I really like the trust principle in the paper, about what we can say about the relationship between credence functions when one person would prefer to use another person’s credences than their own. But I’m skeptical about the concept that seems to initially motivate it, namely, the idea that some people might actually be experts. Does any of this depend on there being such a proposition, or can we do it all in a language without such propositions?
I don’t think there really needs to be anything metaphysically like an expert or a proposition that someone is an expert. It’s really about capturing the relationship between a principal’s credence and some unknown probability function. Certain possible relationships between the two are interesting, and total trust seems to carve things at a particular joint—thinking the unknown function is more accurate and being willing to outsource decision-making and deferring in an easy-to-capture way that’s weaker than reflection,
Nice explanation of the paper!
I really like the trust principle in the paper, about what we can say about the relationship between credence functions when one person would prefer to use another person’s credences than their own. But I’m skeptical about the concept that seems to initially motivate it, namely, the idea that some people might actually be experts. Does any of this depend on there being such a proposition, or can we do it all in a language without such propositions?
I don’t think there really needs to be anything metaphysically like an expert or a proposition that someone is an expert. It’s really about capturing the relationship between a principal’s credence and some unknown probability function. Certain possible relationships between the two are interesting, and total trust seems to carve things at a particular joint—thinking the unknown function is more accurate and being willing to outsource decision-making and deferring in an easy-to-capture way that’s weaker than reflection,