>I think of it in terms of probability per year. The fact that we haven't had a nuclear war yet suggests that this probability isn't super high.
This is super misleading reasoning because of anthropic bias. There are many possible Earths. On all the ones were nuclear war happened, we wouldn't be here to observe it. You can only ever observe the "surprising" result that you continue to exist. So you can't use it as evidence in a probability estimate.
We came so close to nuclear war on multiple occasions. The chance can't be that low.
Can we really apply the anthropic principle so narrowly? Are we to conclude that my risk of death when crossing the street could be really high, and I can’t tell anything from the fact that it hasn’t happened because I wouldn’t be around to think about it if it had?
You can look at how often other people are hit by cars. You can also look at "close calls" where you were almost hit but miraculously saved.
But if you have personally walked into traffic without looking hundreds of times, I'd think you are just lucky. And that using your survival as a single data point is just survivorship bias.
This is super misleading reasoning because of anthropic bias. There are many possible Earths. On all the ones were nuclear war happened, we wouldn't be here to observe it. You can only ever observe the "surprising" result that you continue to exist. So you can't use it as evidence in a probability estimate.
We came so close to nuclear war on multiple occasions. The chance can't be that low.