I think you may be missing my point. The variable we are considering is time. First, where is time being spent? Next, where within the time spent is the most likely time for disconnecting? We begin to ask ourselves "what are the different possible mechanisms of disconnecting?" From answering these questions, we reach the conclusion that the distribution of the time dependent disconnects averages shortly before crota kneeling. Sorry I'm not going into more depth, but my phones keyboard makes typing complicated thoughts a very daunting task.