Yep, my guess is that the key point of failure tipping the system into a runaway cascade was increased latency in ingesting status reports on location/availabil

Yep, my guess is that the key point of failure tipping the system into a runaway cascade was increased latency in ingesting status reports on location/availability of crew. There probably exists an unknown "delay time threshold" in which to process and act on changing status updates. And once the average status update time exceeds that latency it just snowballs because the internal model the system is solving for is sufficiently out of sync with the real world that its "forward fixes" are causing new future failures approaching a 1:1 rate so the system can never catch up.