x
structural cost — LessWrong