gwillen pointed me to this weird paper by Deutsch from the late 90s that claims to derive the Born rule from some harmless-sounding nonprobabilistic axioms and decision-theoretic reasoning. Interesting but weird.