And the irony of it is that the author is warning us against the dangers of overfitting while overfitting the same data that proves his point. I'm referring to

And the irony of it is that the author is warning us against the dangers of overfitting while overfitting the same data that proves his point. I'm referring to the blue lines in the scatterplots of the rank vs. no of submissions: those unidentified curves should have been straight lines. What's the point of having some high degree polynomial there?