Abstract Stop-and-go waves in vehicular traffic are commonly explained as a linear collective instability induced by, e.g., response delays. We explore an alternative mechanism that more faithfully mirrors oscillation formation in dense single-file traffic. The fluctuating drivers' responses, modeled by noise, play a key role in this model; as noise is increased, the base (uniform) flow abruptly switches to stop-and-go dynamics despite its unconditional linear stability. The stop-and-go motion persists only as long as noise is present. We rationalize the instability mechanism in quantitative terms by breaking down the noise into oscillatory driving terms that can be studied separately. An analogy with Kapitza's inverted pendulum is hinted at.
Dufour et al. (2026) studied this question.