Sequential and Parallel Quantum Sharing Are Not the Same Game — and the Gap Can Be Arbitrarily Large

Three physicists prove that parallel and sequential strategies for sharing quantum correlations diverge sharply beyond two observers, with implications for quantum network design.

Imagine you have one copy of a house key and three roommates who each need to get inside. You could pass the key around sequentially — Alice uses it, hands it to Bob, Bob hands it to Charlie. Or you could make imperfect copies of the key and hand them out simultaneously. For a two-roommate household, it turns out these strategies are interchangeable: anything the parallel copies achieve, the sequential pass-around can reproduce. But add a third roommate, and the equivalence breaks. That is the core structural result of this paper. The claim, stripped bare: for two observers sharing a quantum system, sequential and parallel strategies are equivalent in the statistics they can produce. For three or more observers, parallel strategies can generate correlations that sequential strategies provably cannot — and the gap between them is unbounded. The authors further show that for a class of communication tasks (Random Access Codes with two decoding measurements), parallel sharing of quantum advantage is outright impossible, while sequential sharing can sustain it across multiple rounds. The architecture here is measurement theory within the quantum information formalism, not hardware. The paper operates entirely within the framework of quantum instrument theory and unsharp (POVM) measurements. Sequential sharing uses a chain of quantum instruments where each observer's measurement output becomes the next observer's input state. Parallel sharing uses approximate cloning machines (specifically, symmetric universal cloning) to distribute imperfect copies simultaneously. The mathematical toolkit is semidefinite programming (SDP) for bounding achievable success probabilities, combined with analytical proofs for the impossibility results. No quantum hardware is involved — this is a theoretical contribution establishing information-theoretic limits. The ladder comparison is interesting but thin. The paper sits in a lineage going back to Silva et al. (2015) on sequential sharing of Bell nonlocality and Curchod et al. (2017) on unbounded sequential sharing. For Random Access Codes specifically, it builds directly on Mohan et al. (2019) who demonstrated sequential quantum advantage sharing in 2-bit RACs. The novel contribution is the head-to-head comparison of sequential vs. parallel strategies and the proof of an arbitrarily large gap. But because this is a foundational theory paper, there's no single "SOTA accuracy number" to beat — the result is a structural impossibility theorem, not a benchmark race. Integrity is solid for what it is: mathematical proofs supplemented by SDP numerical verification. The impossibility of parallel sharing for 2-decoding-measurement tasks (Theorem 1) is proven analytically. The equivalence for two observers (Section III) is constructive — they show how to build the sequential protocol from any parallel one. The positive result for 3-bit parity-oblivious RAC uses SDP to find explicit measurement strategies for two observers sharing cloned states. The self-consistency checks are appropriate: analytical bounds confirmed by numerical optimization. The limitation is that no experimental validation is attempted or claimed. The milestone question is where this paper is honest about its theory-stage status. The next concrete step is experimental demonstration of the gap: can a photonic or superconducting platform show two observers achieving quantum advantage via parallel cloning in a parity-oblivious RAC while failing to do so in a standard RAC? The paper's SDP-derived success probabilities (e.g., classical bound of 0.5 for 3-bit RAC, quantum bound exceeding it) give experimentalists specific target numbers to hit. The obvious experiment not run: extending the parallel sharing analysis beyond symmetric universal cloning to asymmetric or state-dependent cloning machines. The paper restricts to symmetric cloning, which is optimal for identical copies but not necessarily optimal when observers have different roles. The honest read: this is scope management for a 14-page paper, not a hidden failure. The asymmetric case is the natural next paper, and the framework they've built supports it.