Retrospective on a quantitative productivity logging attempt — LessWrong