Behind the wall, an invisible machine is running. We cannot see the gears, but we hear the floor and the ceiling respond to its vibration. If the floor hums, the ceiling answers with a delay, and if the ceiling hums, the floor answers differently. This asymmetry of response reveals that something inside is rubbing and heating up.

A similar situation arises in the physics of small systems, for instance molecular motors. Thermal noise prevents us from measuring every detail precisely, and the instrument registers only coarse states: "working" or "idle." How much energy is dissipated as heat is not directly visible. The authors of the paper found a way to estimate a lower bound on this dissipation from the asymmetry of the coupling between two observable quantities at different moments in time.

For a steady state, when the flows are constant, the estimate works for any time interval between measurements. For changing processes, it holds when the two moments nearly coincide. In the examples, the estimate captures up to 60% of the true dissipation in a network of four states and about half for a particle on a ring, where only half the path is visible. This approach is useful in biophysics, optical traps, and even for financial time series.