Sorry for the delay in my reply but I noticed your response after checking if she posted about risk mitigation.
I read the paper you linked and had previously listened to the 80000 hours podcast in alternative foods. I am for research into alternative foods. I think it is a good plan B and your methods should be tested to narrow down the uncertainty in their efficacy. However, my plan A would be $1,000 in flour baked into hardtack and a few thousand multivitamins or other food stores.
I do not think it is necessary to split hairs between counterforce and countervalue. I would be amazed if our launch on warning and use it or lose it mentality has changed. The assumption should be MAD instead of NUTS.
In her nukemap she uses a 15 and 500 kiloton example. The orange part is the thermal radius where fires are widespread. The grey circle before that is the 5 psi overpressure where most residential buildings are destroyed. However it does not show the 1-2 psi overpressure range where windows and fires are blown out. In smaller yield weapons the flash and blast radii that snuffs the fire out nearly overlap. I think secondary fires are unlikely to start due to modern building codes.
If it were possible for nuke map to MIRV you could see 33 blasts, each with 15 kilotons instead of her one 500 kiloton blast. And
In reality her 500KT example is just one piece of the MIRV attack from the SS-18 which would carry ten 500KT weapons. This would extend the blast radius over the city where the fuel load is highest. It would be immediately apparent that the same total number of kilotons from many lower yield weapons are more effective for total destruction than a single 5MT weapon.
In the firebombing of Dresden a first wave of high explosive blockbusters opened up the interior of the buildings. A second wave of bombers dropped incendiaries to start many fires in a predetermined pattern on top of the exposed interiors. A third wave of fragmentary bombs suppressed firefighters and civilians from extinguishing the fires.
The order of operations to make a firestorm are blast, fire, and the suppression of firefighters. Firestorms are hard to pull off and the allies waited for ideal weather conditions before attacking. The high temperature pyrocumulous cloud lifts the soot into the stratosphere.
In a nuclear MIRV blast the firestorming sequence of events does not occur. The first blast is a high altitude burst which disrupt the power grid reducing the likelihood of secondary electrical fires. Natural gas fires may still be possible.
The flashes start interior fires through transparent windows. Fire suppressing sprinklers in commercial buildings would be triggered to suppress them. Lastly, the blast from a second up to a few minutes later snuffs out any remaining fires. Fires started on the perimeter of cities are unlikely to have a high enough fuel load to firestorm or move before they burnout.
https://m.youtube.com/watch?v=ztJXZjIp8OA
https://m.youtube.com/watch?v=Q0kxCjiIyBA
I understand your analysis shows 50Tg of soot making it into the lower stratosphere from a large number of moving fires. I still think rubble burning in such a manner is unlikely. Although you discount this by giving it a 20% weight. It is hard to tell since it is uncertainties on top of uncertainties.
Your analysis for the number of deaths makes sense. When US designed MAD they wanted enough firepower to assure the destruction of Russia. This was likely done by taking the number of deaths per kiloton and matching their population. Russia has a population of 144 million. Therefore, your 150 million number for US deaths is likely correct because the Russians are now matching the number of US weapons.
I have read everything you have posted so far and your thorough analysis has piqued my morbid curiosity. I feel that I should be playing devil’s advocate.
I think you might be underestimating the number of initial deaths in a nuclear exchange and overestimating the number of deaths from a nuclear winter with your assumptions. You do acknowledge that you assume only one nuclear weapon per city which downplays the importance of MIRVs.
Smaller and more accurate weapons will not reduce the total number of fatalities but kill many more people. Increasing the yield of a weapon by a factor of ten doubles the size of the blast radius. It is more efficient to use many smaller and more accurate warheads with overlapping blast radii to attack a city. Using a death to kilotons calculation based on Hiroshima is likely not a good assumption. More yield means more deaths but with a MIRV smaller yields kill more effectively.
All buildings including above ground reinforced concrete structures are demolished at a 20 psi overpressure. MIRVs are designed to create this overpressure evenly throughout their target areas. By design MIRVs will trap people under rubble. The mechanism that significantly reduces the probability of cities firestorming also kills the people who are trapped. This scenario is ideal for the attacker by maximizing fatalities and minimizing the likelihood of a nuclear winter. The combustible material is buried under rubble where it smolders generating carbon monoxide and asphyxiates the people who are trapped.
In a firestorm the temperature increases by air from the surrounding areas stoking the fire just like in a blast furnace. The hotter the temperature the higher the smoke is lifted into the atmosphere where it is unlikely to rain out. The smoldering fires in a MIRV deprive the fire of oxygen which reduces the risk of firestorming.
In the dark calculus of nuclear war your model shows it may even be ethical to double tap a city with a MIRV that is firestorming. A firefighting MIRV could be over a firestorming city within minutes. The city might burn for several hours before the deleterious climate effects are irreversible. This should give just enough time for nonbelligerent countries to launch in order to protect themselves from a nuclear winter famine. I think this scenario is most likely if India and Pakistan started a nuclear war, annihilated each other, but used only a single warhead per city, generating many firestorms.
If MIRVs are not used then I would reduce your models by at least a factor of two for the amount of smoke generated. Unless your model already accounts for the differences seen in Hiroshima and Nagasaki. Most of the damage and deaths in Hiroshima were due to the firestorm after a single nuclear weapon. Nagasaki did not firestorm due to terrain masking and would not have contributed as much to a nuclear winter.
I believe a counterforce attack will quickly escalate to a countervalue attack. An initial attack would include runways over 7,000 feet long. It is likely that many airports would be targeted to deny nuclear bombers the ability to use them.
A counterforce attack would include groundbursts against missile silos, Cheyenne mountain and Mt Weather. There are many populated areas nearby that will receive lethal levels of fallout from groundbursts on those hard targets. Therefore, I do not see a difference between counterforce and countervalue targets other than counterforce targets being the initial targets of SLBMs.
I think there are ways to reduce these risks assuming we decide to keep nuclear weapons instead of scrapping them altogether. We should prioritize submarines and SLBMs over strategic bombers, hardened ICBM silos, and cold war bunkers. Nuclear submarines are harder to destroy, they operate far away from populated areas, and can quickly hit their targets. This would significantly reduce the number of groundbursts and deaths from fallout. I give nuclear war a 0.1% chance per year since it does not occur once every ten years and likely not once every 100 years. At 1 in 10,000 years we are talking about the entire history of civilization. I put it at 1 in 1000 years because I cannot rule that out and it is still significant enough to look into ways of reducing risk. I see now that I am even more optimistic than the minority report “super predictors” :)