In the lab, two measurements let hidden information yield work

Imagine a particle held by a tightly focused laser beam, jiggling in a liquid that is not water but a dense, spaghetti-like network of microscopic structures that rearranges around it. That is the stage physicists at the University of Konstanz and the Max Planck Institute for Dynamics and Self-Organization in Göttingen built to test an old idea in a new way.

Maxwell's demon is an imagined microscopic gatekeeper, introduced more than 150 years ago, that uses information about molecules to sort them: it opens a small door for fast molecules in one direction and slow ones in the other, and one compartment grows hotter while the other cools. The apparent paradox was resolved by recognizing that information is a physical resource that can be converted into work. That insight has fed interest in information engines, microscopic machines powered by information rather than by temperature differences.

The catch is that no demon has complete information. Even for a single particle, position and velocity cannot be measured with perfect accuracy. And in the fluids where microscopic machines are expected to work, such as polymer solutions and biological fluids, the internal microstructure shapes the particle's motion while staying out of reach. As the particle moves, it deforms the network, and the network pushes back. The particle can be tracked precisely, but the network's internal dynamics cannot be observed directly, so part of the information needed for optimal control stays hidden and the extractable work drops.

The researchers recover part of that hidden information with a double measurement: they record the particle's position twice, separated by a controlled time delay. Comparing the two snapshots gives indirect information about the state of the surrounding medium, extending the demon's knowledge beyond a single moment. By tuning the delay, the team shows that energy stored in these hidden dynamics can be harvested and turned into useful work. Under suitable conditions, the double measurement allows more work to be extracted than a single measurement would. The reporting does not give a figure for that gain.

For now, the work is a step toward cyclic microscopic information engines, which could run without a temperature difference or any other large-scale energy gradient. That matters most at very small scales, where stable thermal gradients are increasingly hard to maintain. The beneficiaries would be the microscopic machines expected to operate in complex fluids, provided the principle carries from a trapped particle in a lab liquid to a working device. The reporting gives no timeline for that, and the experiment itself is a demonstration, not a machine.