Houshalter comments on Pascal’s Mugging: Tiny Probabilities of Vast Utilities

[Houshalter]

What about optimizing for median expected utility?

EDIT: I wrote up a much more detailed proposal here.