Imagine you have a locked safe, and you believe that any key which opens it must be unique — there's only one way to unlock the contents. That's roughly what 'uniqueness of purifications' means in quantum information: given a mixed quantum state, there's essentially one way to extend it to a pure state on a larger system. This paper proves that if you trap a single nonabelian anyon in a region of a topological material, that uniqueness guarantee breaks. The safe has multiple incompatible keys. The committed claim: a superselection sector containing a single nonabelian anyon violates Haag duality, which is operator-algebraically equivalent to uniqueness of purifications. This is not a near-miss or edge case. The authors provide both a physics-level argument and a rigorous proof using DHR/BF sector theory, and they instantiate the result concretely in Levin-Wen string-net models — the standard lattice models for doubled topological order. To understand the ladder, you need the backdrop. Haag duality has been a working assumption in algebraic quantum field theory since the 1960s and in the quantum-information-theoretic study of gapped phases since the 2010s. The conjecture the paper disproves — that all gapped ground states satisfy approximate Haag duality — was explicitly stated in prior work by Naaijkens and others as a foundational condition for applying split-property techniques to topological phases. No one had produced a concrete counterexample. This paper does. The architecture is pure operator algebra and lattice model theory. The proof strategy uses DHR superselection theory (Doplicher-Haag-Roberts) adapted to lattice systems, specifically the framework of cone-localized endomorphisms on quasi-local algebras. The key structural insight is that a nonabelian anyon's internal Hilbert space creates inequivalent extensions of the local algebra that cannot be detected from within the cone's complement. The Levin-Wen models serve as concrete playgrounds — these are exactly solvable Hamiltonian models on trivalent lattices whose excitations are anyons labeled by a unitary fusion category. Integrity is high for a mathematical physics paper: the result is a theorem, not a simulation. The proof has two layers — an intuitive physics argument and a rigorous sector-theory proof — and the authors verify the construction explicitly in Levin-Wen models. There's no benchmarking to cherry-pick because this is a no-go result: the claim is that something previously assumed possible is in fact impossible. The validation is mathematical proof, the strongest available. Two additional results sharpen the picture. First, the ground state in question does not allow quantum steering despite being pure — a genuinely surprising quantum-information-theoretic consequence, since pure states normally allow maximal steering. Second, the violation extends to approximate Haag duality, killing the softer version of the conjecture as well. This means there are entire gapped phases — not just individual states — where Haag duality fails everywhere in the phase diagram. The milestone question here is conceptual rather than numerical. The immediate unlock is whether the failure of Haag duality forces revisions to existing proofs about topological quantum error-correcting codes that assumed it. Several results in the theory of topological order — classification of phases, stability of the spectral gap, information-theoretic properties of codes — have Haag duality or approximate Haag duality baked in as a hypothesis. Each of those now needs an audit. The next concrete experiment the authors did not run: testing whether their counterexample breaks specific claimed results downstream, particularly in the classification program for 2D gapped phases. The honest read is (c) — they're saving the consequences for a follow-up, because the counterexample itself is already a full paper.
