x
metroskedastic — LessWrong