He doesn't do much better on the other example. Term-frequency saturation is the important part, and he ignores the 1972 paper by Karen Sparck Jones.