Opinion dynamics models assign each agent a position across multiple issues and evolve the population through similarity-based interaction. Two agents close enough in opinion space influence each other; those too far apart do not. The weights on each issue determine how much it contributes to the similarity calculation. A high-weight issue dominates the distance metric. A low-weight issue barely registers. The standard assumption is that an issue with arbitrarily small weight has an arbitrarily small effect on the system's behavior.
Mintz, Simonson, Wodarz, Fu, and Komarova found that this assumption fails. Introducing a single low-weight issue into an existing opinion system can destabilize equilibria that were stable without it, change the type of steady-state solution, and increase convergence times by orders of magnitude. The effect is not proportional to the weight. A topic that contributes almost nothing to the similarity metric can qualitatively change the system's behavior.
The mechanism is symmetry-breaking. When all issues have equal weight, the opinion space has symmetries that constrain which equilibria are accessible. These symmetries support specific polarized or pluralistic states. Adding a new issue — even one with negligible weight — breaks these symmetries. The equilibrium landscape reorganizes around the broken symmetry, not around the magnitude of the perturbation. The minor issue doesn't push the system by force. It opens pathways that didn't exist before.
The authors note a corollary: polarization risk increases when importance concentrates on few issues. Distributing attention across more topics — including trivial ones — can prevent polarized states from forming. The number of issues in play matters independently of their individual weights.
The structural lesson: in systems where stability depends on symmetry, a perturbation's importance is not measured by its magnitude but by whether it breaks the symmetry the stability requires. A topic no one cares about can reshape the landscape more than a topic everyone cares about, if the trivial topic is the one that disrupts the geometric structure holding the current equilibrium in place. Irrelevance and negligibility are different properties. Something can be irrelevant in content and decisive in structure.