A physically important event occurs in about one out of every million ordinary trials. A direct simulation observes no such events in a modest run. What is the best interpretation?