DaVice
9
Yes, that was my point. It would take a large sample size to arrive at 1 recipe for every 4 runs, and he later clarified: 
However, his original statement (and what I quoted him saying) was that it would take “156 runs,” which is not a large sample size at all, in fact it leaves no room for probability and suggests that you will get a recipe every 4 runs, without fail.