The Blackmail Equation — LessWrong