One of my friends on Twitter, Venkatesh Rao, recently challenged people to write one hundred tweets on one topic. I took the bait, and wrote one hundred opinions on nuclear war. It was harder than I thought, but felt like a good exercise. It helped me realized where I had big gaps of knowledge, and also helped me expand some theory that I had been chewing on. Read the thread here or on Twitter.
1) Nuclear War is unlikely to cause human extinction. Working on a post about this now.
2) The second scariest thing about nuclear war, after the massive initial destruction and death, is the uncertainty about what would happen the next few years after…
3) Fallout shelters work and they’re still relevant. In a large scale nuclear war, hundreds of millions would die from fallout, but many of these lives could be saved with more fallout shelters. All it takes is a couple feet of soil and food and water for a week or two.
4) Climate effects from nuclear war, “nuclear winter”, are a real risk but the area is severely understudied. Climate science is hard to start with. It’s even harder when the initial conditions are full of uncertainties. Experts disagree.
5) While there are only 9 nuclear powers, there are at least that many that don’t have nuclear weapons but easily could. They build their own nuclear reactors, train nuclear engineers, and could build nuclear weapons within a few years if they decided to.
6) Dr. Strangelove is the best movie. It’s not only artistically brilliant, it’s also super educational for thinking about the history of the cold war and risks of nuclear war.
7) Nuclear EMP weapons have the potential to disrupt society in significant ways. They’re slightly easier to build than a standard ICBM–it takes getting a nuclear warhead to space, but does not require re-entry shielding required–something NK could do
8) While 7) is true, we don’t know how significant a nuclear EMP attack would be. It would disrupt power and destroy a lot of electronics, but I have seen no careful estimates of damage to GDP–not even rough order of magnitude estimates.
9) Nuclear war is more likely in the next 70 years than it was in the last 70 years. The nuclear taboo is stronger, but there are more actors, and more potential areas for conflict. There are more nuclear arms races happening simultaneously.
10) Radiological weapons are really nasty, but it would not be easy to construct a “Doomsday Device” that spread long-lived radiation around the world. Normal hydrogen bombs are deadly radiological weapons if used as ground bursts, but their fallout is short-lived.
11) India and Pakistan, both with nuclear weapons… Not a good situation. They share a border and have a history of conflict. There is a long history of political instability and factionalism. Yikes.
12) The US and Russia are going through serious (nuclear) relationship issues. The US left the INF treaty this year, and the most significant treaty, NewSTART, expires in 2021. This is in danger of not getting renewed, which would be the worst nuclear treaty lapse ever.
13) When it comes to ICBMs, missile defense is super difficult, and very unlikely to be effective against countries with serious nuclear programs. Because it’s hard to test under realistic conditions, it’s very hard to know exactly how ineffective they would be.
14) It’s dangerous when nuclear weapons get faster or stealthier. Both increase the feasibility of a first strike that takes out much of an adversary’s nuclear capability, and this weakens deterrence.
15) Although stealth capability that improves viability of a first strike is destabilizing, stealth that improves survivability of nukes to a first strike is stabilizing. It’s very good for the world if nuclear submarines continue to be hard to detect and track.
16) Nukes are OP
17) What happens in a nuclear war is NOT predetermined. Limited escalation might be possible, we don’t know! Acting as if it’s predetermined is dangerous because it can become a self-fulfilling prophecy. As in, don’t assume any nuclear war is MAD
18) Assuming a nuclear war can be “winnable” is also destabilizing. For most scenarios, a nuclear war would be strongly net negative for all participants.
19) A precommitment to all-out nuclear retaliation to any use of nuclear weapons creates a stronger deterrent than a less severe precommitment, if believed. However, if it’s not believed, it can weaken deterrence, and even if it is believed, it makes the nuclear war far worse.
20) Nuclear non-proliferation efforts have been surprisingly successful! Most serious thinkers in the 1940s & 50s, such as Herman Kahn, thought there would be 20+ nuclear powers by now.
21) Thomas Schelling and Herman Kahn both made great contributions to our understanding of nuclear competition dynamics. They’re often portrayed as disagreeing on most things, but they agreed on many non-obvious points.
22) The Strategy of Conflict and On Thermonuclear War are both worth reading for those interested in nuclear deterrence theory & risks of nuclear war. On Thermonuclear War, though dated, is the most serious attempt I have seen to model what might happen in a real nuclear war.
23) The Doomsday Machine and Command and Control are excellent reads. The former for understanding the insanity of early nuclear war plans and the bias towards nuclear readiness over safety, and the latter for mistakes in weapons risk management.
24) The Making of the Atomic Bomb is excellent if you don’t mind excessive detail. It’s not just about the bomb–it also covers the fascinating history of physics as we learned WHAT THE UNIVERSE IS MADE OF
25) Human minds aren’t equipped to think well about nuclear weapons or risk of nuclear war. People generally slide off the topic, because it’s too overwhelming to think about, and it doesn’t easily fit into our narratives about life or society.
26) Global zero, as in complete nuclear disarmament, isn’t a stable equilibrium so long as great powers are engaged in military rivalry. However, with sufficiently protected second strike weapons, states could agree to reduce their arsenals to small # of hundreds
27) Governments and militaries aren’t competent enough to manage nuclear risk over the long term. Unless something fundamental changes in geopolitical relations, nuclear war is inevitable in the long term.
28) No state wants to start a nuclear war. The most likely way a war would occur is the result of runaway escalation. If a state believes an all-out nuclear conflict is inevitable, they have a strong incentive to strike first to destroy the other state’s nuclear capability.
29) Civilization would recover after an all-out nuclear war. It may or may not happen quickly. There may be a civilizational collapse where whole technologies & supply chains are lost. Eventually they will be rebuilt.
30) We have very little idea what would happen if there was a nuclear war between India and Pakistan. It would be an unprecedented catastrophe, more casualties than WWII, but economic and social effects are very difficult to model, and I haven’t seen any good attempts.
31) Nuclear terrorism is possible but far less risky than nuclear conflict between states. If any terrorists are capable of stealing & using nuclear material, they will have to be exceptionally well organized and resourced.
32) It’s hard to intuit how much more powerful modern nuclear weapons are compared to those used on Hiroshima and Nagasaki.
The kiloton to megaton advance happened quickly.
1945 Hiroshima = 15 kilotons.
1954 “Castle Brave” test = 15 megatons
1000x in 9 years
33) The two closest calls, the times we came closest to nuclear war, happened during the Cuban Missile Crisis. One was when the captain of a Soviet submarine under perceived threat from depth charges wanted to launch a nuclear torpedo, and was talked down by another officer.
34) The other closest call was when Kennedy almost launched a ground invasion of Cuba. He was under enormous pressure from his military leadership and political advisors to do so. But they didn’t Cuba was defended with tactical nukes. They would have been used.
35) Uncertainty about rival nuclear powers’ capabilities is dangerous, as 34) shows. The perceived “Missile Gap” led to a huge increase in nuclear stockpiles, even though the missile gap did not exist.
36) Public accountability is double-edged sword when it comes to nuclear posturing. The public can push for de-escalation, as in the case with the Nuclear Freeze. However, leaders can face pressure to not look weak, as Kennedy did, and to not “compromise with the enemy”
37) Communication between the leadership of nuclear powers is very important to defuse crisis situations, but this is often when leaders face the worst political pressure to “look tough” and “not compromise”.
38) There should be a hotline between every nuclear power for use during emergencies. This currently exists between the US and Russia, and sort of exists between Pakistan and India. Every other nuclear power should follow suit.
39) I predict Iran will successfully test a nuclear bomb in the next 10 years. 😬
40) Aegis missile defense systems can also be used offensively. The US should not have withdrawn from the INF.
41) The proliferation risk from nuclear power is more significant than the environmental and health risks from nuclear waste.
42) Nuclear power plants and spent nuclear fuel sites are vulnerable targets in nuclear conflicts. Long-lived radiological weapons are difficult because it’s hard to move so much material on a warhead. However, it’s relatively easy to hit a spent fuel site with an ICBM.
43) Nuclear powers should negotiate about which which targets that should be avoided in a nuclear war. Nuclear power plants & fuel sites, dams, etc. should be hit under no circumstances.
44) It seems paradoxical that states could plan for nuclear wars which would kill millions of people, and yet also negotiate in advance to reduce the damage of such wars. It’s less paradoxical than it seems, and if done well would not be destabilizing.
45) Early US nuclear targeting policy was incredibly inflexible and needlessly genocidal. While the US still has the worst of these plans, updated with modern capabilities, they ALSO have plans that could spare cities and avoid all-out nuclear war.
46) Status 6 / Poseidon is an underwater, high yield, autonomous, radiological weapon being developed by Russia. There is a lot to unpack here. First, this may be the first time a nuclear state has announced a weapons system aimed explicitly at radiological area denial.
47) Status 6 may also be the first autonomous nuclear weapons system with the ability to do anything more complex than simply reach its target. ICBMs are “autonomous” in a sense but they aren’t very smart.
48) The underwater arms is less visible than the UAV arms race, but it’s actually the most interesting one. UAVs can be remotely operated, but Unmanned Underwater Vehicles usually can’t. AI enables all sorts of new underwater capabilities.
49) Stealth unmanned vehicles need more autonomous capability than other kinds of unmanned vehicles. Managing risk of great power war in the 21st century will require careful thinking and planning about autonomous vehicles & their interactions.
50) In a certain light, nuclear war would be exciting. 🔥
51) Reliable launch detection capability & missile tracking is a good thing for global stability. Uncertainty in this domain creates more pressure to act on short timescales, which could be detrimental in a crisis.
52) A handful of physicists involved in the Manhattan project, like Szilard and Oppenheimer, contributed more in their philosophical and visionary contributions than they did in their physics work.
53) The political and governance problems of nuclear weapons are more difficult than the technical problems. Leo Szilard & others dared think through the implications and imagined possible solutions, rather than the deterrence bandaid we ended up with.
54) Ultimately, Szilard, Oppenheimer, & the other policy-oriented physicists got politically outmaneuvered. This doesn’t mean they were naive or their work was unimportant. It’s a hard game to play and they were right for trying to play to win, even though their chances were small
55) Ultimately, the question of what to do about nuclear weapons is a question of geopolitical power and governance. Like climate change, nuclear weapons pose risks for every state, but differ in that they are controlled by a small number of actors.
56) Local institutional incentives favor maintaining a state’s advantage in deterrence capability, which is stable in the short term, as we’ve seen. However, the long term, deterrence will fail and nuclear wars will occur.
57) Human institutions are not perfectly rational and nuclear deterrence is not magic.
58) We’ve lost something valuable that some intellectuals in the 40s and 50s possessed: the bravery and creativity required to imagine systems of international governance that would greatly reduce risk of nuclear ear. Check out the collection of essays “One World or None”
59) It’s easy to be cynical about the possibility of real international governance of nuclear weapons. The early efforts towards this never came close to succeeding. That doesn’t mean we should give up.
60) Idealism and cynicism are both strong attractors for those working to reduce risk of nuclear war. The former imagine we can reach Global Zero without huge geopolitical shifts, and the latter believe we can only maintain deterrence. Both are importantly wrong.
61) The study of the effects of nuclear war is severely neglected, given its importance. Funding, talent, and flow of knowledge all contribute to the neglect. Nuclear matters get plenty of media coverage, but very little money is spent on, for example, modeling nuclear winter
62) Classification silos weaken our knowledge of the risks of nuclear war. Even experts with classified access only have access to a fraction of the material relevant to understanding their subject.
63) There are legitimate reasons, from both a national and international perspective, to keep many details of nuclear weapon tech, research, & deployment a secret. However, we are not well calibrated on which info should be public.
64) The Manhattan Project physicists had strong views about the question of how to balance secrecy and openness relating to nuclear weapons research. They preferred far more openness than military and political leaders, but secrecy won out.
65) Details about the construction and deployment of nuclear weapons should continue to be highly classified. There should be more openness in research about effects of nuclear war. On questions of arms control and verification, information sharing should be negotiated.
66) The present and future of nuclear deterrence is in the oceans. Ballistic missiles have gotten more accurate and maneuverable, stealth & hypersonic tech has improved, so land targets are more vulnerable to first strikes.
67) While UUVs won’t completely replace nuclear submarines anytime soon, I expect the undersea arena to be increasingly autonomous as UUVs become more capable. Stealth and lurk capabilities are the obvious advantages. A dormant UUV can sit on the bottom & won’t get tired or bored
68) It’s very good for stability if counterforce strikes (against a rivals’ nuclear weapons) remain prohibitively difficult. For now, the US, China, and Russia are fairly safe from a successful counterforce strike. The situation could get better or worse, depending on the tech
69) Advances in machine learning have the potential to both improve and weaken stealth tech. They will improve autonomous capabilities of unmanned stealth vehicles, but also improve signal processing & detection capabilities
70) The 21st century arms races are more expensive than the 20th century races, involve more actors, and span a broader set of technologies and theaters. We live in interesting times…
71) Wikipedia is a great resource for learning about nuclear war.
Hot take I know 🔥
72) North Korea is not going to disarm under the current regime. Kim Jong-un does not want to go the way of Gaddafi. He won’t trust promises made by Western powers and doesn’t want to be controlled by China either. Nuclear weapons protect him from outside intervention.
73) The US, China, and Russia need to cooperate to prevent more countries from obtaining nuclear weapons! The US and the Soviet Union were able to do this despite their conflict. It’s unfortunate that the current climate is so hostile to this.
74) Arms control is underrated, especially today. I want to see more bright young students study arms control. It’s a dark time for diplomacy, time to turn that around!
75) Art and media definitely have a role to play in reducing risk from nuclear weapons. I’ve been very influenced by Dr. Strangelove and The Day After. I’d like to see films that explore long term solutions to our nuclear problems.
76) The Day After is a great, horrifying film. It’s real hard to viscerally feel what a nuclear war might be like. The Day After gets at that. Set in a small Kansas town, a bunch of average Americans go about their normal lives and then they don’t.
77) Fascinating and sad story about risk management and radiation poisoning: The demon core was a mass of plutonium used for testing that killed two physicists in less than a year.
78) The US – Russia nuclear hotline is a rarely used. The last time it was used? President Obama warning Russia not to interfere on election day. Interesting signaling.
79) Why is nuclear war unlikely to cause human extinction?
The most likely mechanism of extinction is nuclear winter. The worst scenarios look like ~20 C cooling in middle latitudes, with much less cooking in the southern hemisphere. These effects dissipate after 5 or 10 years.
80) If there was a nuclear winter as described, billions of people might die. It would be the worst tragedy humanity had ever seen. And yet that’s not close to killing everyone. At least millions of people would survive in the southern hemisphere.
81) Nuclear winter effects are dependent on a lot of factors, like whether the war is during the winter or the summer. It also depends a lot on targeting. An easy way to avoid a nuclear winter is to avoid targeting cities or anything that might cause a firestorm.
82) The other potential mechanism for human extinction from nuclear war is radiation. However, most of the fallout from a nuclear war would only be lethally radioactive for a week or two, and would be localized to the continent where the bombs detonated.
83) People have hypothesized that long-lived radiological weapons could kill everyone, but 1) it would take a hell of a lot of them and 2) with one possible exception (status 6) no one has any.
84) Even if a nuclear war didn’t kill everyone, maybe the population would be so reduced, and society so dysfunctional, that humanity would be killed by something else. While possible this isn’t likely. There aren’t many ways for millions of people across the world to die at once
85) Perhaps a pandemic could finish humanity off after a nuclear war? Not likely, for two reasons. One, a global pandemic spreads via human transportation. If human society hadn’t recovered, then there isn’t much transport. If it had, then there also is modern(ish) medicine.
86) Natural pandemics are self-limiting after a certain point. It’s much, much harder for a disease to kill 99% of a population compared to 90%, as hosts become scarcer. Humans are hard to make extinct!
87) We often talk about nuclear war as if it could only occur once. But I think nuclear war could occur many times. A dark prediction: World War IV will not be fought with sticks and stones, but rather with nuclear weapons, just like World War III.
88) It’s possible nuclear weapons could be one of many factors that drives humanity extinct over hundreds or thousands of years. Cyclical nuclear and conventional wars, climate change, environmental destruction, etc. could eventually make earth uninhabitable.
89) However, for the slow extinction scenarios, we still need to posit mechanisms of human extinction. By default, humans will weather the storm.
90) When thinking about post-nuclear environments, people often point to fossil fuel depletion as a barrier to rebuilding. This is certainly true, but it’s a factor that will slow rebuilding, not prevent it entirely. Humans are creative, and there are many types of fuel.
91) It’s interesting how some parts of tech stagnate while others progress. The hydrogen bomb of today is basically the same as the one of 1950. Delivery systems, on the other hand, have come a long way and have a long way to go.
92) The biggest nuclear bomb ever tested was the Soviet Tzar Bomba at 50 megatons, though it was designed to go up to 100 megatons. They could have built larger bombs, but there was no point. It was more efficient to build more smaller warheads.
93) It may or may not be possible to build weapons more powerful than nukes, but one hopes if these are possible they will never be developed. Nuclear weapons appear to be at peak destructive capacity for any strategic purpose.
94) The power, reliability, and demonstrability of nuclear weapons may be the reason we see so few biological weapons programs. The militaries most capable of biological weapons programs already have nuclear weapons, so why waste resources on a less useful weapon?
95) The largest biological weapons program to date was the Soviet Biopreparat, with it’s 30,000 employees. It didn’t make much strategic sense, but probably got internal support due to the idea that it could help compensate for nuclear disadvantage.
96) I hope states contemplating bioweapon programs would look at the failed Soviet program and realize the paradox of such a program. If you make it public, you’ll receive strong international condemnation. If you don’t, “the whole point is lost if you keep it a secret!”
97) It’s useful to look to South Africa and the Ukraine for cases where states had or could have had nuclear weapons capability and voluntarily gave it up. The South African case is the strongest, since they developed their program on their own rather than inherited some weapons.
98) If global thermonuclear war seems imminent and you have the resources, get to the southern hemisphere.
99) War Games is a great film. I’m not sure there’s all that much in it to learn about nuclear war, though “the only winning move is not to play” feels like a nice sentiment even if it’s not actionable. But there’s nothing like an 80s teenage hacker movie featuring nukes and AI.
100) Risk of nuclear war is scary but the situation is far from hopeless! Humans are amazingly resilient, and ultimately we have strong incentives to cooperate. With luck, we can hill climb to more stable equilibria, and find a path to longterm peace and stability.