I wonder how they measured them. Is it possible with large enough sample sizes the average values would probably be about right? Sort of like that idea if you get 10,000 people to guess how many beans are in a jar, the average will get closer and closer to the actual value as you go?