The Direction Finder

This runs over the whole graph: 706 claims, 154 typed edges, 20 conjunctions. It reports what the structure actually supports. One result is baked into the view and worth stating up front: the graph is dense within fields and sparse across them — only 1 edge crosses a math/science/humanities boundary, so the cross-domain lens is mostly empty, and that gap is the most actionable finding here. The combos lens is the rich one. The frontiers lens ranks open problems by how much established machinery is already pointed at them — not by forward "unlock" reach, which is genuinely sparse (the corpus is wired toward open problems, not outward from them).