Precision of Sets of Forecasts — LessWrong