Quantum Performance Claims Shift by up to 1.91x When System Boundaries and Normalization Are Fixed

The arXiv preprint performs a systematic audit across quantum computing, QKD, sensing, and clocks, insisting that performance is a joint property of task definition, boundary choice, and comparator rather than hardware alone. It reanalyzes published data to show how correlated errors, interface losses, and duty-cycle effects routinely invert advantage claims once these parameters are matched. The work highlights that LIGO’s 6.1 dB quantum-noise reduction coexists with only 0.534 coincident observing fraction, separating detector gain from delivered service. This approach directly challenges component-level marketing that dominates coverage of quantum roadmaps.

Real-world consequences appear in validation priorities the paper identifies: fabrication yield, calibration covariance, and control latency dominate over headline metrics such as qubit count or squeezing level. Related work in Nature Photonics on photonic sampling and in PRX Quantum on finite-key QKD already documented similar normalization sensitivities, yet mainstream reporting continues to treat raw hardware figures as decisive. The preprint supplies the missing cross-domain taxonomy needed to convert those component records into reproducible system capability.

Credible progress therefore requires reproduced logical workloads, prospectively validated simulations, and repeater links that demonstrably outperform direct transmission under matched availability and lifecycle cost. The framework supplies the measurement protocol to test these conditions before scaling investment.

Next steps include community adoption of standardized boundary definitions in benchmark suites and funding agencies requiring duty-cycle and interface-loss data in all quantum technology proposals.