In 1965 in a paper beautifully titled “Path, Trees and Flowers,” Edmonds developed a bit more complicated method for finding the augmenting paths in general friendship diagrams. … Edmonds goes into a long digression on the nature of an efficient algorithm. While he realizes that no formal definition can completely capture the intuitive idea of efficiency, he suggests a notion of efficiency by having a procedure that uses computation time that is “algebraic” in the size of the problem, for example, 100^4 or 100^2 or 100^12. Later this class of problems would become known as P (for “polynomial,” which replaced Edmond’s “algebraic”) and come to represent problems we can solve efficiently. That’s the P side of the P versus NP question.
— Lance Fortnow, The Golden Ticket