Science and Fiction: a pragmatic approach — LessWrong