In the era of superintelligent systems, societies may face a novel form of systemic self-destruction: economic autophagy. We define economic autophagy as a regime in which highly optimized, algorithmically controlled institutions consume their own structural foundations—human interpretive labor, redundancy, and slack—in pursuit of short-term efficiency. Drawing on feedback control and non-equilibrium thermodynamics, we propose a gain–delay model in which optimization gain (G) accelerates faster than corrective feedback capacity (D), inducing a phase transition from stable growth to structural collapse. We formalize resilience as a stock variable (Rt), provide operational proxies suitable for public disclosures, and propose empirical procedures to estimate the critical damping constant (kcrit). We illustrate the framework with testable predictions in software QA (SHIFT Inc.), high-frequency trading, and just-in-time supply chains. We further propose practical countermeasures—engineered friction, interpretive incentives, and minimum resilience buffers—to operationalize safe bounds for AI deployment. Author's Note: This paper originated from a simple consumer frustration: GPU and memory prices are too high! What started as a complaint evolved into this theoretical model.
Explorer Harvey (Thu,) studied this question.