Dangerous myths about nuclear war

by Michael Brunnbauer, 2022-06-05

In short, even the most civilised nations may burn with passionate hatred of each other. We may see from this what a fallacy it would be to refer the war of a civilised nation entirely to an intelligent act on the part of the government, and to imagine it as continually freeing itself more and more from all feeling of passion in such a way that at last the physical masses of combatants would no longer be required; in reality, their mere relations would suffice - a kind of algebraic action.
Carl von Clausewitz, On War

I fear nuclear war and I can imagine the horrors it entails, but I can only write about it in a detached way. Please do not be offended by my style. You probably have some conception of nuclear war and fear it too. Some things I write might increase your own fear and some might alleviate it. My goal is to fight the wishful thinking combined with fatalism surrounding the topic, which I consider dangerous. Rest assured that I do not want to paint nuclear war as something you can plan for and be safe. No matter how well you are prepared, survival is a matter of luck.

So let's start with the first myth:

"The risk of nuclear war is low"

Estimating the risk of nuclear war is extremely difficult. But all experts will agree, that:

If you do something that can go wrong over and over again, it will eventually go wrong - no matter how low the probability. This is why a long term goal of worldwide abolishment of nuclear weapons is so important. I am in no way implying that past luck or misfortune will affect present probabilities - this is a common misconception. But if you commit yourself to never give up nuclear weapons, you commit to nuclear war sometime in the future.

As long as the situation does not allow worldwide abolishment, it seems to make sense to reduce the severity and probability of nuclear war with arms control and careful choice of nuclear doctrine and posture. At this point, the disagreements already start: Wouldn't reducing the severity make nuclear war more acceptable and therefore more probable?

Let's have a look at the persistent cliche of how it would look like:

"All buttons are pressed simultaneously, humanity goes extinct"

This is how some non-proliferation activists like Daniel Ellsberg like to think of it. As soon as the taboo of nuclear weapon use is broken, the situation immediately and automatically escalates into a spasm war, all buttons are pressed and then the surviving military staff goes home to die with the rest of humanity.

This situation is exactly like with a doomsday device - the ultimate deterrent where nuclear war can only happen by accident. Building a doomsday device would be foolish and I guess this is the point. But this view of the current state of deterrence is in many ways wishful thinking: Deterrence is not guaranteed and never absolute, nuclear war can start intentionally, it has distinguishable outcomes and current arsenals do not threaten all of humanity - arguably not even the survival of the nations involved.

"Cities are military targets"

Not really. Better think of cities as the hostages of nuclear war.

Strategic bombing - like in WWII - has the goal of destroying enemy morale and/or economic ability to sustain its forces. Unless you plan a prolonged conventional war after a nuclear exchange, both is not very relevant in a nuclear war due to its short duration (hours to weeks). Only the direct and immediate effect on leadership morale is important. Furthermore, any desired effect will be negated as soon as you are unable to continue destruction.

This is why, in a nuclear war, the first attack with the intent to maximize civilian casualties ("countervalue targeting") - if it happens at all - will likely be exemplary - and forces have to be held back for effective blackmail.

The first priority for attacker and defender is to limit damage to the own country. The defender may give punishment priority over limiting damage - but that is a bit irrational. There is an incentive for both sides not to cross the "no cities" treshold first and this seems to be current US policy. Both sides may have a different perception of that treshold though. The precision and the yield options of modern nuclear weapons make it easier to reduce collateral damage but there is of course no such thing as a "clean" nuclear war and military targets can be in urban areas.

Limiting damage also entails targeting the military of the enemy ("counterforce targeting"). A first strike might take out many enemy weapons before they can be used. In a second strike, it makes sense to at least attack weapon stockpiles, bomber bases and submarine ports to deny restocking.

Adherents of minimal deterrence will disagree with this. It is a no-first-use and countervalue-only strategy closely related to Mutual assured destruction. This is how I see it:

This is why the US has a nuclear posture that includes counterforce and flexible response options. How this came about historically can be read in this very interesting article (BTW I do not share its optimism over missile defense). The drawback of counterforce is that it makes arms control more difficult.

There are several tresholds that might be respected or crossed in nuclear war and they are all older than the no-first-use taboo:

DemonstrationReal attack
Foreign territoryHomeland

These tresholds show up in Herman Kahn's famous escalation ladder. What your tresholds are and where exactly you draw the lines depends on your views. The escalation ladders of two adversaries might look quite different and that will complicate things. The higher a situation is on the escalation ladder, the higher the risk tends to be that things will unintentionally get out of control. This is a desired effect, as your opponent might back down out of fear even if he can match your escalation. Thus rungs 4-9 are labelled "rocking the boat". It's the boat both you and your opponent are in.

Despite this, the idea that any nuclear war, regardless of circumstances, would quickly reach the topmost rung seems untenable to me. People who say otherwise will often cite war games that went horribly wrong like Proud Prophet. I doubt that people under the extreme stress of a real situation would simply enact the standard policy like in Proud Prophet. It would be nice to know the number of war games that did not escalate out of control vs. the number of those that did.

"Nuclear war is unthinkable"

It is May 2022 and media is full about speculation about nuclear war. Why? Because Putin has made it more thinkable in order to deter NATO countries from helping Ukraine - with some success. The escalation ladder linked above in fact has a "Nuclear war is unthinktable" treshold and Putin is playing with it.

It would be dangerous to assume that nuclear war is always unthinkable for potential adversaries or that they will always think "it cannot be won". Nuclear weapon countries have to work hard for these goals. Maintaining stability of deterrence in a changing world includes permanent thought about nuclear war.

"Nuclear weapons are useless"

They have been used numerous times since the end of WWII to deter what one side or the other regarded as major or minor provocation. Deterring nuclear war is just the basic function. One cannot be 100% sure what the outcome would have been without them and whether a counterfactual conventional war between NATO and the Soviet Union would have been worse than all the factual proxy wars taken together. But the fact that so many countries regard nuclear weapons as desireable - either their own or those of their allies - speaks for itself.

The dilemma is that the more risk a state is willing to take, the more useful nuclear weapons can be. Minimal deterrence will just barely cover the basic deterrence from nuclear attack. The other extreme is a posture that suggests to your adversary that even a minor provocation might cause a first strike from you. It's quite obvious that this is very dangerous and at the same time very tempting for states with lacking conventional forces and/or revanchist attitudes. In the first phase of the cold war, this role of brinkmanship fell to the US with its massive retaliation policy and tactical nuclear war capability in Europe.

These days, North Korea, Russia and maybe soon China are playing with fire to gain political advantage. If we want to distinguish between "defensive deterrence" and "aggressive deterrence", Russia's recent threats accompanying a war of aggression seem to cross that line.

"Nuclear winter"

The science of nuclear winter is extremely contested and ridden with politically motivated worst case assumptions. A typical paper will assume that:

Nuclear winter cannot be ruled out and some kind of "nuclear autumn" causing famine might even be likely. So what would be the worst case? This paper from 2019 posits 150 teragrams of soot injected into the upper troposphere for a full city exchange between the US and Russia. This number of 150 teragrams has come up again and again in papers since the 1980s - with completely different underlying war scenarios (e.g. including China) and contemporary arsenals. Based on figure 2 of the paper I linked earlier, one could argue for 70 teragrams of soot for a US/Russia exchange with 2022 nuclear arsenals.

But let's assume 150 teragrams and sound modelling. One would have to expect below freezing temperatures over much of the northern hemisphere during summer for years. If not prepared for, this could kill a majority of humankind due to famine - but some people would be able to survive even in the worst affected areas and civilization would not be seriously threatened in the southern hemisphere.

Why is it dangerous to buy into nuclear winter uncritically? Assuming all involved parties buy into it, it will increase deterrence and decrease the risk of nuclear war. This will make limited conventional conflict more acceptable (Stability-instability paradox). Also, it will cause fatalism - increasing the severity of nuclear war due to missing preparedness. If one party does not buy into nuclear winter, deterrence becomes asymmetric and the risk of escalation due to miscalculation increases. The best case may be when all parties have doubts about nuclear winter.


No scare quotes here because fallout is real - but probably not as scary as you imagine: When targeting cities, it is more effective to detonate the weapon in an air burst, such that the nuclear fireball is not near the ground and the fallout becomes negligible in terms of casualties because it is distributed mostly globally and delayed instead of locally and early.

A surface or ground burst instead will diminish the range of thermal and blast effects (e.g. 40% less casualties) but add potentially lethal fallout over a range of up to hundreds of kilometers downwind. Whether the added casualties from fallout compensate the diminished thermal and blast effects depends on variables difficult to predict - like wind direction and speed, population density downwind, how many people seek which protection how fast and how long their supplies will last before they have to leave shelter. The planner targeting a city will usually prefer the more calculable air burst option.

Any "nuclear winter" targeting scenario from above assumes air bursts only and has negligible local fallout casualties compared to the direct effects of the weapons. Global life expectancy would shrink due to increased cancer and mutation risk. The additional risk of birth defects would be lower than the additional cancer risk - possibly by an order of magnitude. This study from 1986 suggests an average additional global cancer risk around 1%, but uses contemporary arsenals and assumes no protection from time spent inside buildings.

Annex A contains a calculation of cancer risk based on data from atmospheric testing and more realistic modelling. The average additional cancer risk seems to be between 0.1 and 0.3% in the northern hemisphere for a full exchange of NATO+Russia deployed arsenals using air bursts only. The average impact in the southern hemisphere would be at least an order of magnitude lower.

The 1986 study also separately examines targeting of nuclear facilities to make the fallout worse and more durable. That would dramatically affect the local situation and at least triple the global numbers, but it would not be able to dramatically change the global situation. The doomsday scenario from the Novel/Movie On the Beach is not on the table.

What is the cause of the residual radiation of [thermo]nuclear weapons? The radioactive fission and fusion products of the weapon and any radioactive isotopes created by neutron capture. Neutrons do not travel far in the atmosphere - this is why neutron bombs are small bombs suitable for attacking enemy troops - not cities. Above 10 kiloton yield, the thermal and blast effects always reach further than any prompt radiation. With modern high yield weapons, almost all isotopes created by neutron capture will share the fate of the other fallout or stay near ground zero.

The bomb remains will be vaporized and spread over the fireball. As the fireball expands and cools, the bomb remains condense to particles so small (10nm-20µm) that they are lifted up and carried away with the fireball if it does not come near the ground. This is the case in an air burst. If the fireball is near or at the ground, bigger fallout particles (up to milimeter size) will form due to the increased concentration of vaporized matter and condensation on unvaporized matter drawn into the fireball. These particles will reach the ground in a matter of hours instead of weeks and months. They simply fall out of the moving mushroom cloud - with bigger particles hitting the ground earlier. This is the early fallout (up to 24h).

The delayed fallout (24h up to years) occurs for the small particles (10nm-20µm). A mushroom cloud is a pyrocumulus cloud like those from firestorms but will rise much quicker. The higher the bomb yield the faster it will rise and the higher it will settle. This is why fallout due to rain within the first 24h is unlikely for modern high yield weapons (source). The black rain falling 20-30 minutes after the Hiroshima air burst was caused by the firestorm cloud (often confused with the mushroom cloud) and it is contested whether it contained fallout from the mushroom cloud (note the disconnect between top and stem typical for an air burst).

Delayed fallout will happen via two mechanisms:

It is dangerous to fear fallout to the point of fatalism. If target planners can assume that many people will not seek shelter, they might choose more ground bursts to increase fallout. But even if you live downwind of a ground burst, you are not necessarily doomed.

This is a simulation by NUKEMAP of a 100 kiloton ground burst at Büchel air base in Germany, where the B61 gravity bombs that are shared with Germany are kept. A ground burst makes sense here to destroy the runway and the bunkers with the bombs (no, they will not add to the yield and my guess is they won't contribute to fallout). Let's assume you are ca. 35km downwind in the middle of the 100 rads per hour fallout contour in the district Kell of Andernach. Wind speed is 12mph (19.31 km/h), so the fallout will start to arrive at your location ca. 1.8 hours after detonation. Click on the button "Probe location" and drag the icon with the question mark near ground zero into Kell. On the right, look for the "Information at sample point" section.

In the first 6 months, fallout exposure/dose rates will approximately follow the equation R * t-1.2 once no new fallout arrives (t=0 is the time of detonation). This means that for every 7-fold increase in time after detonation, there is approximately a 10-fold decrease in the exposure rate (seven-ten rule). After 6 months, fallout decays even faster.

The H+1 dose of ca. 203 rads per hour given by NUKEMAP is the normalized dose rate at Kell one hour after detonation. At that time, the fallout has not even arrived but with that trick, we know know R for Kell (203 = R * 1-1.2 = R). After 1.8 hours - when the fallout arrives - the equation gives 203 * 1.8-1.2 = 100 rads per hour. That number is too high because not all of the fallout has settled but let's be on the safe side. 48 hours after detonation, you would measure 1.95 rads per hour and two weeks after detonation, it would be 0.19 rads per hour. These are the recommended times to stay sheltered: At least 48 hours, optimally two weeks. The outside accumulated dose between 1.8 hours and t hours can be estimated by:

5 * R * ( 1.8-0.2 - t-0.2 )

This is 434 rads in the first 48 hours - a dose that would cause radiation sickness and might kill you if you are unlucky. After two weeks, this would climb to 585 rads. A common goal for fallout shelters is a protection factor of 1000 - reducing those doses to 0.434 and 0.585 rads, but even a protection factor of 10 would save you from radiation sickness. Let's assume a protection factor 10, no relocation and no cleanup - not even due to rain. Accumulated dose in the first two weeks is 58.5 rads. If you leave shelter after two weeks and spend your time unprotected until 6 months have passed, you would accumulate an extra 127 rads. You can estimate your additional cancer risk in percent by multiplying your accumulated long term rad dose by 0.055 (source). That's an extra 10% chance of getting cancer - which might need decades to develop and might not be fatal. Not nice but not "envy the dead" territory either.

How do you achieve a protection factor of 1000? With a shielding of ca. 25cm of steel or 84 cm of concrete or 122cm of earth (these numbers will vary with the density of the material). If you have less shielding, your protection factor for thickness d will be approximately e6.91*d/D - where D is the thickness for factor 1000. If you have several layers of shielding (e.g. different walls in a building), the individual protection factors will multiply. In many situations, a windowless basement will provide sufficient protection - so the most important advice is:

Get inside, stay inside, stay tuned (to emergency advice via radio).

Here are some additional points:

If you want real practical advice, read Nuclear War Survival Skills by Cresson Kearny. For the more theory oriented, The Effects of Nuclear Weapons by Samuel Glasstone and Philip J. Dolan still is one of the most authoritative texts.

The living will envy the dead

This quote is attributed to Nikita Khrushchev, but nobody seems to be able to tell where and when he said that. The Day After shows only the immediate aftermath of those who are worst affected by a worst case scenario. The more powerful movie Threads instead follows events in the UK until 13 years after an attack and includes a nuclear autumn/winter scenario. This illuminates the definition problem of the term "survivor" in time and place: Would it be those alive 1 hour after an explosion, 1 month 1 year or 10 years? Would it include people from countries not participating in the war?

Herman Kahn tried to translate that quote into a more meaningful question: "Is a normal happy life precluded for the majority of survivors?". He obviously had a broad definition of "normal". His outlook was mid to long term (starting 1-3 months after the war) and geographically restricted to a country directly affected (the US). I think these qualifications of the question make sense. His answer was cautiously optimistic but the only answer can be: It depends. On the timing and outcome of the war, the preparations made before the war, the attitude of the people and other factors. Let's try to break down the problem - with the basic assumption that problems not listed here will be less relevant:

  1. Breakdown of logistics causing famine.
  2. Long-term breakdown of economy causing famine and exceptionally low living standards.
  3. Breakdown of social order.
  4. Effects of food and water contamination from fallout.
The first two problems dominate. If they cannot be solved, the other problems will not matter much and if they can be solved, the other problems will likely be alleviated to a tolerable degree.

If the attack timing is not too bad, it seems that stored grain in the US could feed 300 million americans for more than a year. The problem is tending these stockpiles and getting them to where they are needed after an attack. This will be very difficult without some pre-war planning and training. Perhaps the most important factor will be the availability of radiation meters to overcome any reluctance of key personnel to leave shelter within the first days or weeks. An all-out attack shortly before harvest would be the worst timing. With current policies, stockpiles would be low and harvesting would be limited by radiation hazard (crops would be mostly undamaged though as they did not have time to accumulate fallout internally and can be cleaned from adhesive fallout).

In a worst case scenario, reestablishing a sustainable "economy" before stockpiles run out is a race against time. It involves salvaging of resources, relocation of people, decontamination of important areas, appropriate sheltering during work and free time, establishment of currency and more. Whether this problem is tackled centrally or decentralized, it will require good leadership.

The last two problems 3. and 4. will be greatly alleviated by the availability of stocked (likely uncontaminated) food and good leadership. They also depend on attitude. If you think the situation will only get worse, you are more likely to subscribe to "might is right". If you expect pre-war radiation safety and living standards, you will be disappointed. Compared to overall casualty numbers, the long term threat from contaminated food and water is small, if not insignificant - and there are possible precautions to reduce it.

The web of interdependency in modern societies and its vulnerability to nuclear war is the main theme of the movie Threads. Many countries depend on food imports and the efficiency of agriculture may well drop by 50% without modern high-tech, fertilizers and pesticides. This can be offset by abandoning livestock and biofuels - but would countries like the United States do that and keep exporting crop after fighting a nuclear war?

In summary: The secondary effects of nuclear war can be massive - just like the positive effects of precautions. "The living will envy the dead" should not become a self-fulfilling prophecy. More resilience to supply chain shocks is already a popular theme that should be pursued. I also would not worry about degrading an adversaries nuclear forces due to (comparably low amounts of) money spent for civil defense. Nuclear war is horrible enough even with some preparation.

"Nuclear overkill"

You can look up the status of nuclear forces in the world at the FAS.

As for the US and Russia the New START treaty from 2010 currently limits the number of deployed strategic weapons to ca. 1550 for each side. When I use the phrase "current arsenals" in this article, I assume that tactical weapons do not play a big role and that stockpiled strategic weapons are not used in a conflict. This may not be the case in a prolonged tactical nuclear war or in a prolonged crisis where the treaty is broken and there is enough time to match stockpiled warheads with delivery vehicles.

With unreasonable assumptions about the course of a war and arbitrary other assumptions, one can construct an "overkill capacity". Let's look at current Russian nuclear forces and assume the deployed and reserve strategic warheads are all fired with 100% success rate at the 331 US cities (population 96,598,047 / area 76,630 km2). The area to which those weapons can inflict at least 5 psi overpressure ("most residential buildings collapse, injuries are universal, fatalities are widespread" according to NUKEMAP) is 146,038 km2 - 1.9 times the total area of the cities. Make of that what you want.

Final thoughts

There is no deterrence without risk. If you do not want to risk nuclear war with a revanchist nuclear power, you basically have to surrender. Even when you escalate in a crisis, you usually hope that the increased risk of things getting out of control will force the opponent to back down.

The best deterrent is a threat that the other side believes you can and will carry out (capability and credibility) and that it thinks will hurt more than any possible gains. If this is not possible or desireable, doubt about how things may play out still can be a deterrent. In practice, crisis and nuclear war have so many uncertainties that doubt will play a central role. A balance has to be sought between looking like a bluffer and an imminent danger that has to be preempted. This balance will have to be reevaluated permanently - the world does not stand still.

It may happen that an adversary looks like a danger that has to be preempted at any cost. We may go into nuclear war fully aware - it's not like there is no historical precedent of all sides eagerly going to war. If such a fight has to be fought out, it seems desireable to me that it would happen conventionally or by some other means that requires more effort and is slower. Ultimately, we want to continually reduce the risk of nuclear war - that is probability and severity - until we have found a better system that prevents mass slaughter without the risk of mass slaughter.

Until this - admittedly utopian looking - future has arrived, I advise everyone neither to assume the worst nor the best outcome, not to be totally unprepared and to join the conversation on nuclear policy.


Send comments to brunni@netestate.de or to this Twitter thread

Annex A - Possible global fallout doses from UNSCEAR data

543 atmospheric nuclear detonations have been reported due to atmospheric testing and combat use. The United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) has estimated the 100-year average accumulated radiation dose from 1945 to 2099 in the northern hemisphere due to global delayed fallout to be 0.506 mSv (external only) and 1.340 mSv (total including ingestion and inhalation). See Table 16 from annex C of the 2000 report). This estimate is based both on measurements of actual deposition and modelling.

Table 6 of the UNSCEAR report lists the "partitioned" fission yield causing global fallout in the northern hemisphere to be 144 Mt. The fission yield of currently deployed NATO+Russian arsenals is ca. 520 Mt. A simple scaling of the dose is not enough though, as the yield mix of the NATO+Russian arsenals will inject a higher fraction of the fallout into the troposphere and lower stratosphere instead of the upper stratosphere. This leads to faster deposition on the ground and thus a higher dose. Another factor is the latitude mix of the atmospheric tests vs. targets in a NATO/Russian conflict - due to the limited latitudinal spread of tropospheric fallout and varying population density.

I assume the following simplified model for global fallout dose in the northern hemisphere:

D = C * ( MT0-20 * 0.17 + MT20-40 * 0.53 + MT40-60 * 0.3 + ML / 10 + MU / 50 )

To calculate C from UNSCEAR data, I set:

This establishes C at 0.147 mSv / Mt for the total dose and 0.055 mSv / Mt for the external dose.

What would this model predict for scenario A in the 1986 study from Shapiro et al.? The numbers provided in the study are MT = 226 Mt, ML = 1234 Mt, MU = 571 Mt and a resulting average 50-year external dose around 15 rads in the northern hemisphere. Assuming the tropospheric fallout spreads uniformly between 20 and 60 degrees latitude, my model predicts an average 12.57 mSv 100-year external dose in the northern hemisphere. UNSCEAR assumes 0.7 Sievert per Gray and a shielding factor of 0.36, while Shapiro et al. assume no shielding - but this still separates both predictions by a factor of 3. For now, I cannot explain this discrepancy and will treat the predictions of my amateur model as a lower bound with the upper bound being 3 times the lower bound.

Now to a scenario with currently deployed NATO+Russian arsenals according to the FAS Nuclear Notebook. The sources provide some leeway so some of the numbers have been chosen rather arbitrarily:

Country Warheads x yield in kt
Russia 487x800, 325x100, 256x100, 94x100, 330x200, 96x200
US 200x335, 200x300, 616x90, 384x455, 200x150, 100x400, 100x170
France 40x300, 10x300, 160x100, 80x100
UK 120x100

The yield partitioning is derived from table 5 of UNSCEAR annex C:

Total yield in kt Troposphere Lower stratosphere Upper stratosphere
10080 20 0
20014060 0
70080 6200

Partitioning of yields not in the table will be done according to the relative fractions of the best match, which is found as follows:

The fission fraction of all warheads is assumed to be 0.5. Again, it is assumed that the tropospheric fallout spreads uniformly between 20 and 60 degrees latitude. The model predicts an external 100-year average dose of 6.93 mSv and a total dose of 18.53 mSv in the northern hemisphere for a full exchange with air bursts (source code). The total dose amounts to an additional cancer risk of 0.10%.

Changes to this document