Micro feedback loops and learning — LessWrong