Imagine you're packing a suitcase for a long trip. You might think that coordinating your outfits across all 14 days — this shirt pairs with those pants on Thursday because you'll rewear the jacket from Monday — would let you pack more efficiently than just optimizing each day's outfit independently. In quantum information, the equivalent hope has haunted the field for decades: maybe by entangling inputs across many uses of a noisy channel, you can squeeze out communication rates higher than what any single use allows. This paper says: if your packing strategy treats every day as interchangeable (permutation invariance), the elaborate cross-day coordination buys you nothing in the long run. The committed claim: for any finite-dimensional memoryless quantum channel, the optimized coherent information per channel use converges to the single-use maximum when inputs are restricted to have full permutation symmetry. Superadditivity — where multi-use correlations beat single-use optimization — can still happen at finite block lengths, but it washes out asymptotically. This is a structural impossibility result, not a computational one. The mechanism leans on de Finetti reduction, a workhorse technique that says permutation-invariant states on many copies look approximately like mixtures of product states. The authors exploit this to show that any gain from correlations gets diluted as the block length grows. They also establish a strong converse for entanglement generation: if you try to communicate above the single-use coherent information rate using permutation-invariant inputs, your fidelity doesn't just fail to reach 1 — it crashes to zero. This is the "cliff edge" version of a capacity bound, not the gentle slope. Beyond the headline communication result, the paper identifies a shared variational structure underlying several known multiplicativity and additivity results — completely bounded norms, sandwiched Rényi channel divergences, entanglement-assisted communication rates. Rather than treating these as separate theorems requiring separate proofs, the authors show de Finetti reduction plus permutation covariance gives you a unified machine that reproves all of them. This is the methodological contribution: not new results on those fronts, but a single proof architecture that makes the pattern visible. The integrity picture is characteristic of mathematical quantum information: the validation is proof-based, not experimental. There are no benchmarks to cherry-pick and no simulations to tune. The strength of the result depends on the strength of the assumptions — full permutation invariance is a genuine restriction. The open question the field actually cares about is whether general (non-permutation-invariant) inputs can achieve superadditive capacity asymptotically for memoryless channels. This paper does not resolve that question; it resolves the symmetric sub-case cleanly. The output entropy result extends the symmetry story further: for permutation-invariant inputs, correlations cannot sustain an asymptotic reduction in output entropy per use below the single-use minimum. This holds for Rényi entropies with order greater than one and for von Neumann entropy. It's a natural companion to the capacity bound — you can't extract more information and you can't compress the output entropy, both for the same structural reason. The gap this paper explicitly leaves open is the big one: does asymptotic superadditivity exist for general inputs on memoryless channels? The Hastings counterexample showed minimum output entropy is not additive, and superadditivity of coherent information at finite block length is known. But whether correlated inputs can beat single-use capacity in the limit remains one of quantum information theory's central open problems. This paper sharpens the boundary by closing off the permutation-invariant route definitively.
