Rough calculations: Fermi and the art of guessing — LessWrong