Modelling Continuous Progress — LessWrong