@johnnycaffeine The war in Donbas, which Russia intervened in, started two weeks after that timestamp.

Disagree with other commenters saying that Bosnia and Herzegovina, Ukraine, and Georgia are the most likely states to join NATO in the near future. Conditional on NATO expanding before 2024, I'd give it >90% chance it's Sweden and/or Finland. All three of the NATO aspirant countries have significant obstacles to joining, which make it look implausible for any of them to join in the near future. Georgia and Ukraine would need to either resolve their boundary disputes, or convince NATO to accept them despite their boundary disputes. Bosnia and Herzegovina...

My biggest disagreement with this is Bosnia and Herzegovina, which I'd put at more like 2% likely to join. Their Serb component can veto NATO membership, and is firmly opposed to joining. And NATO probably wants to be able to trust that Bosnia and Herzegovina isn't about to collapse in another civil war before letting them in. Presumably you're rating them pretty highly because of their membership action plan, but the MAP doesn't actually remove any of these obstacles to joining. I also think you're overestimating the probably of countries leaving. If ...

@johnnycaffeine If your odds are 1%, then saying 5% has expected value 73*99% - 249*1% = 69.78, but saying 1% has expected value 77*99% - 404*1% = 72.19, so the latter is better for you. Unless you're saying that you value metaculus score nonlinearly, perhaps?

— edited by alexmennen

@Tamay At some point, Russia will either invade Ukraine, or de-escalate and withdraw their troops. Presumably this will happen one way or the other before the end of the year. We can already see that Russia isn't in a big hurry to invade Ukraine, so it's not clear to me that time passing with neither an invasion nor de-escalation provides much evidence that they will eventually de-escalate rather than invade.

@tryingToPredictFuture It's technically possible that someone could believe there's an 18% chance through May and 46% chance from then through the end of the year, but past May is a long time to maintain a military buildup before invading. Conditional on a Russian invasion of Ukraine during 2022, I expect it's probably before the end of May, and I predict that most of the Metaculus and GJI predictors would agree.

— edited by alexmennen

Source on Ukraine not mobilizing?

The hypothesis that Ukraine doesn't take the possibility of invasion seriously doesn't seem like a very good defense. How confident would you have to be that Russia would invade in order for it to not be worth the costs of mobilizing, and how plausible do you think it is that they could justifiably be that confident?

@(archon1410) Despite the phrasing as "for some constant epsilon" in the abstract, the paper on arxiv also gives a specific bound, though they only state bounds below 3.9995, so I'm not sure why the nature article stated a weaker bound than what the paper says. "However, let us mention here for the interested reader that we will give two different proofs of Theorem 1.1, the first (which is a little simpler) with ε = 2^−10, and the second with ε = 2^−7. It will be clear from the proofs that these constants could be improved further with some additional (...