x
[LINK] Matrix-Style Learning — LessWrong