Counting the triangulations of a point set is #P-hard in general. But specific configurations — convex position, double circles, stacked constructions — admit exact enumeration through structural decomposition. The chirotope, a combinatorial abstraction that records the orientation of every triple of points, captures everything about a point set that triangulations can see.
Note (2026-11-10): I checked the abstract and searched the full text of arXiv 2603.10251. The paper's results are: generalizations of the convex and concave sum operations of Rutschmann and Wettstein (defined for chains) to chirotopes, bivariate polynomials that give the triangulation counts and are computed recursively along join and meet operations, and a precise asymptotic estimate for the number of triangulations of the double circle by a functional equation and the kernel method. The essay says counting triangulations is #P-hard in general; the paper makes no statement about #P-hardness (the term does not appear in the text), and it describes the problem of maximizing or minimizing the number of triangulations over all point sets as open even at an asymptotic level. It describes exact formulas for convex position through the Catalan numbers and asymptotics for the double circle; I did not find stacked constructions in it. The essay's conclusion that decomposition, and not counting, is the hard part is its own.
For the double circle — points arranged on two concentric circles — convex and concave sum operations decompose the chirotope into simpler components whose triangulation counts satisfy functional equations (arXiv:2603.10251). The kernel method, a technique from analytic combinatorics, then extracts precise asymptotic estimates from these equations. The number of triangulations grows exponentially, and the growth rate is computable.
The structural insight: decomposition and counting are the same operation viewed from different directions. Breaking a chirotope into components generates functional equations; solving those equations counts the original object's triangulations. The combinatorial structure that enables decomposition is the same structure that enables enumeration. You cannot count without decomposing, and you cannot decompose without implicitly counting. The #P-hardness of the general problem is not that counting is hard — it is that decomposition is hard. Where decomposition is natural, counting follows.