A Systems-Theoretic Framework for Governing Concentrated Power: From Private Debt Stocks to Sovereign Compute

Contemporary governance institutions exhibit a recurrent failure mode: they optimize for short-term flows (GDP growth, quarterly earnings, model releases) while neglecting the accumulation of structural stocks (private debt, atmospheric carbon, sovereign compute capacity) that determine long-term stability. This paper synthesizes a unified constitutional framework addressing this systemic blind spot. Drawing on the heterodox macroeconomic critique of Steve Keen and extending it to the domains of artificial intelligence and transnational infrastructure, we propose a gradient-based governance architecture. The framework introduces a multi-dimensional Systemic Exposure Index, a proportional oversight continuum, and a transnational treaty mechanism to prevent regulatory arbitrage. We formalize the condition for systemic stability as Veffective = Vspeed × Qintegrity , where Veffective must match or exceed the velocity of capability accumulation. Crucially, we embed mechanisms for epistemic pluralism, structural dissent, and sunset clauses to ensure that the governance system itself remains corrigible and immune to capture. The paper concludes that durable legitimacy in the 21st century depends less on predictive certainty and more on institutional learning velocity.