Quantum sensing pushes measurement precision toward the Heisenberg limit — sensitivity scaling as 1/N² rather than the classical 1/N. The standard approach distributes entangled photons across spatial modes. More modes, more photons, more precision. But the spatial resources are fixed: you have a finite number of sensors, a finite number of arms.
This paper exploits a resource that was sitting unused: time. By entangling measurements across repetitions — temporal multiplexing — the authors achieve Heisenberg scaling simultaneously in photon number N, spatial modes M, and measurement repetitions R. The sensitivity approaches 1/(NMR)², a triple product that previous protocols never achieved because they treated repetitions as classical averages.
Note (2026-11-11): checked against the paper's full text. The scaling, the optimality within Gaussian states, the homodyne readout and the persistence under loss match the paper. Two statements go beyond it. (1) The essay says the authors 'achieve Heisenberg scaling simultaneously' in photon number, spatial modes and repetitions. The paper says the sensitivity can asymptotically approach simultaneous Heisenberg scaling: in the large-repetition limit the error approaches the 1/R^2 form, and the time-multiplexed scheme approaches Heisenberg scaling in the number of repetitions as R increases (the scaling is in the variance of the phase estimate). (2) The essay's 'temporal entanglement is free' and 'doesn't require new hardware' is not stated. The paper's scheme injects squeezed vacuum at each time step and it proposes a loop-based photonic sensing architecture as an experimentally feasible realization; I found no statement about hardware cost. It is a theoretical proposal, with no experiment reported.
The protocol is experimentally feasible: it requires homodyne detection and a looped photonic architecture, and the advantage persists under optical loss. The Bogoliubov transformation formalism proves optimality within Gaussian states — you cannot do better with this class of quantum light.
The insight is that temporal entanglement is free. Every previous distributed sensing protocol entangled across space but averaged across time, discarding temporal correlations that carry metrological information. Repetitions are not just statistics — they are quantum resources. Using them as such doesn't require new hardware. It requires recognizing that the dimension you were averaging over was the dimension you should have been entangling across.