CEV is not even guaranteed to output anything, if there is insufficient convergence.
So eliezer’s rigid list must be defined by some other criterion, presumably decided by him. For example, he could use his own Individual extrapolated volition with a certain preferred measure over the free variables of extrapolation, unless CEV gave a coherent output, in which case use CEV.
Another problem with using CEV is that its output could be sensitive to minor changes to the human population of earth. But in Eliezer’s metaethics, “Should” is a rigid designator, so he would have to pick a specific time, correct to the exact second, and that time had better be in the past, otherwise “Should” would be nonrigid—you would be able to change “Should” by doing something that affected people.
I thought CEV was not what defined “right”-eliezer, but a useful heuristic for approximating “right”-eliezer, with a built-in hedge that straight majoritarianism doesn’t have.
CEV is not even guaranteed to output anything, if there is insufficient convergence.
So eliezer’s rigid list must be defined by some other criterion, presumably decided by him. For example, he could use his own Individual extrapolated volition with a certain preferred measure over the free variables of extrapolation, unless CEV gave a coherent output, in which case use CEV.
Another problem with using CEV is that its output could be sensitive to minor changes to the human population of earth. But in Eliezer’s metaethics, “Should” is a rigid designator, so he would have to pick a specific time, correct to the exact second, and that time had better be in the past, otherwise “Should” would be nonrigid—you would be able to change “Should” by doing something that affected people.
I thought CEV was not what defined “right”-eliezer, but a useful heuristic for approximating “right”-eliezer, with a built-in hedge that straight majoritarianism doesn’t have.
If not CEV, then what? I define “right” to be criterion-set X, where X is, um, … oh, is that the time?