Not using a priori information for Russian propaganda — LessWrong