oh, interesting, I assumed the data came from interruptions (that seemed obvious) but I'm surprised you had some specific negative measurements. How do you decide the magnitude of the number? Just counting how long both parties are talking?