I have a slight lingering confusion about how the assumption that agents have knowledge about other agents’ beliefs interacts with Aumann’s Agreement theorem, but I think it works because they don’t have common knowledge about each other’s rationality? I suspect I might also be misunderstanding the assumption or the theorem here.
From what I could gather from talking to Critch, this is based off thinking that Aumann Agreement relies on assumptions that don’t actually hold about human reasoning, but that the probability theory in this paper is pretty similar to how human reasoning works. Technically, you could get it from uncommon priors.
From what I could gather from talking to Critch, this is based off thinking that Aumann Agreement relies on assumptions that don’t actually hold about human reasoning, but that the probability theory in this paper is pretty similar to how human reasoning works. Technically, you could get it from uncommon priors.
Actually in the podcast he chalks it up to no common knowledge of disagreement due to no common knowledge on what exactly high-level words mean.