Would the fact that he/she does not have the future text in his sample/population and that he uses this dataset as a sample of all the southparks to be ever written (in a prediction mode) make this make sense