I don’t particularly have strong opinions agreeing or disagreeing with this estimate, but want to use it as an opportunity to note that Fermi estimates like this often don’t work well when the conditionals aren’t well laid out.
To explain, let’s say we accept that there is a 5% chance that a NATO country will invoke Article 5 by the end of the year. The question is where most of that probability mass is. If, as seems plausible, it’s almost all in cases like “Poland has an accidental incursion by Russian troops in Ukraine / a Russian missile goes astray, and Poland wants to make it absolutely crystal clear that they are not willing to get dragged into the war,” the Article 5 declaration isn’t actually indicating a war with NATO. Even if 100 troops die, I’d suspect that almost all of the probability mass is on smaller and quickly resolved incidents.
The next question is a conditional—conditional on a war, what is the probability it turns nuclear. What is “a war”? Well, we are probably drawing from the set of events that we would consider prototypical of wars. But “a war” is not what we care about—we care about the probability conditional on the specific war, and if most such wars are the kind that are quickly resolved, the 25% estimate isn’t appropriate—especially when we then move on to the implied probability.
Oh, and anyways, everyone anywhere near a major city, or near a nuclear power plant, should probably have a bottle iodine pills for radiation poisoning, just in case. They cost like $5.
I don’t particularly have strong opinions agreeing or disagreeing with this estimate, but want to use it as an opportunity to note that Fermi estimates like this often don’t work well when the conditionals aren’t well laid out.
To explain, let’s say we accept that there is a 5% chance that a NATO country will invoke Article 5 by the end of the year. The question is where most of that probability mass is. If, as seems plausible, it’s almost all in cases like “Poland has an accidental incursion by Russian troops in Ukraine / a Russian missile goes astray, and Poland wants to make it absolutely crystal clear that they are not willing to get dragged into the war,” the Article 5 declaration isn’t actually indicating a war with NATO. Even if 100 troops die, I’d suspect that almost all of the probability mass is on smaller and quickly resolved incidents.
The next question is a conditional—conditional on a war, what is the probability it turns nuclear. What is “a war”? Well, we are probably drawing from the set of events that we would consider prototypical of wars. But “a war” is not what we care about—we care about the probability conditional on the specific war, and if most such wars are the kind that are quickly resolved, the 25% estimate isn’t appropriate—especially when we then move on to the implied probability.
Oh, and anyways, everyone anywhere near a major city, or near a nuclear power plant, should probably have a bottle iodine pills for radiation poisoning, just in case. They cost like $5.