x
history-of-alignment — LessWrong