Who checks a quantum computer's work when it beats every classical machine?

When a quantum machine is said to have beaten every classical computer on Earth, who actually checks the answer? A new arXiv preprint sketches one scalable way to do that checking, for a narrow slice of problems called non-equilibrium dynamics, the fast-changing, out-of-equilibrium systems that classical computers also struggle with. The verification holds only under a stated error model.

The proposal, posted as arXiv:2607.14212, leans on a special class of quantum states called stabilizer scars. These states occupy a middle ground: hard enough to be interesting, structured enough that a classical computer can simulate them in full. That dual nature lets researchers estimate the fidelity of a quantum simulation against a known answer, even as the simulation itself scales beyond what classical machines can keep up with. The trick builds on earlier stabilizer-scar work and on the broader theory of quantum many-body scars, a class of non-thermal states embedded in otherwise chaotic quantum systems.

Under a physically motivated error model, the paper argues, the fidelity measured on these scar-based states also bounds the fidelity of nearby, classically intractable simulations. The grade comes from a problem a classical computer can solve; the result certifies work on a problem it cannot.

The paper proposes a verification tool, built on a narrow class of states. The fidelity bound is conditional, and the work has not been peer reviewed. It does offer one concrete thing: a scalable benchmark for one corner of the quantum-advantage race.