x
Relative likelihood — LessWrong