Regarding the point about the expectation of the future being undefined: I think this is correct and there are a number of unresolved issues around exactly when we should apply expectations, how we should treat them, etc.
Nonetheless I think that we can say that they’re a useful tool on lots of scales, and many of the arguments about the future being large seem to bite without relying on getting far out into the tails of our hypothesis space. I would welcome more work on understanding the limits of this kind of reasoning, but I’m wary of throwing the baby out with the bathwater if we say we must throw our hands up rather than reason at all about things affecting the future.
As a minor point, I don’t think that discounting the future really saves you from undefined expectations, as you’re implying. I think that on simple models of future growth—such as are often used in practice—it does, but if you give some credence to wild futures with crazy growth rates, then it’s easy to make the entire thing undefined even through a positive discount rate for pure time preference.
Regarding the point about the expectation of the future being undefined: I think this is correct and there are a number of unresolved issues around exactly when we should apply expectations, how we should treat them, etc.
Nonetheless I think that we can say that they’re a useful tool on lots of scales, and many of the arguments about the future being large seem to bite without relying on getting far out into the tails of our hypothesis space. I would welcome more work on understanding the limits of this kind of reasoning, but I’m wary of throwing the baby out with the bathwater if we say we must throw our hands up rather than reason at all about things affecting the future.
To see more discussion of this topic, I particularly recommend Daniel Kokotajlo’s series of posts on tiny probabilities of vast utilities.
As a minor point, I don’t think that discounting the future really saves you from undefined expectations, as you’re implying. I think that on simple models of future growth—such as are often used in practice—it does, but if you give some credence to wild futures with crazy growth rates, then it’s easy to make the entire thing undefined even through a positive discount rate for pure time preference.