Then you go back and read the source so you can try to run through the calculation backwards in your head to estimate, relative to the surrounding comments' timestamps, what the score might be based on its position?