Inherited Improbabilities: Transferring the Burden of Proof — LessWrong