[SEQ RERUN] Justified Expectation of Pleasant Surprises — LessWrong