🎰 Rigged-Gear Detector

Can you catch a rigged mechanism by watching only one gear? A geared Brownian machine, live.

Demonstrating the result of "Force-Torque Reciprocity and the Inference of Concealed Dissipation in a Geared Brownian Machine" (arXiv:2607.06131). Two overdamped rotors share one periodic cam potential. A visible dial (all you can measure) is geared to a hidden rotor sealed inside the machine. Because the shared potential obeys force-torque reciprocity (τ_int = −ℓ·F_x), the paper proves you can reconstruct the hidden rotor's entropy production — its concealed dissipation — exactly from the visible dial's statistics alone. No timescale-separation assumption; the usual information-flow correction terms vanish identically. Every number below is computed live from a real coupled-Langevin simulation, not faked.

The visible dial is all a regulator can watch. The hidden rotor is sealed — secretly cheated with an extra torque. Drag the slider to rig it harder. The inference engine estimates the hidden rotor's concealed dissipation from the dial only, and plots it against the true value (which you're not supposed to be able to see). Hit Reveal to confirm the estimate was right.

The Cheat — hidden drive torque τ_drive

0 = a fair machine (thermal equilibrium, zero net dissipation). Turn it up to secretly drive the concealed rotor.

FAIR
Inferred hidden work rate 0.00 /s — consistent with an honest mechanism.

Concealed dissipation (hidden rotor)

Inferred (dial only)
0.000
True (sealed rotor)
0.000
Reconstruction error
—
Inferred secret torque
0.00

What the dial reveals

Dial mean drift ⟨αΊ‹⟩
0.000
Inferred hidden torque ⟨τ_int⟩
0.000

Via reciprocity ⟨τ_int⟩ = −ℓ·⟨F_x⟩, the sealed rotor's mean torque is read straight off the visible dial — without ever opening the box.

Inferred vs. true — converging live

inferred (visible only) true (hidden) analytic target