra ton-mile demand for tankers sent the tanker market into boom, and the landed cost of oil in the West doubled. The Canal closure showed the utilities how vulnerable oil was. They started scrambling to build coal (and nuclear) plants. It was a smart move; but in the next few years fossil was hit by one blow after another. The nationalization of Libyan oil in 1969 led to leap frogging oil prices, culminating in the Oil Crisis of 1973 brought on by the Yom Kippur war. At the same, coal was blasted with a slew of new pollution regulations. The unions piled on as the UMW recognized labor now had massive bargaining power. Coal suffered a series of costly strikes, both union and wildcat.

All this should have been great news for the new guy on the block, nuclear. Nuclear was insulated from both oil price and fossil pollution regulation. But nuclear’s costs rose in lock step with coal’s. Bupp and Derian’s Light Water is the standard history of early American nuclear. These authors were befuddled:

Coal seemed to be just competitive with nuclear power from light water reactors at about 25 to 30 cents/mbtu in 1970; it still seems to be competitive at about four times that price in 1976.\cite{bupp-1978}[page 97] [Emphasis in the original.]

They need not have been. In 1967, a new omnipotent player had emerged. In 1954, Congress had given the AEC complete and unfettered control over nuclear, both nuclear weapons and nuclear power. As Truman put it, atomic power was “too important to be made the subject of profiteering”.\cite{ford-1982}[p 41] The AEC had to both implement Mutually Assured Destruction, and promote and regulate nuclear power.

Prior to 1961, the Nuclear Power Regulatory Division was just another box in the AEC organization chart, reporting to the General Manager, who had to balance nuclear power’s risk and cost in exercising the AEC’s autocratic power. Here’s how Irvin Bupp described this period.

At first, the AEC simply never took the job of regulating nuclear power seriously. Regulatory policy was explicitly based on the peculiar proposition that because nuclear technology was so obviously hazardous, the companies manufacturing and purchasing it would regulate themselves. \cite{bupp-1985}[p 1968]

No, Irvin there was nothing “peculiar” about this system. This is the system that gave us the Turkey Points and the Oconees. If the plant owners have to pay for any harm they cause, they will find the right balance between cost and risk

But in 1959, the AEC had quietly adopted Linear No Threshold(LNT) hypothesis, a radiation harm model that claims our bodies cannot repair damage to our DNA, a claim that denies indisputable biology. The AEC did this in part because it heightened fear of the bomb. But LNT also massively magnified the fear of a nuclear power release. In 1961, the Regulatory Division was split off from the rest of the AEC. It now reported directly to the Commissioners. The Division’s job was to prevent a release. Period. Cost was not a consideration.4

The only limit to this unbridled, dictatorial power was the cost of fossil. If the Division pushed the cost of nuclear above the cost of fossil, license applicants would disappear and the Division would have nothing new to regulate. After 1968, it would also lose the revenue from the application review fees, on which it was dependent. As the cost of coal rose, the Regulatory Division quickly pushed nuclear’s cost up against the new limit.

With the doubling and tripling in coal prices between 1967 and 1975, the utilities’ goal became do whatever you have to do to get nuclear plants built. Rather than pushback against new regulations, no matter how silly or superfluous, and engage in time consuming negotiations, the utilities decided to accept anything they thought they could pass on to the ratepayer.

Congress has explicitly given the AEC not only the power to make the rules but to change those rules as it saw fit. Regulatory uncertainty made fixed price contracts impossible. The last turnkey contract was signed in 1968. But there was an out. In this period, all the US utilities were regulated monopolies. By signing cost plus contracts, the vendor transferred the regulatory risk to the utility, and the utility transferred the risk to the ratepayer. A regulated monopoly is a system in which the more money the utility spends, the more money the investors make. The incentives are all wrong. Costs and build times explode. But the regulators are happy; the vendors are happy; the investors are happy; and the ratepayers are screwed.

Over time most of the Western countries have adopted something similar to the American system. The results have been predictably horrible. Current newbuilds in the West cost more than $15,000 per kW (ten times the cost of Oconee and her sisters in real terms) and have construction times of more than ten years. The strike price for Hinkley Point C is an impoverishing 133 pds/MWh (18 cents/kWh). By denying our ability to repair DNA damage, we have effectively rejected the blessings of nuclear power.

Figure 8. Hinkley Point tombstone and recent US and European costs and build times. The inscription is wrong. Human welfare is our overriding priority. Suppose you told a football team: Guys, your overriding priority is not getting hurt. If any of you get hurt or do anything that might get you hurt, you are out of football for life. Now go out there and win this game. How well do you think that team would perform?

Enlightenment and Action

1) Nature has provided us with a source of electricity with a energy density that is 100,000 times that of fossil. A source that requires little land, and produces almost no pollution and no CO2. A source that was and can be cheaper than coal, much cheaper. It’s called nuclear fission.

2) Nature has provided us with a DNA repair system that can handle the unique hazard associated with this source. She had to do this to protect us from our own oxygen based metabolism.

You can call this double gift, providential. You can call it, anthropic. You can call it, dumb luck. But what you don’t do is reject it. By denying the second gift, humanity has rejected the first. This is an act of auto-genocide.

In Hindu theology, Lakshmi is the Goddess of wealth and prosperity. Ganesh is the God of wisdom and the remover of obstacles. The basic idea of pairing them, as I understand it, is receiving a fortune doesn’t mean much unless we have the knowledge to use it wisely. I think all of us can agree there is a lot of wisdom in that precept.

It is certainly true when it comes to nuclear power. Nature provided us with a source of electricity that would have been unimaginable 100 years ago. That source has an energy density 100,000 times larger than fossil fuel. That means it requires very little of the planet’s precious resources. In a syllable, it can be cheap. And in fact nuclear power was cheap, 3 cents/kwh cheap in today’s money when it was still moving down a steep learning curve in the late 1960’s. On top of that, fission power uses up little land, and produces almost no pollution and very little CO2. It is hard to imagine a more bountiful gift.

And that’s only half the blessing. Fission like electricity and high pressure steam and fire itself comes with a unique new hazard: RADIATION. But Lakshmi, Nature, evolution --- I don’t care what we call it --- doubled down. She not only gave us fission; she also provided us with a DNA repair system that can easily handle dose rates hundreds of times above background, dose rates that are unlikely to be exceeded by any member of the public in even a very large release.

What she did not give us was the knowledge to use this double gift wisely. In fact, due to the ominous way fission power came to us, we chose to massively exaggerate the harm associated with radiation, in order to decrease the probability of the Bomb being used again. This led to the Intolerable Harm Lie, a big release of radioactive material would be so catastrophic that it cannot be permitted to happen.

Therefore, we gave control of nuclear power to an all powerful, purblind regulator, whose overriding priority is preventing a release. In the NRC’s world view, making people poorer is unimportant. Global warming is not their business. Harm from alternate sources of electricity is not a consideration. Evacuation and exile costs can be disregarded. The only thing that counts is radiation harm, based on a harm model that is biological nonsense.

As a result nuclear power in the West now costs five or more times what it can and did cost. What should be our cheapest source of electricity is impoverishingly expensive. By denying our indisputable ability to repair DNA damage, we have effectively rejected the blessings of nuclear power.

That’s where Lord Ganesh must come in. We need him to give us the wisdom to recognize and accept the second half of the blessing. But we need more than wisdom. We need action. Ganesh is depicted holding a lotus and an axe. The lotus represents enlightenment. In this case, the recognition that we have been lied to and the rejection of that lie. The axe represents Ganesh’s role as the remover of obstacles. Sixty years of universal acceptance of the Intolerable Harm Lie has created a coven of very powerful special interests whose existence depends on that Lie. We will need a very big axe to remove those obstacles.

For the full story and replacements for both LNT and the current auto-genocidal regulatory system, get the Humanity has been Twice Blest book.

Here are my predictions. The response will not say that LNT is biological nonsense. It will not acknowledge that LNT is the no DNA repair theory whose foundational premise is contradicted by well established, indisputable biology. It will not point out that the radium dial painters and the Taipei apartment dwellers clearly falsify LNT.

The response will embrace the Muller Trap and accept the false LNT or no threshold dichotomy with something like INL’s:

However, it is important to stay focused on the central issue with LNT: that all radiation exposure, without threshold, will create deleterious health effects.p[3]

It will offer the deeply flawed INWORKS study as support for LNT, and ignore the Life Span Study (LSS) of the bomb survivors on which the Japanese and American taxpayers have spent nearly a billion dollars over 70 years, despite the fact that until recently the LSS study was the NRC’s Goldstandard. What changed was the Radiation Effects Research Foundation admitted their LSS data is non-linear. It will point out that the LSS exposures were acute dose rate profiles, but LNT claims there is no difference between acute and chronic profiles. It will also ignore the Nuclear Shipyard Workers Study by far the best controlled of the nuclear worker studies.

It will not say what counts is not the cumulative dose, not the average dose rate, but the entire dose rate profile. In fact, it will never use that term. Instead it will repeatedly talk about the “dose”, implicitly accepting LNT. There will be no mention of DNA repair time. What recommendations it comes up with will be based on the biologically meaningless and medically dangerous period of a year.

It will not offer a well defined replacement for LNT. Therefore, the NRC has no way of converting an acceptable risk to dose rate limits. Despite this, it will call for some relaxation of the current limits, probably 50 mSv/y for nuclear workers and 5 mSv/y for the public, without offering a compelling argument for these numbers.

It will make a few handwaving generalizations about not requiring ALARA. But since it offers no replacement for LNT, there will be no way to do the benefit-cost analysis needed to actually tell the regulator where to stop. Whether or not ALARA is dropped in name, regulator incentives will remain unchanged and so will the regulator’s behavior.

There will be no discussion of the deadly downside of LNT, the unnecessary evacuations and exiles, the millions of lives shortened by fossil pollution, and the failure to raise billions out of poverty.

The response will end up neither endorsing nor repudiating LNT, but call for more study, preserving the status quo.

I have had no contact with the people drafting the reconsideration. This will be an interesting test of your preacher’s infallibility.

In Hindu theology, about which I know very little, Lakshmi is the Goddess of wealth and prosperity. Ganesh is the God of wisdom and the remover of obstacles. The basic idea of pairing them, as I understand it, is receiving a fortune doesn’t mean much unless we have the knowledge to use it wisely. I think all of us can agree there is a lot of wisdom in that precept.

It is certainly true when it comes to nuclear power. Nature provided us with a source of electricity that would have been unimaginable 100 years ago. That source has an energy density 100,000 times larger than fossil fuel. That means it requires very little of the planet’s precious resources. In a syllable, it can be cheap, dirt cheap. And in fact nuclear power was cheap, 3 cents/kWh cheap in today’s money when it was still moving down a steep learning curve in the late 1960’s. On top of that, fission power uses up little land, and produces almost no pollution and very little CO2. It is hard to imagine a more bountiful gift.

And that’s only half the blessing. Fission like electricity and high pressure steam and fire itself comes with a unique new hazard: RADIATION. But Lakshmi, Nature, evolution --- I don’t care what we call it --- doubled down. She not only gave us fission; she also provided us with a DNA repair system that can easily handle dose rates hundreds of times above background, dose rates that are unlikely to be exceeded by any member of the public in even a very large release.

But what she did not give us was the knowledge to use this double gift wisely. In fact, due to the ominous way fission power came to us, we chose to massively exaggerate the harm associated with radiation, in order to decrease the probability of the Bomb being used again. This led to the Intolerable Harm Lie, a big release of radiative material would be so catastrophic that it cannot be permitted to happen.

Therefore, we gave control of nuclear power to an all powerful, purblind regulator, whose overriding priority is preventing a release. As a result nuclear power in the West now costs five or more times what it can and did cost. What should be the cheapest source of electricity is impoverishingly expensive. By denying our indisputable ability to repair DNA damage, we have effectively rejected the blessings of nuclear power. We must renounce that lie.

That’s where Lord Ganesh needs to come in. We need him to give us the wisdom to recognize and accept the second half of the blessing. But we need more than wisdom. We need action. Ganesh is depicted holding a lotus and an axe. The lotus represents enlightenment. In this case, recognition that we have been lied to and the rejection of that lie. The axe represents Ganesh’s role as the remover of obstacles. Sixty years of universal acceptance of the Intolerable Harm Lie has created a coven of very powerful special interests whose existence depends on that Lie. We will need a very big axe to remove those obstacles.

A PDF of the Twice Blest poster can be downloaded from the Gordian Knot Group website.

Nuclear promoters and their dupes are spouting all kinds of nonsense about factory built reactors. Factories won’t cut it. If you are talking factories, you are talking tiny or you are talking components. Consider NuScale, the current leader in sucking up our money. What actually gets built in a factory is a 23 meter tall chimney, Figure 1. Because of the chimney’s low power density and abysmal efficiency, we need 12 chimneys to produce 977 MWe of steam.

Figure 1. The NuScale chimney. Don’t be fooled. This thing is 23 m (75 ft) tall.

These 12 chimneys sit in an immense, site built swimming pool, Figure 2. The overall structure is larger than the horribly overbuilt EPR fission island which produces 64% more power. And we have not even begun to talk about the turbine hall. If NuScale is factory built, then so is the AP1000 whose components are built in a factory. Both designs cost over $15,000/kW, about ten times what the 1960’s built plants’ cost in real terms. Technological advances in the last 60 years should have cut the real costs by at least a factor or two.

Figure 2. The NuScale swimming pool. The little white squares are the chimneys. The pool is about 125 meters (410 feet) long, 55 meters (180 ft) wide, and 60 meters (197 feet) deep. It is not factory built.

If we are going to build nuclear power plant off site in a controlled, assembly line fashion, we need a shipyard, or more precisely a nuclear plantyard. A plantyard would produce 5000 ton superblocks. These superblocks, perhaps 15 per 1 GWe plant, would be loaded onto barges and transported to the site, or better combined together at the yard into two or three islands. The islands would be floated onto heavy lift vessels or towed to a shoreline or near shore site, and ballasted onto the bottom or moored.

So what would it take to build plantyards in the USA. Here are some thoughts.

Enact the Nuclear Reorganization Act

An absolutely necessary condition of success is replacing the NRC with a balanced regulatory system based on a realistic view of radiation harm. As long as nuclear is controlled by an omnipotent regulatory system based on the false premise that a large release is intolerable, there can be no real progress. If a large release is intolerable, then there should be no nuclear. We must also avoid the American tort system. I will not rehash all the arguments here. Check out the Twice Blest book. If we don’t pass the NRA, there’s no point in even thinking about American plantyards. But passing the NRA will not be sufficient.

Don’t obsess about wage rate, but we must have labor discipline.

Wages are close to a non-factor. We have this image of a shipyard as a human beehive, hundreds of men swarming over a hull. In fact, a modern yard can be a lonely place. I’ve been in a panel line hall that stretches for maybe 400 meters. I was nearly alone. Direct labor is about 15% of the cost of the ship. Labor productivity is much more important than wage rates.

Samsung can build ships in Korea at less than one-fifth US yard cost. When Samsung was thinking about resurrecting the defunct Avondale yard near New Orleans, they were surprised to find that wages on the Gulf Coast were less than what they were paying at home. But the US regulatory maze and uncertain labor discipline sent them home in a hurry.

Korean yard labor has very strong unions. The average fully built up wage is in the neighborhood of $40/hr. Strikes are rare but, when they do occur, they are, to put it politely, militant. You do not want to be a scab trying to cross a Korean yard picket line. The unions sign 5 year contracts. Once they sign a contract, they live up to it. The unions closely monitor yard economics. Every 5 years they make sure they get their cut.

One result is, if somebody is not pulling his weight, his union buddies let him know that’s unacceptable. And if that doesn’t work, a union steward is brought in to explain to him that by hurting yard productivity, you are costing everybody money. Deadbeats either get the message or are removed from the yard work force, one way or another. The result is remarkable productivity, and surprisingly little resistance to automation as long as it is gradual. We must foster a similar system.

Control EPA, OSHA, and the ambulance chasers

Here’s Nathan Drake, who was Manager of Environmental Compliance at the NASSCO yard in San Diego.

Nowadays environmental compliance is as integral to shipbuilding as welding, and I say that with a straight face. The regulatory landscape is so restrictive that in some other yards, every gallon of paint, every pound of welding wire/rod, each hour that a diesel compressor runs, must be accounted for daily. I don’t say this lightly when I say that some facilities are facing existential risks for up to a quarter of a pound of hexavalent chromium emissions over an entire year.

The NRA would have to be expanded or supplemented by a companion act to bring the plantyards under the same regulatory system as the plants themselves. NEPA must be waived, and any disputes between the EPA and the yard adjudicated by the Nuclear Arbitration Board. Fines must be based on actual emissions, not paperwork.

Ditto for OSHA. As for the ambulance chasers, it’s Worker’s Comp and that’s it.

Nuclear plantyards, not shipyards.

While the plantyards would use much the same technology as a shipyard, they would be focused on a totally different market. Shipbuilding like all mature, truly competitive businesses is highly cyclic. In the US, at the first slump, the yards turn to their sugar daddy, the US Navy, adopt Navy (aka NRC style) controls and systems and are ruined forever.1 If we can start building nuclear plants at near should-cost, it will be 50 years before the first big nukebuilding slump. Think the great railroad boom of the second half of the 19th century.

Multiple yard competition

Having said that, we must have competition between the yards. We need that competition to push nuclear down to its 1960’s should-cost, and then start pushing the should-cost down. That means multiple yards. When the Koreans decided to compete with the Japanese in shipbuilding they did not set up a flagship yard. They made sure they had at least three big yards and made sure those yards had to compete with each other on an even basis. That competition quickly forced all the Koreans yards to improve both cost and more importantly quality. It wasn’t long before the Koreans surpassed the Japanese in both and then kept improving.

Korea in the 1970s was cash starved. This is not the case in the US. We don’t need public funding. We just need an even playing field, the right rules, and assurance the rules won’t change. The money will flow in.

Abandon concrete.

Structural concrete and shipyard production do not go together. Yard built plants would need to be all steel. Concrete would be relegated to a few non-structural tasks such as turbine foundation mass.

This will require us to drop the fantasy that we can protect a nuclear plant from modern weapons with concrete. The only defense against a precision, bunkerbuster is a realistic understanding of radiation harm and adequate buffer zones.

Do not attempt to use the current yards

The management and much of the labor force of the current big American yards is beyond redemption. They are infected by a variant of the same disease as the US nuclear establishment. A shipyard is cheap relative to the value of its output. In 2014, Samsung was considering building a greenfield yard in Indonesia. The project’s projected cost was 950 million dollars. The yard could turn out 500,000 light tons per year worth at least two billion dollars. Yard CAPEX is less than 15% of the cost of a ship. A plantyard will be simpler and turn out a higher value product.

By going green field, we get not only the latest robotic technology; but we can tailor the yard to what a nuclear power plant needs. For one example, nuclear plants, except for pressure vessel heads which can be formed in a hot press, need not require any double curved steel. But unlike most ships, the plants may require very large cylinders. Just as importantly, we can train a new work force. No unlearning of bad habits will be required; and no mass firings of people who think stick welding is an advertisement for yard productivity, Figure 3.

Figure 3. In my 4 years in Korean yards, I can’t recall ever seeing any stick welding.

Bring in the Koreans

We must have the Korean yard production system. Think Toyota in North Carolina. In the yard system, detailed design not only does the working drawings, but just as important the production scheduling down to per shift detail. This includes scheduling each sub-block and block lift by crane. The weight and center of gravity of each lift is calculated and the lifting lugs are part of the design. Scaffolding which will be required in final erection is installed at the block level. Detailed design and production scheduling cannot be separated.

I’m envisioning three yards designed and managed by Hyundai, Samsung, and Hanwha respectively. Let the Koreans bring in as many people as they want, and let them stay as long as they want. Get Westinghouse completely out of the picture. The Westinghouse/KEPCO “deal” must be abrogated.

No tariffs. No forced localization. No subsidies.

The yards must be free trade zones. They must be able to bid anyone on the planet on an even playing field. About 60% of the cost of a ship is purchased material. Korean yard buyers scour the world with Walmart like zeal. Finally no subsidies. Given nuclear’s inherent cheapness and the need, if the plantyards require subsidies, we haven’t set the rules right.

The Shortcut

That’s a long and very daunting list. There is a short-cut. Build the plants in Korean yards and import them. But we would still need to pass the Nuclear Reorganization Act.

Figure 1. No. The welfare of humanity is our overriding priority

Sticks and carrots

People are always pestering me for concrete examples of the cost of NRC style regulation. I turn them off by telling them it doesn’t work that way. It’s about incentives and motivation. What are the sticks and what are the carrots? Suppose you tell a football player, your overriding priority is not getting hurt. If you get hurt or do anything that might get you hurt, you are out of football for life. Now go out there and win this game.

When you go from a competitive environment where it’s build better/cheaper or die to an environment which is ruled by an autocratic regulator’s goal to prevent a release, everything goes to hell. Paperwork and process trumps substance erecting massive barriers to entry. Cheap becomes “unsafe”.

Everybody’s motivation gets wrong headed, not just the regulators. Incumbents work harder on protecting and deepening their artificial paperwork moat than they do on their product. Workers forced to follow ridiculous, wasteful procedures and sit around waiting for a series of sign offs on an obvious fix rationally decide if the bosses don’t care about doing the job right neither should they. Fixed price contracts become infeasible. The vendors’ goal becomes milk each project for as much money as possible for as long as possible. And the next thing you know, plants take three or four times as long to build as they should and cost five or more times what they should.

My inquisitor walks away shaking his head and saying to himself, where are the facts, he doesn’t have a real argument.

The Thought Experiment

Sometimes I try the thought experiment. Every year about 30 people are killed in Korean shipbuilding. In my four years in Korean yards, I never saw anybody doing something that I would regard as unsafe, or wouldn’t do myself, and I’m scared of heights. The yards have a complete set of safety rules. Our inspectors got in trouble on several occasions for breaking them. But working at height, around all sorts of moving machinery --- the yard is a cacophony of horns and beeps and bells --- with cables and sparks everywhere, and continually lifting heavy loads is an inherently dangerous enterprise. Bad things happen.

Suppose the Korean government decides this is intolerable. One death is one too many. They would set up an all powerful bureaucracy, the Office of Shipbuilding Deaths (OSD), whose overriding goal is preventing yard deaths and injuries. The OSD would write up a set of rules and requirements with which the yards must comply in order to get a license to build ships. The OSD would require each yard to apply for a license by documenting in detail how they will comply with all its rules and regulations. The yards would switch their best talent to writing these license applications and the scores of supporting documents. Their very best designers and engineers would move to other industries. They would be replaced by lawyer types, skilled in weasel wording and the manipulation of bogus probabilities.

If the government is really serious about eliminating shipyard deaths, they would even allow the OSD to change the rules on the fly. Fixed price contracts and guaranteed delivery would become a thing of the past.

The OSD would deploy inspectors to each yard with the power to stop work if the inspector saw anything he thought violated the license, which includes such amorphous requirements as As Safe As Reasonably Achievable (ASARA), which he has autocratic authority to interpret. Job after job would come to a halt while the yard came up with new procedures, and then got them approved after lengthy negotiations.

The OSD would quickly figure out that the main cause of yard deaths and injury is all the lifting that is going on. They would designate transporters as the Best Available Technology (BAT) for moving stuff around. Heavy lifts whose hoist height exceeded 30 cm, would be outlawed. The yards would be required to build shallow ramps to move the blocks down into the building dock.

I know what you are thinking. This time you’ve gone too far. You’ve ruined your argument by getting silly. Maybe, but this was precisely what happened in the UK with the ABWR. Boiling water reactors have elevated spent fuel pools. To remove the cooled elements, they must be lowered down to ground level. But the Office of Nuclear Regulation has a lift limit of 30 cm for the transport casks, which was based on PWR requirements. One of the solutions that was proposed was a 230 m long ramp. I’m not making this up.\cite{fingleton-2025}[p 58] After a long squabble a compromise was reached.

Even without the ramps, the finely tuned machine that is a Korean shipyard would be destroyed. The costs of the ships would skyrocket and the quality deteriorate. But when I try this approach, my doubting Thomas walks away saying “The Koreans would never be that stupid”. I don’t even have a chance to say “Yeah, but we ...”.

A Concrete Example

It seems we must have a concrete example. Let’s go with the cost of dry cask storage. Under NRC regulation, each dry cask costs about a million dollars. Each cask requires about 30 tons of steel and 30 tons of concrete. Remember the spent fuel has already aged 4 years. Most of the the volatile stuff is gone. We are going to put the casks in the ground or in a vault. If there is a cask failure, nothing much happens,

Using standard construction cost factors, we are talking at most $50,000. Since we can produce these on a repetitive, automated, assembly line basis, we should be able to beat that. But we do need some neutron absorber (plastic) and some inert gas, so maybe we end up at a bit more than $50,000. Indeed Jim Hopf’s company was profitably making dry casks to NRC requirements for about $90,000 in the late 1990’s, before they were black flagged by the NRC for using competent vendors that, unlike incompetent Shaw Industries, did not have an N-stamp. So let’s very generously estimate a dry cask’s should-cost at $100,000. You can have a fancy car for that price.

Figure 2. Histore Pad. 580 tons of spent fuel per acre

Table 1 compares the overall cost of dry cost storage for a $100,000 cask with that for a $1,000,000 cask. This is based on the Holtec Histore system, Figure 2, and Holtec numbers for the non-cask costs of the system. a real discount rate of 5%, a cask life of 40 years, and storing the spent fuel forever. The $100,000 cask system cuts storage cost by a factor of 3.

Thanks to nuclear’s insane energy density, dry cask storage is cheap per kWh. Even at NRC’s ridiculous cask cost, it’s only about 1% of nuclear’s did-and-can-cost of around 3 cents/kWh. Despite this, the NRC which has pushed nuclear’s does-cost up to around 15 cents/kWh with measures that have had no measurable effect on the probability of a release has chosen to put off this tiny cost for about 20 years by using dense packing.

Figure 3. Open-racking left; dense-packing right.

Dense packing means that, if the spent fuel pool is drained, which could happen by simply losing pool cooling, then there is a excellent chance we will have a spent fuel pool fire and a very large release. This almost happened at Fukushima, vonHippel and Schoeppner have computed some of the plumes associated with such a release from the Peach Bottom plant near Baltimore in different weather patterns, Figure 4.\cite{vonhippel-2016} Under LNT and EPA rules, millions would be unnecessarily evacuated. The resulting chaos would make Fukushima look like a picnic.

This could be avoided by going back to open-racking where the fuel elements are spaced so that air cooling is sufficient to keep the fuel elements from bursting and catching fire. That was the AEC/NRC’s original plan. It was a good plan.

Figure 4. Cesium-137 ground contamination from a postulated dense-packed spent fuel pool fire at the Peach Bottom plant north of Baltimore. The figure shows four different weather patterns.\cite{vonhippel-2016} The red area would have to be evacuated under EPA recommendations. This would be a tragedy of epic proportions. 1.0 MBq/m2 is far below the level at which there will be any detectable harm.

Why would the NRC abandon open-racking for dense-packing? At least part of the reason was bureaucratic infallibility. Originally dense-packing was just a short term fix to a temporary problem. When it morphed into a universal, standard operating procedure, NRC was faced with

a) admitting they had screwed up and allowed an unnecessarily risky situation to develop, or
b) claiming that it’s all part of the plan.

Deskpots are never wrong. It had to be (b).

Another big contributing factor is the Price Anderson suicide pact. Under Price-Anderson, the costs of a really big release are spread over all the US plants. If an individual plant decided dense packing is crazy, and paid for open-racking, he would still be on the hook for a big spent fuel pool fire at any of the other 95 plants. He sees only 1/96th of the benefit of his investment.

But an important factor is that getting rid of open racking is at least three times as expensive as it should be because of the paper work required to build a simple steel and concrete can.

In short, NRC style regulation is the reason every nuclear power plant in the US uses dense-packing, which is by far the most likely, non-weapons path to a Chernobyl or larger sized release in the USA. Such a release coupled with the NRC’s immoral, indefensible defense of LNT will cause panic, evacuation, and exile that will shake the country to its roots. Is that a big enough cost for you?

As we begin to realize that global warming is almost certainly not an immediate existential problem, people like Greta, who must have an immediate existential problem, have moved on to other more immediate concerns. There has been a collective sigh of relief. But the fact that even the IPCC has admitted that the infamous RCP8.5/SSP5-8.5 scenario on which so many apocalyptic claims were based is not going to happen does not mean that the long run problem has gone away.

We are putting more CO2 into the air than ever before, Figure 1.

Atmospheric CO2 is still rising. After a crazy jump in 2024, atmospheric CO2 rose by ‘only’ 2.1 ppm, Figure 2. The decline in the increase was partly due to an increase in CO2 absorption by the planet.

Figure 2. Growth in atmospheric CO2 1960 to 2025.

In 2025, the global average temperature dropped slightly from 2024’s record breaking number, Figure 3; but that just means we are going through an La Nina/El Nino oscillation, in which the ocean soaks up and then releases heat. The long term trend is clear. Global warming is very much with us.

Figure 3. Global average temperature, 1940 to 2025

Figure 4. One set of estimates of sea level rise.\cite{zheng-2026}

My main concern is sea level. Sea level rise is accelerating. There’s considerable uncertainty in the numbers complicated by changes in land level and the switch from tidal gauges to satellites. But most sources put the annual rise in the first half of the 20th century around 1.4 mm per year. By the end of the century, sea levels were rising at about 3 mm/y. Current numbers are in the 3.8 mm/y range. Recently the acceleration has been around 0.8 mm/year per decade. If we heroically assume that the acceleration stabilizes at this level, we get the red line in Figure 5. A 2 meter (6.6 ft) rise by 2200 would be a very expensive problem.

Figure 5. Sea Level Rise at current and no acceleration

Figures 1 and 2 make a mockery of Net Zero. There is no way we are going to stop putting CO2 into the atmosphere. The digression into wind and solar has been an impoverishingly expensive flop. Wind and solar can nibble at the edges in certain areas; but the only way we can put a real dent in global warming is a whole lot of cheap, 24/7, low CO2 electricity everywhere. The only way that can happen is nuclear at its did and can cost of around 3 cents/kWh in today’s money.

Even if we came to our senses tomorrow and stopped claiming via LNT that our bodies can’t repair radiation damage, accepted the dual blessing of DNA repair and a source of electricity whose energy density is 100,000 times better than fossil, the absolute best we could hope for is to cut our CO2 emissions by 50% by 2100.

This would require cheap nuclear pushing fossil out of electricity generation (25%) except for a bit of peaking and backup, electrifying all building heating (6%) and light road transportation (8% battery powered vehicles only cut life cycle emissions by about 2/3rds) and using nuclear to produce green or turquoise hydrogen for ammonia (fertilizer)(2%), hydrocracking(2%), and steelmaking(6%). And that’s the easy half.

The temperature would keep rising. A lot of future ice melt is already baked in since it takes a very long time for a huge chunk of ice to melt. (That’s one of the reasons global warming is not an immediate problem.) The time for the Greenland and Antarctic ice sheets to equilibrate to a temperature rise can be millennia. I’m guessing the best we can do over the next 177 years is zero acceleration. I’ve plotted that assumption on Figure 5. It leads to a mean sea level rise of about 0.75 m (2.5 ft) in 2200. That’s a lot more manageable than 2 meters, but will still be a very big problem in a lot of places. And that may be the best that can happen. Finally, there’s no reason to expect sea level rise will magically stop in 2200. In 2200, the annual rise is 18 mm/year in the constant acceleration scenario.

My point is just because some unbalanced prophetesses and opportunistic charlatans have decided that global warming is not happening fast enough to meet their psychological or financial needs does not mean that the physics have changed. My Social Cost of CO2 has not changed and neither should yours.

In early 2023. the Gordian Knot News published a piece entitled How much time do we have? arguing we almost certainly have some time --- if you must have a number call it a century --- to respond to global warming. But we need to use it wisely. We must reduce CO2 emissions in a way that does not further impoverish the poor. The only way we can do that is with cheap nuclear, cheap as it was and could be again. That too has not changed.

The only way we can get back to did-and-can cost nuclear is to free it from the dictatorial, auto-genocidal control of a release obsessed NRC. Pass the Nuclear Reorganization Act.

Figure 1. Aalo’s Critical Test Reactor. 84 years of progress. No messy carbon dust.

The American nuclear establishment has an amazing ability to congratulate itself for abject failure. Every morning the GKG’s inbox is filled with clippings eulogizing all the good things that are happening in the new nuclear golden age,

MOU’s are being signed right and left. Cooperation seems to have gone viral. Nuclear power conferences are experiencing record attendance. The NRC is streamlining all sorts of processes, setting record after record. Nuclear IPO’s are over-subscribed. Billions of investor dollars have been funneled into paper designs, mostly for toy reactors. Quite a bit of this money has leaked out to the promoters. The taxpayer is being bilked at record levels to help these people out. 38 countries have “committed” to tripling nuclear power by 2050. Should be easy since most of them have no nuclear.

The World Nuclear Association just published its 2026 annual report. Here a typical quote:

This strong performance has been matched by unprecedented global engagement. Record attendance at our 50th World Nuclear Symposium and the successful launch of the World Nuclear Supply Chain Conference demonstrate the accelerating interest in nuclear energy. Coupled with our expanding presence in international forums and deeper collaboration with sectors from finance to technology, these milestones signal a structural shift in global support for nuclear.

The glossy report is full of photos of skirts and suits sitting around big tables and preaching into microphones. There are no pictures of a nuclear power plant.

Seems just about everyone agrees the USA needs a lot more dispatchable power NOW. Many people are starting to realize that fossil power costs are on the way up. This is old news to Europeans. We have a proven 60 year old, gigawatt scale technology that did produce and could produce electricity at 3 cents/kWh in today’s money. The Koreans have a current version that can get close to that in the right regulatory environment.

Figure 2. Nuclear Power Plants under Construction.

So how many commercial nuclear power plants are under construction in the US? Zero. How many have been ordered by American utilities? Zero. We have “milestones” and “frameworks” and “targets” and “goals” and “visions” and “initiatives” and no plants. That’s the new nuclear golden age.

Dow Chemical has an agreement to build 400 MW’s of X-energy’s gas reactors for its own use; but that’s contingent on X-energy’s getting through the NRC licensing process. X-energy is currently engaged in the “pre-application engagement” stage. This has involved 17 “topical reports” and 13 “white papers”, some of which have gone through 3 revisions based on NRC’s feedback, which valuable advice is costing X-energy $300 per man-hour. X-energy started the pre-application engagement progress in 2018. I don’t know when they will start the actual license application process.

Under the much ballyhooed DOE Reactor Pilot Plan, we may have a couple of tiny reactors going critical by July 4th. A toy scale, Potemkin nuclear village. As far as I can tell of the original 11 players, all but three are no longer claiming they will meet the dead line.

1) Aalo appears to be the most likely to “succeed” with its Critical Test Reactor. The CTR is vaguely reminiscent of Fermi’s original pile Figure 1. Like Fermi’s pile, the CTR will produce so little power --- we are talking watts --- that it needs no cooling.

2) Antares may also make it with its Mark 0. The Mark 0 is also a zero power test facility.

3) Oklo’s plans seem to change dramatically with each press release, and they put out a lot of press releases. The current push to first criticality appears to be Groves, which is medical isotope reactor, no electricity.

So to celebrate the country’s 250th anniversary we will repeat what the Greatest Generation did 84 years ago. Expect a barrage of self-congratulatory pronouncements.

So far the new nuclear golden age has been golden for a few bureaucrats, consultants, NGO’s, and promoters. It has not been golden for either ratepayers or taxpayers. I see no sign that this is about to change.

In fact, things are getting worse. NuScale, the first of the vaunted SMR’s, has proven even more expensive than the most expensive conventional designs. The Financial Times just reported that Sizewell B will be even more expensive than the astronomically expensive Hinkley Point C, showing the value of replicating a design when the regulator is in charge. The one Free World reactor that might have been able to break through the regulatory logjam has been incarcerated in Korea. And that’s under a supposedly pro-nuclear administration. The Trump administration is a great believer in Executive Orders. It could replace LNT by an EO, but it consciously chose not to. Imagine the progress we will have under a less pro-nuclear President.

Maybe it’s time to heed Greta’s advice. Pass the Nuclear Reorganization Act.

Ban and MAD

In the 1950’s and 1960’s, everyone was petrified of the Bomb, and with good reason. The Bomb had been employed just a few years back with horrific effect.1 Current bombs were 100 times more powerful. The US and Russia were locked in an existential struggle, which periodically flared into proxy wars and eye ball to eye ball confrontations. Both sides had thousands of nuclear armed missiles locked and loaded.

There were two schools of thought about how to handle this perilous situation.

a) Ban the Bomb.

b) Mutually Assured Destruction (MAD)

For the Banners, the first step was to stop nuclear weapons testing. They were led by the Rockefeller Foundation whose management felt personally responsible for the Bomb. In their support of theoretical physics in the 1930’s, they had funded almost all the Manhattan Project greats. Worse, they had single handedly funded the cyclotron at Berkeley. This proved critical to the creation of the Bomb. The Foundation was determined to make amends.

The tool they chose was the health hazards of radioactive fallout. Their problem was the dose rates associated with test fallout were a small fraction of the natural background radiation which all of us experience each day.

The Rockefeller Foundation and its allies decided to argue that radiation produced genetic damage and that damage was unrepairable. Radiation damage just keeps building up. Therefore, the harm was proportional to the total dose, regardless of how rapidly or slowly that dose was incurred. This is like saying taking 365 tablets of aspirin at once is the same as taking 1 tablet per day for 365 days. This no repair hypothesis is called the Linear No Threshold model or LNT.

If LNT is correct, then the Banners could aggregate the tiny increases in dose rate due to test fallout over hemispherical populations and over decades to argue that Bomb testing was invisibly killing millions of people worldwide. The Foundation expertly (and unscrupulously) promoted LNT with all its resources.

The MADers argued that banning the Bomb was politically impossible. Their solution was to make the Bomb so big and so scary that it would never be used. It was in their interest to see LNT adopted. They also wanted to keep testing bigger and bigger bombs. Their stance boiled down to: radiation is as extremely harmful as LNT claims; but that just means we need to make sure the Bomb is never used by making the weapon too horrible to use.

The Atomic Energy Commission(AEC) was in a bind. In 1954, Congress had given the AEC total autocratic control over both nuclear weapons and nuclear power, a decision that makes about as much sense as giving the Department of Defense control over fossil power, on the grounds that both conventional weapons and fossil fuel plants use the same basic chemistry. The AEC had to both implement Mutually Assured Destruction, and promote and regulate nuclear power. Their quandary was captured by Karl Darrow, an important member of the Manhattan Project, who wrote to a colleague: “I take it that there are two main objects. One is to please the public with the prospect of beneficial uses of atomic power, and the other is to scare it out of its boots by threatening it with new weapons.”

The Foundation and it allies were successful.2 In 1959, the National Council on Radiation Protection, many of whose members were funded by the Foundation, recommended that radiation protection standards be based on LNT. In the same year, the AEC quietly adopted LNT. LNT became the law of the land.

LNT is a biology denier

There was only one problem. LNT is wrong. Our bodies do have the ability to repair radiation damage. The damage that we should be and are most concerned about is damage to our DNA. DNA is the part of the cell that contains the information we need to function. If that information is altered, a cell can go rogue and turn into cancer. Radiation damage can do just that.

But so can our own bodies. Humans are oxygen breathers. We combine oxygen with other chemicals to produce the energy we need to function. This is called metabolism. Our oxygen based metabolism allows us to do some pretty amazing things. But as a by product, we are constantly ripping up our own DNA. It turns out that our own internal metabolism produces over 25,000 times as many Double Strand Breaks as average background radiation. Double Strand Breaks are the kind of DNA damage that can lead to cancer.

Nature had to come up with a solution to this carnage. Otherwise we would not be here. Her solution was a panoply of usually very effective repair processes. Google “DNA Repair”. Go ahead do it. You will get around 137 million hits. DNA repair has been studied in mind numbing detail. Three Nobel prizes have been awarded in the field of DNA repair. LNT’s foundational premise that radiation harm is unrepairable is flat wrong. This is indisputable.

Radiation is a bit like aspirin. If we get a whole bunch of radiation at once, our repair processes become overwhelmed, and we see an increase in cancer incidence. However if the radiation dose rate is not too high, our repair processes can keep up. In that case, we can detect no increase in cancer, even when that dose rate is incurred for decades.

Thanks to hundreds of millions of dollars of research, we have a pretty good idea what the dose rate is at which we start seeing an increase in cancer. Because our DNA repair processes must handled internal damage rates that are many thousands of times above background radiation damage rates, that dose rate is well above the dose rate that almost any member of the public will receive in even a release as large as Fukushima. In fact, we have detected no increase in cancer at Fukushima. Even at Chernobyl, we have detected no radiation induced harm to the public, other than thyroid cancer due to kid’s drinking badly contaminated milk.3 This could easily have been prevented by confiscating the milk, as was done at Fukushima. In short, thanks to the remarkable DNA repair processes a providential Nature has bestowed on us, even a very large release will produce little or no detectable radiation harm to the public.

While LNT’s unrepairable predictions are roughly correct when a large dose is received all at once, LNT over-estimates the cancer incidence by factors of a 1000 or more when the same dose is spread over months and years, as it is in a nuclear power plant release. This has caused the radiophobia that both the Banners and the MADers intended. As a result we have turned control of nuclear power over to a autocratic, all powerful regulator who has been told a big release is intolerable. His overriding priority is to prevent a release. The result is the current does-cost of nuclear power in the West is five or more what it did and can cost.

But why take any radiation risk at all? That’s an easy one. Humanity must have cheap nuclear power to flourish. Thanks to its insane energy density, 100,000 times larger than fossil, nuclear power was and can be cheap, dirt cheap, 3 cents/kWh cheap in today’s money for reliable, 24/7 power that takes up almost no space. That cheap electricity could make all humanity wealthier. The worst health hazard of all is being poor. Wealthier means healthier. As a big added bonus, all this power comes with no pollution and very little CO2. By making nuclear power far more expensive than it should be, LNT is killing people.

The first step in getting back to nuclear’s did and can cost is to replace LNT with a radiation harm model that recognizes our indisputable ability to repair radiation damage.4 LNT was not mandated by Congress. It was adopted administratively by the AEC/NRC, and then by the EPA. LNT could be replaced by an Executive Order.

Figure 1. DDI Model response to Okuma high maximal Dose Rate Profile with fixed repair period. A Done DDSB is a DDSB that is either correctly repaired, misrepaired by rejoining the wrong ends, or not rejoined in which case the cell cannot replicate and cause cancer. It simply dies.

Non-technical people tend to lump engineers and scientists. They are both geeky types. The both have this nasty habit of lapsing into an unintelligible tongue written in undecipherable glyphs. There’s not much difference. Nothing could be further from the truth.

Engineers are taught from day one that everything’s a tradeoff. Your job is to come up with a good balance of performance, cost, reliability, and safety. The key question is: how well does it work? In modern parlance, they are non-binary.

Scientists on the other hand are very binary. They are on a search for “truth”. A theory is either right or its wrong.

When it comes to radiation harm models, SNT is multiple orders of magnitude more realistic than LNT when the dose rate profile is spread more or less evenly over a long period. If implemented, it would result in reasonable radiation regulation. The engineer says “Wow. Big improvement. What are we waiting for?”

The scientist is not impressed. As far as she is concerned, both models are wrong. LNT denies our indisputable ability to repair DNA.1 SNT claims that what’s going on in the cell today has zero impact on what happens in the next day. (LNT makes the same assumption.) That’s almost certainly wrong. There’s no real reason to switch from one wrong model to another wrong model. We need more research to find the correct model. My grant application is in the mail.

So what can we do to mollify this bunch, and still remain simple enough to be implemented? One possibility is a Double DSB Inventory (DDI) model. The basic premise of a DDI model is: for cancer the key problem is closely spaced Double Strand Breaks in our DNA. Two closely spaced DSB’s can be viably misrepaired, by rejoining the wrong ends, leading to mutations which can lead to cancer. There’s quite a bit of experimental data that supports this hypothesis. So what we need to do is keep track of how many currently unrepaired DSB’s the cell is dealing with and where they are on the DNA.

As long as the cell’s repair processes are able to keep the current inventory of DSB’s near normal levels, the likelihood of cancer will remain near normal levels. But if the dose rate is high enough, the repair processes won’t be able to keep up with the damage. The inventory of unrepaired DSB’s will start to grow, and the probability of another DSB very close to an existing DSB will climb steeply, and with it the probability of a viable misrepair. If the damage rate if high enough, this process would lead to a very rapid increase in the probability of cancer.

To implement this concept, a DDI model divides the DNA into a large number, M, of slots. It then steps thru time. For now we will assume a time increment of a day. Going into day n, we have b_m DSBs in the mth slot. The probability of repairing or otherwise safely disposing of each existing DSB will depend on the repair period. If the repair period is 1 day, then the number of DSB’s in each slot will reduce by 1 each day, with a floor of zero. The probability of a new DSB in each slot will depend on the dose rate in day n. If each mGy produces k DSB’s, then 1 mGy will produce k new DSB’s, randomly spread over the M slots.2 Keep track of the total number of done DDSB’s --- DDSB’s that are either rejoined or disposed of by not rejoining and allowing the cell to die --- and assume the probability cancer is proportional to that figure.

This line of thinking results in a Markov chain. What happen in day n depends on the inventory of DSB’s in the cell in day n-1. Repair periods are no longer independent. They are chained together.

Figure 2. Two Dose Rate Profile’s of 50 acute daily doses of 5 mGy. Both LNT and SNT claim these two DRP’s result in the same cancer incidence. SNT is conservative for the top profile. It’s still more conservative (less accurate) for the bottom.

Such a model has some attractive properties. SNT (and LNT) thinks the two dose rate profiles shown in Figure 2 have the same cancer mortality. That’s almost certainly not true. The damage may be the same; but in the bottom profile, the cell has twice as long to repair the damage. A DDI model can distinguish between two such profiles.

The simplest DDI model assumes a fixed repair period. For a constant dose rate profiles, in an earlier post we found that the number of DDSB’s was close to quadratic in the dose rate if the repair time did not depend on the dose rate or the current number of DSB’s. But what happens when you give a DDI model the kind of dose rate profile that would be incurred in a big release.

Figure 1 shows the DDI response to the GKG’s worst possible public dose rate profile at Fukushima. Ignore the numbers. All the DDI parameters here have been picked out of the air. We are just trying to get a qualitative feel for how DDI models work. For these parameters, early in the release, the repair processes can’t keep up with the damage. The number of DDSB’s soars. As the dose rate falls, the repair systems halt the climb, and then pull the DDSB’s down. Once the dose rate is low enough, the system is able to keep the DDSB’s at nearly background levels. Nearly back ground is not at background. If you look very closely at the flat portion of the DDSB curve, there is a Monte Carlo ripple; but any increase in cancer incidence will be undetectable.

In the real world, the repair period is almost certainly not independent of the dose rate or number of currently unrepaired DSB’s. Bissell et found, that the DNA repair period depends on the number of Double Strand Breaks (DSB) which the cell has to deal with, which in turn depends on the dose rate.\cite{bissell-2011}[Figure 3] The effect is not subtle. A factor of 20 increase in the dose pulse increases the repair period by a factor of 10.

In A Better SNT we found that, if the repair period increases linearly with the dose rate, then for a constant dose rate profile, the number of DDSB’s is slightly more than cubic in the dose rate. This matches Rowland’s best fit to the dial painter data.

Here the numerical value of the exponent has to be in the range 2.7 to 4.1, with the value 3.15 for the exponent appearing to be the best fit to the [dial painter] data. Needless to say, it is difficult to find any physical meaning for such a dose response function.\cite{rowland-1996}[page 3]

Turns out it is not difficult at all, if you focus on how our bodies repair DNA damage.

Figure 3 shows the DDI response to the Okuma high maximal profile under the assumption that repair times increase with the number of currently unrepaired DSB’s. The response is qualitatively the same as Figure 1; but it takes a lot longer for the body to get thing under control. But once it does, it can keep the number of DDSB’s at very close to background levels.

Figure 3. DDI Model response to Okuma high maximal DRP with mild increase in repair period with increasing number of DSB’s. Vertical DDSB axis different from Figure 1.

But it was near thing. If you increase the slope of the repair time increase with the number of DSB’s to a slightly larger number, then things get totally out of control. Figure 4 shows that DDI models can runaway. We can get ourselves into a situation where the number of DDSB’s just keep increasing, even tho the dose rate is decreasing.

Figure 4. DDI Model response to Okuma high maximal DRP where the increase in repair time results in an explosion of DDSB’s. Vertical DDSB axis totally different from Figure 3.

I find all of this mildly interesting; but DDI and similar models (perhaps DDI with an initial hormetian period) are a dead end policy wise. There are at least three problems;

a) Full Definition.
b) Calibration.
c) Building a Consensus.

Full Definition

How do you turn the above handwaving into an actually usable model? For just one example, what does a slot mean? I’ve made it sound like DNA is a 2-D object, just a long string. In fact DNA lives in a 3-D world. Most of the time the molecule is wound into a tight coil around spools called histones. What does contiguous mean for such a structure? How close do two DSB’s have to be before they become a Double DSB capable of misrejoining? I could easily see august committees debating such issues for decades.

Calibration

Suppose you could somehow get agreement on all these issues. The next step would be calibrating the model to get it to to match real world cancer incidence. The more complex the model, the more parameters you must choose. One of LNT’s big practical advantages is it’s a one parameter model. Once the establishment decided to go LNT, they only had to pick the slope. It turns out one parameter is not enough to even qualitatively model how our bodies work. SNT needs three. But to go to more, opens up all sorts of opportunities for endless debate.

Consensation

Here’s the sad fact. For complex processes such as DNA damage and repair, there is no “correct” model. You can find a fault with any model. And for every fault, there will be a number of possible fixes. This leads to a fractal process in which the number of competing models grows exponentially. And since scientists are humans, just about every such model will find a group of supporters. Building a political consensus behind a single model in such an environment is impossible.

The Plea

So here’s my plea to the choir’s scientist section. There is no correct radiation harm model. Fortunately, for the purposes of regulation and compensation, we don’t need one. We just need a model that’s good enough, one that leads to reasonable radiation protection regulation and a generous but not socially destructive compensation for radiation exposure. LNT fails miserably to meet this low bar. SNT meets it. Get behind SNT for regulatory purposes. After we replace LNT with SNT, you may return to the lab, and bitch about how SNT isn’t really right all you want to.

Table 1. NRC estimate of reactor “fuel weight”

The NRC has released proposed 10 CFR Part 57, Licensing Requirements for Microreactors and Other Reactors with Comparable Risk Profiles for comments. I have read through all 496 mind numbing pages. I expect to be amply rewarded when the collection plate comes around.

Let’s start with what I like about this proposal. If and only if you qualify under Part 57:

The expensive CAT teams disappear.(57.12)

Attacks “by an enemy of the United States, whether a foreign government or person” are not a plant’s responsibility.(57.12) Subpart J does requires protection from “radiological sabotage”. This seems to mean enough cameras, etc so that the plant can quickly alert local law enforcement to a threat, but the wording leave a lot of room for interpretation, and as always the NRC is the sole interpreter.

Remote operation is not automatically outlawed.(57.60)

In theory, you can have an on site staff of zero, if you can get it approved.

Nor is load following.

A plant can be “operator independent”.(57.391)

This means the NRC has determined that no operator operations are required to keep the plant from inflicting 10 mSv outside the site boundary. Such plants can be run by a GLRO (Generally Licensed Reactor Operator). “General licence” is NRC jargon for automatic license. For example, thanks to NRC’s largess, we all have a general license to have a few grams of uranium. In this case, it is any employee of an operator independent plant that has been through the plant’s training program, not an NRC approved and administered program.

N-stamp not a requirement

There is a pathway, Subpart F, for getting around the N-stamp. A plant can use a “commercial grade” component if it is “dedicated”. This seems to mean officially approved by the buyer. This approval imposes a defect reporting requirement on the buyer, which on paper is not unreasonable, but could turn into a mountain of paperwork.

No Emergency Planning Zone

The EPZ is replaced by a nebulous “site boundary”. I did not find a clear description of what the site boundary is. It seems to be whatever the plant has fenced off.

Environmental Impact Statement can be waived

The NRC might not require an EIS if no surface water or groundwater is used in cooling, and the site is “a previously disturbed area”. This may contravene NEPA, and I don’t like requiring dry cooling towers.

No resident NRC inspectors.

Save a little money.

No requirement for a simulator at each plant.

Save a chunk of money.

All this means that qualifying under Part 57 gives the qualifiers a big advantage over non-qualifiers. It does not mean that the qualifiers will produce cheap electricity.

So how do you qualify? What does Microreactors and Other Reactors with Comparable Risk Profiles mean. This ponderous phrase is repeated 87 times in the document. Comparable is a chameleon word. It can means something like “similar”. Or it can mean “capable of comparison”. A 1 GW plant can put out 1000 times as much power as a 1 MW microreactor. A 1 GW plant is comparable to a microreactor. I just did it. Even though Part 57 has many pages of definitions, comparable is not defined. I suspect that was not an oversight.

It turns out there are two requirements to be comparable. Here is Section 57.25 in its entirety.

To be eligible for a construction permit and operating license or a manufacturing license under this part, an applicant must demonstrate that its nuclear reactor or nuclear plant design and operation meets the following entry criteria:

(a) An evaluation of the applicable radiological consequences shows with reasonable assurance that any individual located in the unrestricted area following the onset of a postulated accident that bounds a broad range of design basis accidents would not exceed 1 rem (0.01 Sv) TEDE for the duration of the accident; and

(b) The total inventory of thorium, uranium, and plutonium contained in the nuclear reactor or any individual nuclear reactor that is part of the nuclear plant must not exceed 10 metric tons.

First I need to point out that basing (a) on total dose rather than the dose rate profile proves that the NRC is still using LNT despite all the blather about getting rid of this bit of biological nonsense.1 Part 57 reaffirms LNT which implies a big release is intolerable.

(a) is an exercise in probability manipulation. There are any number of PRA hired guns which will help you make the number, in part by playing with the site boundary. Their job will be made easier by the fact that the wording makes it pretty clear that the (a) criterion is based on a single isolated reactor. (Does any one really believe that you can have a major problem with one Nuscale module with zero probability of it not spreading to any of the others in a 12-pack? Check out the big pool in The NuScale Scam.) (a) will be no problem for any well connected applicant.

The real hurdle is (b). Inventory of heavy metal is a very, very strange metric. If you must boil down a design’s worst case hazard to a single number, the obvious choice is the maximum fission product inventory. That’s what the Nuclear Reorganization Act uses and the NRA is certainly not original in that regard.

The NRC helpfully provides Table 1 to let us know which designs qualify. It’s probably just a coincidence that the upgraded NuScale just makes the cut. But ten tons is a lot of thorium and uranium. I did a back of the envelope that convinced me that a somewhat downrated ThorCon power module (557 MWt, 258 MWe normally) could make the number.

Nuclear reactors don’t have a fuel throttle. A stable reactor will put out as much power as you take out. More coolant flow, more power out. The best reactor designers on the planet are a group in northern Italy called Milano Multiphysics. They are led by a guru named Manuele Aufiero.2 If tomorrow I called them, and said “Hey, Manu. Here’s a weird one. I want you to design me the most powerful reactor you can that never has more than ten tons of heavy metal in the reactor.” They’d think I was nuts; but, if I gave them enough money, they would do it. Don’t know what they would come back with, but I’m confident we are talking north of 500 MW electric. Just need to go through the fuel more quickly.

I’m not sure what to make of this strange requirement. The obvious thought is we are dealing with some very dim lights who think this is the way to put NuScale over the top and justify all the money we taxpayers have blown on this boondoggle. But I can’t believe Chris Wright and the ex-DOGERS are that dumb. I guessing they know the metric is highly elastic; and after “proving” Part 57 via NuScale or some other toy reactor, they will spring the trap and apply Part 57 to useful sized reactors.

This is not going to work. The NRC bureaucrats are still completely and absolutely in charge. Part 57 is more rubbery than the more prescriptive Parts 50 and 52. This means implementing Part 57 will require all sorts of ad hoc decisions, and those decision will be made by an omnipotent, unappealable regulator. There is considerable wording in Part 57 which attempts to put restrictions on the regulator; but it always comes with an out. Usually it’s something like unless required “to provide reasonable assurance of adequate protection to public health and safety or common defense and security.” This wording goes back to the original Atomic Energy Act. We know how that went.

The basic problem is the regulator’s incentives have not changed. They know they will be judged on their ability to prevent a release. Right now they are getting a lot of heat from the Trump administration to show “progress”. So there will be some cosmetic improvements, lots of claims of “firsts” and “beating schedules”, and a trickle of approvals. But US nuclear will remain prohibitively expensive. The regulators will have plenty of help from the incumbents. The last thing they want is the kind of competition needed to drive nuclear’s does-cost back down to its should-cost.

The only solution is a regualtory system which rather than pretending bureaucrats are self-sacrificing saints uses our selfish goals to produce a societal welfare maximizing balance of cost and risk. Pass the Nuclear Reorganization Act.

Figure 1. US gas reserves. The last two years have seen a historic decline despite rising prices.

The Gordian Knot News is not pro-nuclear. The GKN is about solving the Gordian Knot of energy poverty and global warming. Humanity has been blest with a source of electricity that can be and was dirt cheap, a source that produces no pollution and nearly no CO2. In fact, humanity has been twice blest. Nature has also provided us with a DNA repair system, that can easily cope with radiation dose rates a hundred times larger than average background radiation. Unless we accept both those gifts, the GKN wants no part of nuclear power. You can’t solve energy poverty with expensive electricity. Unless you are cheaper than coal, don’t bother.

Our argument has gone no where. The solution is to regulate nuclear in much the same way we regulate other less beneficial and more hazardous activities. But that solution has been rejected. Nuclear in the West just keeps getting more expensive. The Trump administration was probably our only hope to begin to change this process. It has been a dismal disappointment.

Our appeals for the poor have fallen on deaf ears. For the parasites and grifters which a regressively regulated, taxpayer funded nuclear has attracted, the goal is to make nuclear as expensive as possible by eliminating any competition. Privileged climate Chicken Littles are perfectly content with impoverishingly expensive nuclear. Even when the non-poor start suffering in places like Germany and the UK, the required changes are not even considered. Nobody seems to care how expensive nuclear is. Maybe we need a new argument. How about this? Without nuclear power at its should-cost, the West will be subjugated by China. Is this an outrageous statement? Let’s see.

The Iran War may be the last war fought with manned weapons. Future wars will almost certainly be fought exclusively or nearly so with drones and guided missiles. He who has enough such weapons to wipe out the other guy’s drones and missiles wins. The Ukrainians have shown it’s not how many people you have; it’s how many and how good your unmanned weapons are. So far the world’s most expensive navy by a factor of three ($350,000,000,000/year) can’t open a Strait against a country that has no navy and no air force.

Table 1 lists some of the key materials that go into unmanned weapons and their control.1 See the Appendix for some notes on these materials.

An interesting feature of all these products is their raw material is neither rare nor expensive. The raw material for graphite is coal tar or pet coke. The raw material for silicon is high grade sand. The problem is turning that ingredient into the high tech product, and that takes a lot of energy. In most cases, most of that energy must be in the form of electricity. To a large extent, world dominance will be based on cheap electricity. Cheap electricity means you have more aluminum (see WWII), more titanium, more silicon, which means more unmanned weapons than the other guy. Cheap electricity means you have the most powerful AI.

The other striking feature of Table 1 is China totally dominates the US in the production of these critical materials. A Ukraine style war between the US and China today would be no contest. But the Chinese need not be in any hurry. Everything is going their way. They merely have to wait until the advantage is so overwhelming, that no sane administration could not capitulate to whatever they demand. The US will be a subject nation within 50 years.

Thanks to George Mitchell and private property rights, the US has been blessed with quite a bit of relatively cheap gas. But we are running through it quickly. Proved gas reserves, which had been rising since about 2000, fell from about 700 trillion cubic feet at end 2022 to 584 at end 2024, Figure 1. The Reserves Production ratio is around 15 years. The gas will last longer than this; but, unless George Mitchell has a Second Coming, only at increasing cost and increasing cost increases.

Figure 2. US natural gas electric power price

The EIA electric power gas price averaged over $4/Mcf for 2025, and the trend is up, Figure 2. Given the volatility and prospects, a utility might reasonably assume an average price over a gas plant design life of 40 years of at least $5/Mcf (2025 USD). That translates to a baseload LCOE of 4.24 cents/kWh. And that could end up being low, way low. Nuclear’s can, did, and should-cost is less than 3 cents/kWh. In the long run, dominance will depend on cheap nuclear. The Chinese aren’t 5 years ahead of us in cheap nuclear. They aren’t 10 years ahead of us. They are 20 years ahead of us, and gaining a year every year.

But the Chinese aren’t much better than us at regulating nuclear. The griping about the Japanese tossing 6 grams of innocuous tritium into the Pacific Ocean was not entirely a political stunt. The Chinese reacted to Fukushima by shutting down their nuclear program for over a year, and reducing all the targets by more than 20 GW’s. The Chinese have a deep seated fear of floods. They have been very slow to move nuclear inland along the rivers. The State Council regularly approves twice as many plants as the NNSA (National Nuclear Safety Administration) licenses.

China not only imported nuclear technology, they imported foreign nuclear regulation. At first, each technology was regulated according to the regulation of the technology’s origin. The EPR was regulated according to French rules. The CANDU was regulated according to Canadian rules. The VVER was regulated according to Russian rules. The AP1000 was regulated according to NRC rules. But as the NNSA has gained bureaucratic strength, the rules have been “harmonized” toward the US rules.

There is only one explanation that I can come up with for why the Chinese government allows this behavior. They have bought the Two Lies. They are petrified of a release, or at least their people’s response to a release. As a result, their program is not going nearly as fast as it could, build times stagnate around 6 years, and costs are at least double should-cost.

This is an opening. If we come to our senses and rely on the same system that gave us fracking and the WW1 and WW2 winning petroleum that came earlier, we can have nuclear at its should-cost, as we did in the late 1960’s. That means we can have electricity cheaper than China. We can afford to build enough drones, guided missiles, and computer power, so that at a minimum we can be a larger Switzerland. We should be able to maintain hegemony over at least the Western Hemisphere. Pass the Nuclear Reorganization Act.

Notes on Selected Materials

Synthetic Graphite

Figure 3. A single Chinese graphitization facility.

Synthetic graphite is a foundational material for many high tech industries. Synthetic graphite is required for the anodes consumed in aluminum production. They are also needed for electric arc furnaces used in recycling steel, producing steel from Direct Reduced Iron, and many other purposes. Synthetic graphite is a key component of lithium ion batteries. Graphite moderated reactors must use synthetic graphite. Without synthetic graphite, not much else happens. Near as I can tell, China can produce 13 times more synthetic graphite than the US.

Producing synthetic graphite is highly energy intensive, requiring 11,000 to 12,000 kWh per ton, 75% electricity. The final step, graphitization, is done in long furnaces, in which 100 volt, 10,000 amp power is passed through baked carbon heating the material to well over 2000C for several weeks.

Aluminum

Aluminum is the backbone of aviation. The American superiority in aluminum production was an essential factor in our victory in World War II. The key to this superiority was cheap hydro power from the Tennessee Valley and Columbia River dams, which attracted dozens of smelters. Now they are just about all gone. There are none in the Pacific Northwest. One site near my old home in Washington is being used to store road salt. And our hydro power is maxed out, Jacobson notwithstanding.

China produced 45 million tons of Al in 2025. The US produced about 0.8 million tons. We did recycle 3.4 million tons. I guess the motto for the next war will be “Beer Cans to Blackjacks”. (The Boeing Blackjack is a small fixed wing drone.)

Producing 1 ton of primary aluminum requires about 15,000 kWh of energy, almost all electricity, very high reliability electricity. For practical purposes, shutting down a Hall-Heroult smelter is not an option. If the alumina and cryolite freeze, you can basically write off that cell. To catch up to the Chinese, we would need 77 GW. And it better be cheap. We are competing with Icelandic geothermal.

Titanium

Titanium is both a super-metal and an energy hog. It’s light. It has great strength. And it is nearly immune to many forms of corrosion. Titanium is essential to high efficiency turbines.

Currently most titanium is produced by the Kroll process. Titanium oxide, the natural form of most titanium, has proven very difficult to reduce (get rid of everything but the metal) directly. The Kroll process converts the oxide to a tetrachloride and then reduces the TiCl4 with magnesium. The magnesium chloride is then converted back to magnesium by electrolysis, which requires roughly 25,000 kWh of electricity.

The result of the Kroll process is a titanium sponge, which needs a lot more energy mostly heat to turn it into a usable material. People are working on simpler replacements for the Kroll process; but, even if they are successful, they will require at least as much electricity.

China is currently producing about 230,000 tons of titanium sponge per year. US is producing zero. We shut down our last 13,000 ton/year plant in 2021. Currently, all our titanium ore goes to pigment. TiO2 makes a nice bright white, which will come in handy for making surrender flags.

Carbon Fiber

Carbon fiber is another energy hog, 80,000 to 90,000 kWh per ton, most of which need not be electricity. Carbon fiber is unmatched when it comes to strength to weight ratio. It is the preferred material for advanced drones.

The aforementioned Boeing Blackjack is not aluminum and it’s not really a Boeing product. Windsurfers were early adopters of carbon fiber. In the 1980’s and 1990’s windsurfing craze, a number of tiny start ups specializing in carbon fiber products, clustered in the small town of White Salmon, WA. White Salmon is in the middle of the Columbia River Gorge, a windsurfing mecca.

Among them was an outfit called In-Situ. In-Situ specialized in hand launched fixed wing drones which commercial fisherman used as aerial fish spotters, replacing manned aircraft. In-Situ’s secret was its carbon fiber drones were light enough so they could carry a camera and enough fuel to make everything work. It was not long before Boeing came calling, bought In-Situ, and turned In-Situ’s cheap, fish spotting Jeep into a very expensive Humvee called Blackjack.

China produces about 90,000 tons per year of carbon fiber to the US’s 35,000. To add injury to insult, China is now producing T1200 (8000 MPa) carbon fiber at multi-hundred ton scale, something the US can’t do. Weihai is the new White Salmon. (Guangwei, the company that made the breakthrough, began as a fishing rod maker.)

Silicon

Extremely high purity silicon is the foundation for all our electronics ranging from nanoscale memory chips to the 500 amp, 5 kV bipolar valves needed for HVDC transmission.

Going from silica to monocrystalline silicon involves three main steps, all of which require electricity:

1) Silica (SiO2) is reacted with high grade carbon (sometimes charcoal) in an electric arc furnace. This results in metallurgical grade silicon and CO2.

2) This silicon goes through a vapor deposition process involving electrically heated silicon rods to create polycrystalline silicon (polysilicon). At this point, we’ve consumed about 170 kWh/kg.

3) If you want single crystal silicon, the polysilicon is put through the Czochralski process. The polysilicon is electrically melted by induction heating and a single crystal is carefully drawn from the melt. This will cost you another 210 kWh/kg.

As near as I can tell, China has a polysilicon production capacity of well over 2 million tons per year but in 2025 actual production was only a bit under a million tons due to market over-supply The US is producing about 35,000 tons. I can’t find any production figures for mono but China totally dominates the business, with about 93% of the world market.

Figure 1. Childhood thyroid incidence in Belarus, 1986-2006.

In the three major releases of radioactive material to date from nuclear power plants, the only statistically significant radiation harm to the public that has ever been detected is a rather dramatic increase in thyroid cancer found in the young who were living near Chernobyl, Figure 1. The big problem in interpreting this jump is separating the apparent increase simply due to far more intensive screening from the real increase due to radiation.

At Fukushima, about 300,000 children received intensive thyroid screening. 191 thyroid cancer cases were operated on.\cite{yamashita-2018}[Table 2] With no special screening, the childhood thyroid cancer incidence rate is around 1 per million. The incidence was 0.00064 or 640 in a million. Intense screening resulted in a 600-fold increase in diagnosed cases.

There is no evidence that radiation has anything to do with this increase. Thanks to stringent control of contaminated food at Fukushima, almost all these kids received zero to near zero thyroid doses. There were no locational differences between kids closer to the plant and those farther away. The age distribution was inconsistent with a radiation induced spike. When a control program was instituted in Aomori, Yamanashi, and Nagasaki prefectures, nowhere near Fukushima, the detection rates were similar.\cite{hayashida-2013} Toki et al did a detailed study of the Fukushima child thyroid data.\cite{toki-2020} They found no statistically significant relationship between thyroid incidence and I-131 contamination.

However in a Belarus study, 16,213 kids were screened, and 87 had thyroid cancer.\cite{zablotska-2011} Assuming their screening was as intense as the Japanese, we would have expected 0.00064* 16213 = 10.5. In a Ukrainian, 13,127 kids were screened, and 45 had thyroid cancer.\cite{tronko-2006} We would have expected 0.00064* 13127 = 8.5. And there was a strong relationship between the size of the dose rate profiles and cancer incidence. I think something like 90% of these cancers was radiation related.

Almost all of these thyroid DRP’s has been traced to the consumption of milk contaminated with iodine-131. 20 to 30% of any iodine that is ingested ends up in the thyroid gland. An adult thyroid gland has a mass of only 15 grams, one five thousandths of a 70 kg body, and a child’s thyroid is smaller still. The thyroid dose is multiplied by a factor of 1000. At Chernobyl total dose to the thyroid ranged as high as 48,000 mGy.

More importantly, I-131 has a half-life of 8.04 days, and weathering cuts the half-life of iodine in the milk to about 5 days.1 These high total doses were incurred over a relatively short time, a few weeks. The dose rates could be as high as 2000 mGy/d.

Table 1. Zablotska breakdown of Belarussian thyroid cancer data

Table 2. Tronko breakdown of Ukrainian thyroid cancer data,

To estimate the dose rate profiles, the GKG uses a model which takes as input:

1) The peak I-131 concentration in the milk.
2) The ramp time in days it takes to get to that peak.
3) The time in days it stays at that peak. During this period, the increase in concentration is assumed to be matched by the decay. After this period, the milk contamination decays at the effective half-life of I-131.
4) Thyroid mass and uptake.
5) The amount of milk consumed per day,

The first three parameters allow us to compute an idealized milk contamination profile. These are shown as the dashed lines in Figure 2. The last two parameters allow us to convert that contamination profile into a thyroid dose rate profile, the solid lines in Figure 2.

Figure 2. Milk Contamination (read right) and Thyroid Dose Rate Profiles (read left)}

Zablotska et al screened 11,600 Belarussian kids from the area around Chernobyl and found a mean thyroid dose of 560 mGy.\cite{zablotska-2011} Tronko did the same for 13,127 Ukrainian kids and came up with a mean dose of 730 mGy.\cite{tronko-2006} Close to 3000 children were above 1500 mGy. The highest measured cumulative dose was 47000 mGy. Figure 2 shows that, under our assumptions, it would take something like 30,000 to 40,000 Bq/L milk to get to the mean numbers. For the high end kids, the milk contamination must have been well above 100,000 Bq/L.

At Windscale, the highest milk contamination was 50,000 Bq/L on a farm 15 km from the reactor.\cite{henriksen-2013}[page 126] Chernobyl released 1000 times more I-131 than Windscale.\cite{tromedy}[p 166] I have been unable to find any milk contamination numbers for the areas close to Chernobyl. But in the Plavsk region of Russian, 500 kilometers from Chernobyl, 1000 to 5000 Bq/L were measured.\cite{iaea-2008} I would expect contaminations at least 20 times higher 30 km from the plant, on the edge of the Exclusion Zone.

The 50,000 Bq/L profile peaks out at just under 40 mGy/day. The kids who were drinking 100,000 Bq/L milk, got over 70 mGy per day. The Odds Ratio for the Zablotska and Tronko studies are shown on the graph. Because the incidence is low, the statistical power is quite poor. But we start seeing clearly significant increase in cancer at around 20,000 Bq/L. This corresponds to thyroid dose rates in the 10 to 15 mGy/d.

The cumulative doses in these dose profiles, shown in the column labeled mGy, are still far below the worst case thyroid doses observed at Chernobyl. The average dose in the Belarus cohort was 560 mGy. Under the assumptions of Figure 2, this corresponds to a peak milk contamination of 29,500 Bq/L and a peak dose rate of 21 mGy/d.

A lot of kids around Chernobyl exceeded 20 mSv/d to the thyroid. As a result, we have seen as many as 4000 thyroid cancers, and may end up with as many as 160 premature deaths. The one thing we must do in a release is prevent kids from drinking highly contaminated milk.

An SNT Model of Thyroid Cancer

We have fitted a 5 parameter logistic to the Belarus thyroid cancer data. The results are shown in the lower right corner of Figure 3. Astonishingly, it turns out the best fit is to assume a repair period of a day and use a curve which for all practical purposes is an exponential with an upper asymptote of only 0.0010, meaning even if the kid got an infinite dose in a day, the increase in cancer incidence for that day would be 0.0010. The lower hook is there; but it is so small compared to the thyroid doses these kids got it plays almost no role. The authors of the Belarus paper also found that an exponential fitted their data far better than linear, Figure 4. In fact, a linear fit was statistically rejected\cite{zablotska-2011}[page 183] They then proceeded to ignore that inconvenient fact, by focusing only on the low end, where the data is very roughly linear, the opposite of the usual LNT ploy. If you narrow the dose range of interest enough, you can always find a portion of a curve that is sort of linear.

Figure 3. SNT versus LNT on thyroid cancer.

Figure 4. Exponential response to thyroid dose, Zablotska Figure 2. Note the massive error bars. Clinical childhood thyroid cancer is very rare, so the statistical power of the data is very poor.

Focusing on the cancer incidence columns in Figure 3, we see that LNT does an extremely poor job of replicating the thyroid cancer data. There is no LNT slope that can fix this. The 5 parameter logistic has no problem fitting the data. But the fit is very different from the fit to the RERF cancer data which we use to estimate cancer from external radiation.

The WHO says the Disability Adjusted Life Years associated with thyroid cancer is two years.\cite{who-2020} If we apply that to the SNT increase in cancer, we obtain the rightmost column in Figure 3. The expected Disability Adjusted Life Days for the Belarus cohort ranges from 3 to 15 days. Using the US dialysis standard, that a life day is worth $350, the compensation at Chernobyl would run from about \$1000 to $5000 per child. If we assume 50,000 children were involved, the total compensation would be very roughly \$150,000,000. It would have been far better to prevent the consumption of contaminated milk, as was done at Windscale and Fukushima.

Why the Different Response

We are left with the obvious question: why the big difference between the RERF LSS data and the thyroid data? Part of the explanation is the factor of 1000 magnification in dose rate puts us in the upper hook of the S. For the non-thyroid cancers, we are down in the bottom hook of the S and the response is better than quadratic in the repair period dose. For the upper end thyroid cancers we are well up into the upper hook of the S, and the response is far lower than linear in repair period dose. In this topsy-turvy region, fractionation is bad. If you have to get a dose, get it all at once.

It’s also true that clinical childhood thyroid cancer is extremely rare, so even if the repair period incidence flattens out at 1 in a thousand, that’s a thousand fold increase.

But that does not explain the big difference in the inflection point, the daily dose at which we move from the lower hook to the upper hook. For the RERF data that is 180 mSv in a day. For the best thyroid fit, it is only 0.6 mSv in a day. This seems to say that the thyroid is much more radiosensitive than the body as whole. The entire S is mushed to the left.

It’s also possible that it takes a lower dose rate to begin wiping out pre-existing thyroid nodules like an ablation. We don’t know. But one thing we cannot do is ignore the data, and pretend the response is linear. And as long as we prevent kids from drinking contaminated milk, we won’t need a separate compensation program for ingestion. Simply buy up the milk, turn it into cheese, and age it for about 3 months.

Figure 1. “...the United States pledges before you, and therefore before the world, its determination to help solve the fearful atomic dilemma - to devote its entire heart and mind to finding the way by which the miraculous inventiveness of man shall not be dedicated to his death, but consecrated to his life.”[Dwight Eisenhower, 1953-12-08]

This post requires some background. The reader must be familiar with the KEPCO/Westinghouse agreement and the Nuclear Non-Proliferation Treaty, which says something quite different than most people think.

Your preacher’s vacation is over. No more ignoring the elephants in the room. Speaking of elephants, US nuclear’s critical problems --- the NRC and the American tort system --- have been joined by geopolitical maneuvering. This could be an opportunity for American nuclear; but right now it’s another crippling manmade barrier.

By far the best Western PWR is the Korean APR1400. The Korean can build an APR1400 at home for around $3000/kW. In 2012-2020, they built four in the UAE, a very expensive place to do business, for a fixed price of $4500/kW. The plant were completed on schedule, and the Koreans earned bonuses.

The US competition is the Westinghouse AP1000, a cramped design using undersized piping which is difficult to impossible to repair. The AP1000 is a scaled up AP600. Somehow Westinghouse uprated the AP600 by 67% without changing the containment diameter. The last AP1000 built in the West cost over $15,000/kW.

Westinghouse lawyers knew they could not compete with the APR1400. They colluded with the US Department of Energy and conned the Korean national utility, KEPCO, into signing an “agreement” under which KEPCO has to pay Westinghouse something like one billion dollars for every APR1400 KEPCO exports from Korea. See The Most Dismal News Possible for the ugly details.

It is not clear this agreement is enforceable. It’s probably a violation of anti-trust. It’s probably a violation of World Trade Agreements. It most certainly is an egregious violation of the Nuclear Non-Proliferation Treaty which both the USA and South Korea have signed and ratified.

So here’s the plan. The Koreans incorporate the Energy Justice Institute (EJI), This is a wholly Korean owned entity in which KEPCO has no interest. EJI purchases APR1400’s from KEPCO, and sells them to the rest of the world. KEPCO has exported nothing.

Westinghouse will scream bloody murder. Trump will bluster and threaten. So the Koreans go to the World Court in The Hague. EJI not only argues that they have not violated the letter of the agreement; but the agreement itself is invalid, since it clearly contravenes the Non-Proliferation Treaty (NPT). Here’s the full text of Article IV.

1. Nothing in this Treaty shall be interpreted as affecting the inalienable right of all the Parties to the Treaty to develop research, production and use of nuclear energy for peaceful purposes without discrimination and in conformity with Articles I and II of this Treaty.

2. All the Parties to the Treaty undertake to facilitate, and have the right to participate in, the fullest possible exchange of equipment, materials and scientific and technological information for the peaceful uses of nuclear energy. Parties to the Treaty in a position to do so shall also co-operate in contributing alone or together with other States or international organizations to the further development of the applications of nuclear energy for peaceful purposes, especially in the territories of non-nuclear-weapon States Party to the Treaty, with due consideration for the needs of the developing areas of the world.[Emphasis mine.]

The KEPCO agreement is a manifest violation of fullest possible exchange. Since both the USA and South Korea are signatories of the NPT, neither Westinghouse nor KEPCO could legally sign such an agreement. I think EJI will win at the World Court.

Westinghouse will also threaten any potential KEPCO customers with all sorts of legal actions. But they will have to do it in the customer’s national court. Americans are not exactly popular in Europe right now. An awful lot of people are unhappy with Trump’s strong arm tactics. I have a hard time imagining any European court not quickly tossing Westinghouse out on its rear. Now is the right time for the Koreans to call Westinghouse’s bluff. The big winner will be humanity.

Figure 1 sketches a possible layout for a pair of yard built steel APR1400’s on an artificial island. The Auxiliary Building (aka Fission Island) has been thinned and lengthened to allow it to be built in a standard VLCC dock. The site shown has a 10 m mean water depth, a spring to neap tide range of 5 m, a Lowest to Highest Astronomical Tide range of 10 m, and a design storm surge/tsunami of 15 m. These are quite extreme numbers which might be found in a place like Dogger Bank in the North Sea. The resulting breakwater height for this site is 27.5 m. The whole island is backfilled to the level of the top of the breakwater, 17.5 m above MSL.

Most sites will have smaller tide ranges and a smaller storm surge. One such site is the Gulf of Mexico. Figure 2 shows my guess at where the first American APR1400 should go. It’s next to the Sabine Ship Channel about 7 miles off Texas. The Channel is currently being dredged to 48 ft MLLW (14.6 m). I envision a chain of islets along the Channel.

Figure 2. Dual APR1400 Islet on the Sabine Ship Channel. The black rectangle is a plant.

The conditions in the GOM are quite mild. The astronomical tide range is less than a meter. In the Gulf the wind has more influence on the water depth than the tide. A big hurricane storm surge is about 6 m. For these conditions, the breakwater height is 16.25 m. The top of the breakwater is 6.25 m above MSL. 14.6 meters is way too shallow for the Boka Vanguard float off. The final tow in will be something like 40 nm.

To hold the buildings down in the design storm surge, spreaders are attached to the sides of the buildings. These can be seen in the Fission Island cross-section. If the building tries to float, it will need to lift a wedge of soil that widens at the angle of repose as you move up from the spreader.

The spreaders are also built in the yard. They are floated in first, then the building is floated in and ballasted down by filling the double bottom and belt with magnetite. Finally,the spreaders are pushed into position, ballasted down and welded to the hull sides. The process is reversed for decommissioning or replacement. We also have the option of bringing in a large floating crane, Figure 3. These monsters have an outreach of over 100 m, and can service any point in either the Fission Island or Power Island.

The power line to shore is routed on the opposite side of the islet to facilitate crane access to the FI and PI, and to avoid float in/float out conflicts. Depending on location, it can be insulated buried, uninsulated on towers or some combination. If the site is more than about 50 km from shore, it may make sense to go HVDC, this will requires an extension of the island on the harbor side for the conversion equipment.

Figure 3. Hyundai 10000. The boom is 185 m long.

The US Gulf and Atlantic Coasts are rich in suitable sites. Sabine Bank off of Texas is just about perfect. But just about any large bay or offshore shoal will suffice. Table 1. lists some of the possibilities.

This concept is far from original. A whole plant yard was built in Jacksonville for the Offshore Power Systems plant, just off the East Coast of Florida. But the project took too long to get started, and ran into the exploding costs of the 1970’s, and was cancelled.

Figure 4. Offshore Power Systems. Dual Westinghouse floater off Florida. Yard was built but then the project was cancelled.

If the reactor itself is well below sea level, all sorts of emergency cooling options become available. The most obvious is natural circulation using a heat exchanger near the water surface. We will only need to get the plant down to the point where this loop can handle the decay heat, and then our grace period becomes infinite. Some sites will be able to use the tide to allow the condensers to be a heat sink, even if all the seawater pumps are lost.

Reality Check

None of this is going to happen. The NRC will not allow a Korean yard to build the plant the way it wants too. I doubt if any of the yards’ vendors have an N-stamp. Neither do the yards, The yards and the Classification Societies have developed a pretty complete Quality Control system. For example, every ladle is sampled at the mill, and each plate made from that steel is so labelled, They and many of their vendors have jumped through the ISO 9001 hoops, but purely for marketing purposes. They will change those procedures overnight if a better way of doing something requires it. There is no way the yards will accept the restrictions and delays of an N-stamp QA system.

Yard tests wait for no one. If the Class or owners inspectors don’t show up for a scheduled test, that test is deemed passed. This is absolutely critical to maintaining the very tight production schedule. The result will be a glaring gap in the paperwork. Try taking that to the NRC. The NRC will throw up innumerable hurdles and roadblocks. The result will be a Vogtle.

Forgetting about the NRC for a moment, consider the Corps of Engineers. The Corps has gone from the country’s number one despoiler of the environment to a greenie bureaucracy that regards all US navigable waters as their personal nature preserve. When we tried to build a small dock in the Florida Keys meeting all the stringent local requirements, the Corps refused to sign off saying we were blocking a navigable waterway. The water depth at the end of the dock was 4 feet.

Congress authorized the Sabine Channel deepening in 2013. But the Corps sat on it. Nothing much happened until 2025 when Trump yanked the project from the Corps and gave it to a local group called the Sabine Neches Navigation District (SNND). They just (2026-04-14) signed a $239 million contract to go to from 40 to 44 feet with completion in 2028. The 44 to 48 foot contract is expected shortly with completion in 2029. The Corps will be a big problem for any nuclear plant in US navigable water.

Finally, we have Trump’s mercantilism. God knows what the tariff will be on an imported nuclear plant. Worse, we can be sure that the dismal restrictions that the Trump administration has imposed on the Koreans exporting the APR1400 apply to exports to the US, even though the “logic” behind the restriction was preventing unauthorized disclosure of US nuclear secrets to the importing nation. The Queen of Hearts would have a tough time swallowing that one. But that’s how it is.

The only way this works is we totally rethink the way we regulate nuclear. Metanoeite. Pass the Nuclear Reorganization Act. We will also have to bypass the Corps, and decide whether we prefer cheap electricity or funnelling taxpayer and ratepayer money to the lawyers at Westinghouse.

Figure 1. Vogtle submodule construction at Shaw. To somebody who has been in a Korean shipyard, this looks like a photo out of the 1930’s.

Figure 2. 100,000 ton per year panel line. How many people can you find in this photo?

I’m an engineer. That means I was put here to design and build things. The last thing I want to do is harp about a misguided regulatory system, which has turned a providential gift which should lift humanity to new heights into a drag on ratepayers and taxpayers and a haven for parasites and grifters. I need a break. A chorister has asked me if we could build a conventional big PWR in a shipyard. This would combine the low technical risk of a 75 year old technology with the amazing productivity of a world class shipyard. This question gives me a chance to go back to what I should be doing instead of moonlighting as an ineffective JV Jeremiah.

Forgetting about all the regulatory nonsense, the short answer is no, not directly. That idea literally won’t float. A large building dock requires a float out draft of less than 12 m. Table 1 shows the weight and plan area of the main components of the APR1400.

For a naval architect, these weights are preposterous. Our ULCC’s, which were 366 m long and 66 m wide could carry 440,000 tons of oil, and easily survive in sea states that generate up to 1G accelerations away from the roll center weigh about 48,000 tons dry.

The problem is the nuclear obsession with concrete. Concrete not only has near zero tensile strength, it’s brittle, and the cure times are an eternity in a shipyard that built each of our ULCC’s in less than a year. It’s hard to impossible to repair. And thanks to beer can thick rebar, it’s also extremely labor intensive. Concrete is crap.

Figure 3, One man using 48 welding machine to make a stiffened panel.

The yards build everything out of stiffened panels. Figure 3 shows one man tending 48 welding machines on a panel line which will turn out 200 tons of panels in a shift.

Figure 4. Robot welding web to panel. Usually the yard will drop a gang of robots into each compartment to do a bunch of welds simultaneously. This photo is not up to date.

To do a structure like a double bottom, the yards take a stiffened panel, weld the webs with run transversely to the stiffeners using robots, Figure 4, flip this sub-assembly over, and slide it onto another stiffened panel. At this point there is no overhead access, so the yards slide the robots in from the side on temporary rails, Figure 5.

Figure 5. Final web to panel welds for a double bottom. At this stage, the robots must be slid into position on temporary rails, usually one robot per stiffener spacing. They are semi-autonomous. Once they have the green light, they feel around to figure out exactly where they are, and won’t start welding until they are satisfied. So they don’t all light up at the same time. It’s very human like behavior.

The result is it take less than 2 man-hours per ton of steel to produce a double bottom like structure. The weld quality is beyond human capability to produce consistently.

In contrast, Figure 1 shows the panel line at the Shaw Industries facility in Lake Charles where the Vogtle sub-modules were made. When I first saw this photo, I thought I had been transported back to the 1930’s.

If we are going to build our APR1400 in a yard, we must use the yards’ methods. The only unusual structure is the containment. The yards hate double curved surfaces. They can live with single curved. The role their own big pipes. But where they shine is flat plate.

The APR1400’s containment is a 60 m tall cylinder with an ID of 47.5 m. The structure has to take about 4 bar gage internal. The same pressure that a ship would experience at a depth of about 30 m.

Figure 6 sketches how I would build the containment on a panel line. The circle has been turned into a dodecagon.1 Each of the 12 sides is almost exactly a tanker double bottom upended with the stiffeners oriented horizontally, like the tendons in the concrete APR1400. The dome would be 12 single curved segments, a bit like St. Peters but hemispherical.

The inscribed diameter of the polygon is the same as the APR1400 inside diameter. We have a bit more interior space and can use exactly the same in containment arrangement. We will need slightly larger brackets in the corners to support the polar crane.

I have sized the scantlings (plate thicknesses, stiffener depth, thickness, etc) based on the bottom structure of a ULCC, which has roughly the same design pressure, and far higher primary stresses. The APR1400 containment thickness is about 1.5 m of concrete plus a steel liner for leak tightness. (1.5 m of nuclear grade concrete leaks????) My containment has a 3 m gap between the inner and outer shells. This can be filled with sand or water after the structure is in place. ThorCon found that water is a surprisingly good barrier against fast moving impacts. It does a good job of spreading the impact energy over a sizable portion of the structure.

I’m confident this structure is conservative. Of course, this will have to be confirmed. When you add up all the steel, it comes to about 12,500 tons. Our containment will cost the yard less than $12,500,000. A wild guess at the primary loop weight would be 5000 tons. The single heaviest component is the steam generator at 781 tons dry. This would be no problem for the yards. The base plate of a big diesel is over 1000 tons.

We need a light draft of less than 12 m to float out of a big building dock. The APR Fission Island is a nearly square building with the containment in the center, Figure 7; The Koreans call this the Auxiliary Building. The Koreans are lousy at coming up with catchy names. Most good engineers are.

Assuming our Fission Island requires the same plan area as the APR1400’s 102 x 87 meter Auxiliary Building plus an extra 250 m2 for our thicker containment, the Fission Island needs about 9000 m2 plan area. On this footprint, to get a float out draft of 11 meters or less, the light weight must less than 102,000 tons. Coincidentally, that’s about the lift limit of the Boka Vanguard, the biggest heavy lift ship of 110,000 tons.

I can’t see any problem in achieving this. The containment steel weighs about 12,500 tons. Based on scaling the ULCC midship section, the rest of the FI steel will be 29,000 tons. There’s no way the nuclear steam system and its supporting machinery can weigh 60,000 tons. And the ULCC’s had to be designed to far higher loads than the Fission Island will ever see. Our problem will be sinking this thing.

Figure 8. Boka Vanguard loaded with 53,000 ton Jack/St Malo platform. The platform is over 90 meters on a side.

The 100m by 90 m length and beam is well within the capabilities for the Boka Vanguard, Figure 8, which has transported a 107 m diameter offshore platform, We could go with exactly the same Fission Island layout. But only a few build docks can handle the 90 meter beam. It will make sense to rearrange the Fission Island into a longer, narrower structure, to allow more yards to compete for the job.

The 110 m long 62 m wide turbine hall will be child’s play for the Vanguard. I’d probably extend it a bit, and include the 30 x 30 m gas insulated switchgear building.2 I do not understand the APR1400’s turbine hall weight. ThorCon’s problem was sinking the Power Island, despite the turbine’s 3000 ton foundation slab.3

An upper bound on the PI steel weight is 30,000 tons. The turbine itself will weigh about 3300 tons including condensers and generator. It will sit of a 10,000 ton slab. All the other stuff cant weigh more than 5000 tons. An upper bound of the Power Island dry weight might be 50,000 tons, which would give a float out draft of about 7.5m. We have plenty of margin.

The Compound Building is about 72 x 50 m and 25 m high. It will have a steel weight of less than 10,000 tons. The Vanguard should be able to handle both the Fission Island and this structure in a single voyage.

We transport the Fission Island and Power Island in two lifts, and float off near the site. We will probably need about 12 m depth for the final tow. I’m envisioning an artificial island or peninsula in 5 to 20 meters of water, with some dredging to bring the three structures in, ballast down, protect with a breakwater, and backfill.

Ideally we want the RPV or most of it to be below sea level, which opens up all kinds of possibilities for emergency cooling. This will require a FI draft of about 16 m at Lowest Astronomical Tide or a displacement of around 150,000 tons. I’m afraid the Fission Island light weight could be as little as 50,000 tons. We may need as much as 100,000 tons of ballast.

Iron ore has a density of about 5 tons/m3 and a cost of around $100/ton. If we end up needing 20,000 m3 of ore, it will fill up the double bottom, plus about a 3 m high belt around the containment. We will blow $15,000,000 on ballast. We will need to do something similar for the Power Island, to ensure the sea water intake is well below Lowest Astronomical Tide. But we are still below a 100 million dollars for steel and ballast. According to KHNP numbers, all the stuff inside the turbine hall will cost about 400 million. and the the Nuclear Steam Supply System will cost 1.5 billion.\cite{choi-2017} (Both numbers are far higher than they should be.) We are talking about 2 billion dollars for a 1.4 GW plant.

This is all back of the envelope. It will have to be confirmed by doing the actual design. But thanks to recent advances in heavy lift capability, if and only if we go to all steel construction, I’m confident that technically we could build a 1 GW+ Pressurized Water Reactor in a shipyard, and gain the astounding productivity that the world class yards have had to develop in the fiercely competitive environment that they face. We could quickly get back to $2000/kW and less using the same basic technology that the late 1960’s plants used. Build times will start out at around two years and quickly come down to one year. The TG will be the long lead time component.

But this is all dreaming. Shipyard productivity depends on three basics:

1) No one can unilaterally dictate the rules. Everybody involved knows the rules and the rules can’t change in the middle of the game.

2) Total freedom to build the ships the way the yard wants to and change that process as it sees fit, as long as the ships perform to spec. This includes freedom to buy equipment and material from anybody willing to provide it. And freedom to decide on its own quality enforcement system.

3) Intense competition over an extended period, not just between the yards, not just between the yards’ vendors and the vendors’ vendors, but also between the Classification Societies.

We have chosen to not allow these three basics to exist for nuclear power.

Until we build nuclear plants like the Koreans build commercial vessels, attempting to build plants, big or small, in a yard, will accomplish nothing but screw up the yard. Pass the Nuclear Reorganization Act.