A Science of Governability for Autonomous Systems: Toward a Foundational Discipline for Governable Autonomy

As autonomous systems become increasingly capable, adaptive, interconnected, and persistent, the challenge of governance extends beyond regulatory compliance and operational safety. Existing disciplines such as safety engineering, reliability engineering, cybersecurity, systems engineering, and AI assurance address important dimensions of system performance, yet none directly focuses on the persistent ability to influence, constrain, supervise, verify, modify, and legitimately control autonomous behavior throughout a system's operational lifecycle. This paper argues that governability constitutes a distinct and fundamental system property that may warrant the emergence of a distinct scientific discipline: Governability Science. We define governability as the sustained capacity to maintain legitimate and effective control over autonomous systems under evolving operational conditions. Building on prior work in runtime governance, governance infrastructure, continuous governability, governability economics, and the governability gap, the paper proposes a foundational framework for the scientific study of governability. The paper identifies principal domains of Governability Science, including governability dynamics, governability metrics, governability verification, governability infrastructure, governability economics, governability failure science, and governability institutions. It further introduces preliminary principles describing how governability emerges, degrades, and can be preserved within autonomous ecosystems. The central thesis is that governability may become as fundamental to the autonomous age as reliability became to the industrial age and cybersecurity became to the digital age. As autonomy evolves into critical economic and societal infrastructure, Governability Science may emerge as a foundational discipline for ensuring the long-term legitimacy, controllability, and sustainability of autonomous systems. Keywords: Governability Science; autonomous systems; runtime governance; governance infrastructure; continuous governability; governability metrics; governability failure; AI assurance; EU AI Act; safetycritical autonomy