Ege, if you find the framework helpful, I’d love to hear your estimates for the factor probabilities 30%, 70%, 80%. I’d also be very interested in seeing alternative endpoint classifications and alternative frameworks. I sense that we both agree that it’s valuable to estimate the nuclear war risk, and basing the estimate on a model that decomposes into pieces that can be debated separately rather than basing it on just gazing into our belly-buttons and tossing out a single probability that feels right.
I’d probably estimate the three factors at ~ 10%, ~ 50% and ~ 10% respectively, so my probability of all out nuclear war between Russia and the US is like ~ 0.5%. Overall, I think I still roughly endorse my reasoning in the following Metaculus comment I wrote in early March:
My sense of this is that the conflict in Ukraine is not going to move us much from the base rate of this event, at least in the short term. I think the following is a factorization of the direct contribution of the situation in Ukraine to this probability:
1. 100 fatalities as a direct result of clashes between NATO and Russian forces anywhere. (~ 2%) 2. Conditional on (1), any NATO country triggers Article 5 to defend against a Russian attack on its territory. (< 50%) 3. Conditional on (2), Russia offensively detonates a nuclear weapon in the territory of any NATO country. (< 50%) 4. Conditional on (3), full-scale nuclear war between Russia and the US. (< 50%)
The product is less than 0.25%. I also think giving 50% to all three of 2, 3, 4 is a significant overestimation of the product, so really the probability should be considerably less than 0.25%.
This is just a factorization of one chain in a Bayes net. I think the base rate of 0.4%/year is quite reasonable: nuclear weapons have been around for 80 years, so a Jeffreys prior gives a base rate of roughly 1⁄160 = 0.625%/year for a nuclear war overall. I think if a nuclear war happens in 2022 it will involve Russia with high confidence, so overall I think this provides an additional 0.5% risk or so of going from the root node to (3) in the Bayes net. The transition (3) → (4) probably has the same chances in either case, so in the worst case scenario the situation in Ukraine doubles the risk of nuclear exchange between the US and Russia in 2022.
I think this doubling is an exaggeration because (2) and (3) both being around 50% seems unlikely. If we put those somewhere around 25% instead, we only get a relatively small increase compared to the base rate, so I think the forecast on this question should be somewhere around 0.2%.
The military situation changing substantially means (1) is now more likely than I had thought in March, so maybe I would now update it to something closer to 5%, but even in this situation I can’t really endorse a risk of all out nuclear war that’s significantly greater than 1%.
I think your middle number is clearly too low. The risk scenario does not require that NATO trigger article 5 necessarily, but just that they carry out a strategically significant military response, like eliminating Russia’s Black Sea Fleet, nuking, or creating a no-fly zone. And Max’s 80% makes more sense than your 50% for he union of these possibilities, because it is hard to imagine that the US would stand down without penalising the use of nukes.
I would be at maybe .2*.8*.15=.024 for this particular chain of events leading to major US-Russia nuclear war.
I don’t think it’s hard to imagine, I can imagine it quite easily. 80% just seems overconfident to me on this question. NATO has no actual obligation to respond to any nuclear use in Ukraine, and I don’t see why you’re so confident that NATO would respond to Russian use of e.g. tactical nukes in Ukraine by attacking Russia directly. It’s not that I think this is unlikely, but in my opinion 80% is just too high of a confidence in what NATO would do in such an unprecedented situation.
That said, this is the part of Tegmark’s forecast that I disagree with the least, because the difference between 50% and 80% is quite small for the purposes of this calculation. I think it’s much more important for him to justify his 30% and 70%, and I assume you would agree with me about that.
The reasoning is that retaliating is US doctrine—they generally respond to hostile actions in-kind, to deter them. If Ukraine got nuked, the level of outrage would place intense pressure on Biden to do something, and the hawks would become a lot louder than the doves, similar to after the 9/11 attacks. In the case of Russia, the US has exhausted most non-military avenues already. And US is a very militaristic country—they have many times bombed countries (Syria, Iraq, Afghanistan, Libya) for much less. So military action just seems very likely. (Involving all of NATO or not, as michel says.)
“A Russian nuclear strike would change the course of the conflict and almost certainly provoke a “physical response” from Ukraine’s allies and potentially from the North Atlantic Treaty Organization, a senior NATO official said on Wednesday.
Any use of nuclear weapons by Moscow would have “unprecedented consequences” for Russia, the official said on the eve of a closed-door meeting of NATO’s nuclear planning group on Thursday.
Speaking on condition of anonymity, he said a nuclear strike by Moscow would “almost certainly be drawing a physical response from many allies, and potentially from NATO itself”.
“-Reuters
I have heard of talk that the US might instead arm Ukraine with tactical nukes of its own, although I think that would be at least comparably risky as military retaliation.
There is a non zero probability that even if NATO can’t come to a decision, the US would just respond unilaterally, so while it’s likely not 80% I would say the probability of significant retaliation is probably quite high?
My point is that a forecaster can have the level of precision where they say 50% is much too low and 80% is much too high. I agree that 50% vs 80% only makes a 1.6x difference in the final number, which is fairly small when you and Tegmark differ by 30x.
Ege, if you find the framework helpful, I’d love to hear your estimates for the factor probabilities 30%, 70%, 80%. I’d also be very interested in seeing alternative endpoint classifications and alternative frameworks. I sense that we both agree that it’s valuable to estimate the nuclear war risk, and basing the estimate on a model that decomposes into pieces that can be debated separately rather than basing it on just gazing into our belly-buttons and tossing out a single probability that feels right.
I’d probably estimate the three factors at ~ 10%, ~ 50% and ~ 10% respectively, so my probability of all out nuclear war between Russia and the US is like ~ 0.5%. Overall, I think I still roughly endorse my reasoning in the following Metaculus comment I wrote in early March:
The military situation changing substantially means (1) is now more likely than I had thought in March, so maybe I would now update it to something closer to 5%, but even in this situation I can’t really endorse a risk of all out nuclear war that’s significantly greater than 1%.
I think your middle number is clearly too low. The risk scenario does not require that NATO trigger article 5 necessarily, but just that they carry out a strategically significant military response, like eliminating Russia’s Black Sea Fleet, nuking, or creating a no-fly zone. And Max’s 80% makes more sense than your 50% for he union of these possibilities, because it is hard to imagine that the US would stand down without penalising the use of nukes.
I would be at maybe .2*.8*.15=.024 for this particular chain of events leading to major US-Russia nuclear war.
I don’t think it’s hard to imagine, I can imagine it quite easily. 80% just seems overconfident to me on this question. NATO has no actual obligation to respond to any nuclear use in Ukraine, and I don’t see why you’re so confident that NATO would respond to Russian use of e.g. tactical nukes in Ukraine by attacking Russia directly. It’s not that I think this is unlikely, but in my opinion 80% is just too high of a confidence in what NATO would do in such an unprecedented situation.
That said, this is the part of Tegmark’s forecast that I disagree with the least, because the difference between 50% and 80% is quite small for the purposes of this calculation. I think it’s much more important for him to justify his 30% and 70%, and I assume you would agree with me about that.
The reasoning is that retaliating is US doctrine—they generally respond to hostile actions in-kind, to deter them. If Ukraine got nuked, the level of outrage would place intense pressure on Biden to do something, and the hawks would become a lot louder than the doves, similar to after the 9/11 attacks. In the case of Russia, the US has exhausted most non-military avenues already. And US is a very militaristic country—they have many times bombed countries (Syria, Iraq, Afghanistan, Libya) for much less. So military action just seems very likely. (Involving all of NATO or not, as michel says.)
“A Russian nuclear strike would change the course of the conflict and almost certainly provoke a “physical response” from Ukraine’s allies and potentially from the North Atlantic Treaty Organization, a senior NATO official said on Wednesday.
Any use of nuclear weapons by Moscow would have “unprecedented consequences” for Russia, the official said on the eve of a closed-door meeting of NATO’s nuclear planning group on Thursday.
Speaking on condition of anonymity, he said a nuclear strike by Moscow would “almost certainly be drawing a physical response from many allies, and potentially from NATO itself”. “-Reuters
https://news.yahoo.com/russian-nuclear-strike-almost-certainly-144246235.html″
I have heard of talk that the US might instead arm Ukraine with tactical nukes of its own, although I think that would be at least comparably risky as military retaliation.
There is a non zero probability that even if NATO can’t come to a decision, the US would just respond unilaterally, so while it’s likely not 80% I would say the probability of significant retaliation is probably quite high?
If you think it’s higher than 50% but lower than 80%, it seems like there isn’t much room there to me?
50% vs 80% is a huge difference, 4x in odds terms.
Not for the purposes of Tegmark’s calculation. Did you check how he uses this number?
My point is that a forecaster can have the level of precision where they say 50% is much too low and 80% is much too high. I agree that 50% vs 80% only makes a 1.6x difference in the final number, which is fairly small when you and Tegmark differ by 30x.
I agree with this, but for the reason you specified I think that precision would be of greater utility elsewhere.