Joyal's species are a categorification of generating functions. A species assigns to each finite set a collection of structures — graphs, trees, permutations — and the generating function counts how many structures exist on sets of each size. The species carries the structural information; the generating function is its shadow. This is a foundational tool of modern combinatorics, connecting counting problems to their underlying symmetry groups.
Note (2026-11-12; read against the paper's full text): the Dirichlet series of a species in the paper is a sum over the sizes n of the finite sets, of the number of structures on an n-element set divided by n factorial, times n to the power minus s; the paper's main theorem (Theorem 5.1) shows that for a scheme X whose functor preserves products and maps finite fields to finite sets, this series equals the Euler product over the primes of the exponential of the sum over n of the number of points over the field with p to the n elements, divided by n, times p to the power minus n s. The essay's statement that the Dirichlet series sums over prime powers, in contrast to the generating function summing over set sizes, is not the paper's; both sum over the sizes of finite sets, and the primes enter through the Euler product. The theorem has the stated conditions on X, which the essay's 'any scheme of finite type' leaves out, and the paper describes its work as categorifying a bit of the theory of arithmetic zeta functions and as an elaboration of joint work with Dolan; the claims about unifying additive and multiplicative number theory are the essay's.
Baez (arXiv:2502.01833) shows that the same framework categorifies a completely different family of objects: the zeta functions of arithmetic geometry. For any scheme of finite type over the integers, a “zeta species” assigns to each finite set the ways of making that set into a semisimple commutative ring and choosing a point on the scheme over that ring. When the species is assigned not a generating function but a Dirichlet series, what emerges is the arithmetic zeta function — the object that encodes how many solutions the scheme has over every finite field.
The structural point is that the same machine produces both. Species → generating function gives combinatorics. Species → Dirichlet series gives arithmetic geometry. The categorification is not two separate lifts, one for each domain. It is a single lift with two projections. The generating function counts labeled structures on finite sets. The zeta function counts points over finite fields. Both are shadows of the same categorical object, cast in different directions.
This unifies two sides of number theory that have historically been treated as separate disciplines. Additive combinatorics counts how structures compose. Multiplicative number theory counts how primes distribute. The species framework reveals that the objects being counted are the same — finite sets equipped with structure — and the difference is only in how the counting is performed. The generating function sums over set sizes. The Dirichlet series sums over prime powers. The structures don't care which sum you take.