Graphs Maps Trees by Franco Moretti may have the highest thoughts-inspired-per-word ratio of any book I've read recently. Of course, it helps that there's not many words. This is basically three essays (one per word from the title) and an afterword by a geneticist for the scientist's perspective. Moretti argues that we should employ scientific techniques to assess trends and cycles in literature. Before visions of the graph from the first page of that textbook in Dead Poets Society start running through your head, I should point out that Moretti clearly puts this as complementary to traditional analysis oriented on a single text. Rather, this looks at each novel as a leaf on a tree, or a point on a graph, so overall trends may be seen, necessarily simplifying and missing something to see something else.
In Graphs, Moretti shows how genres tend to have a lifespan, and admits he doesn't know exactly why, but it may have something to do with broad patterns of social cohesion often called "generations." (He's puzzled, as am I, by why groups would organize like this when babies are pretty much born constantly, but that's just a tangent).
In Maps, Moretti shows how plotting the events in a series of village novels according to their location shows that the events eventually fragment and dissolve away from the village itself.
In Trees, Moretti shows how the way clues are treated in Sherlock Holmes stories can be sorted into categories that branch off like a tree. I have trouble seeing the real utility of the tree structure in this, it's more an if-then than a tree necessary, but his next tree shows how the person and tense used in different types of novels can change even within a sentence to show certain kinds of branching, and this is very interesting.
I have some bones to pick on the way science is used in the tree chapter, but I think that deserves a post of its own. On the whole, I like Moretti's approach, and I'd like to take some of it myself. The most important thing I learned from him may not be the techniques or ideas, but the sense of humility and openness, that this "scientific" approach is only one possibility while the traditional analysis still holds. Good to remember.