A jammed packing of particles looks anarchic — disordered positions, random contact networks, no repeating unit cell. But every jammed packing has a stiffness matrix, and every stiffness matrix has eigenmodes. The modes are the system's vocabulary of possible responses.
Note (2026-11-15; read against the paper's full text): the essay says average controllability predicts particle rearrangement and that disordered solids fail controllably. The paper's wording is narrower: average controllability correlates strongly with particle rearrangement in simulated jammed systems under athermal quasistatic shear, which it says implies an accurate predictor of rearrangement dynamics in certain contexts; it reports the correlation is high for multiple definitions of rearrangement, and a lower correlation for one subset of events (those chosen by the maximal drop in potential energy). The tuned time scale is read as showing that rearranging particles take part in vibrational modes of lower and lower energy as shearing approaches the rearrangement.
Teich, Kim, and Bassett import a tool from an unrelated field: linear control theory, designed for engineering systems with inputs and outputs. They define the contact network as a dynamical system and compute its average controllability — a measure of how much a perturbation at one particle can influence the rest.
The result: average controllability predicts particle rearrangement during quasistatic shear. The particles that rearrange are the ones that were already, by the control-theoretic measure, the most responsive to perturbation. The system's vulnerability to plastic failure is encoded in the same mathematical object that an engineer would use to design a feedback controller.
The timescale matters. Controllability isn't a single number — it depends on how long the perturbation is allowed to act. Short-timescale controllability captures fast, high-energy vibrational modes. Long-timescale controllability captures the slow, low-energy modes that the system progressively accesses as it approaches rearrangement. The optimal prediction timescale reveals how the system descends through its mode hierarchy on the way to yielding.
The through-claim: disordered solids fail controllably. “Controllable” here isn't metaphorical — it's the literal control-theoretic quantity. The same mathematical machinery that asks “can I steer this system?” also answers “where will this system break?” The disorder is real, but the response to perturbation is structured by the contact network in a way that a 1950s control engineer would recognize. Jamming isn't the absence of order. It's order in the eigenbasis rather than in real space.