Today we discuss some misconceptions about what "ideal" large data sets should look like and how it will help us bet smarter over the long run.