Imagine you have a machine that takes in rough gemstones and guarantees a minimum carat yield per stone based on how brightly they fluoresce under UV light. The brighter the glow, the more carat you're guaranteed — but the question is: what's the worst-case yield at each brightness level? This paper answers exactly that question, except the 'gemstones' are multipartite quantum states, the 'UV glow' is the MABK Bell score, and the 'carat yield' is GHZ extractability — the largest squared overlap with a GHZ state achievable via local quantum channels. The committed claim: for every number of parties n ≥ 3, the exact minimum GHZ extractability compatible with a given MABK Bell score is the affine interpolation from 1/2 at the biseparable bound to 1 at the quantum maximum. This had been known for 3, 4, and 5 parties. The paper settles the previously unresolved case of six or more parties with a uniform analytic proof that works from four parties onward. This is a completeness result — no numerical optimization, no approximation, no 'up to epsilon.' The architecture is pure operator algebra on tensor products of arbitrary local dimension with arbitrary binary measurements. The method leans on a carefully constructed family of 'normal states' (states where certain marginals are maximally mixed) to reduce the optimization landscape, then proves the affine bound by showing any state achieving a given Bell score must have at least that much GHZ overlap. The key structural move is showing that one qutrit paired with qubits at all remaining parties suffices to attain every point on the bound, with measurements held fixed as the score varies. Integrity is about as clean as it gets in mathematical physics: the validation is analytic proof, not simulation. Every claimed bound comes with a rigorous derivation. The paper also proves a robustness result — when extractability exceeds the affine minimum by a small amount, the state is close (in trace norm) to the known optimal mixture, with the distance bounded by a constant times the square root of the excess. The exponent 1/2 is proven optimal, and the constants are uniform across party number for any fixed interior interval. There's no cherry-picking possible when the answer is an exact function. For the qubit-only case, the paper determines a separate, strictly higher minimum extractability curve. This is a meaningful distinction: restricting local dimension genuinely changes the certification landscape. The classification of all states attaining the bound at minimum product of local support dimensions (for ≥4 parties, strict interior scores) adds structural insight beyond just the numeric bound. The milestone question is subtle because this is a foundations result, not an engineering one. The immediate unlock is that any experimentalist performing an MABK Bell test on n parties now has a tight, analytic formula converting their Bell score to a guaranteed GHZ fidelity — no lookup tables, no numerics, no party-count restrictions. The practical gap is that real experiments contend with finite statistics and imperfect measurements, and this paper's bounds assume ideal measurements. Bridging to device-independent certification with finite data is the natural next step. The obvious experiment not run is extending this to other Bell inequalities beyond the MABK family, or to other target states beyond GHZ. The paper's technique is tailored to the MABK structure (binary measurements, specific symmetry). The honest read is (c): the authors almost certainly know how to state the generalization and are saving it, because the MABK/GHZ case is already 35 pages and the techniques likely require substantial new machinery for other inequalities.
