We characterize chronic doomscrolling as a stable wrong attractor in a threat-vigilance dynamical system. The baseline attractor—which we call S^0, calibrated informed agency—represents a state in which information consumption is proportional to actionability: the agent monitors the world at a rate commensurate with their capacity to respond. Doomscrolling, denoted W, is a qualitatively distinct attractor in the same system, maintained by the news-feed-compulsion architecture—an engineered feedback loop that exploits ancestral threat-detection circuitry to sustain engagement far beyond informational return.
We identify nine propositions, P1 through P9, characterizing the structure of this attractor, its basin geometry, the constructive-action foreclosure mechanism by which W becomes self-sustaining, and the perturbation paths available for return to S^0. The core finding is that doomscrolling is not a failure of willpower or a bad habit but a wrong-attractor problem: the architecture selects for it, maintains it against perturbation, and forecloses the constructive actions that would dissolve it.
Return to S^0 requires not merely reduction in scrolling time but basin-exit: a shift in the attractor landscape itself, accomplished through behavioral scheduling, deliberate information rationing, and reactivation of constructive-action pathways. We close with implications for individual practice and platform design.
Phronesis