Corrigibility as a Structural Precondition for Digital Public Infrastructure: A Cybernetic Framework

Digital Public Infrastructure (DPI), including identity systems, payment rails, data exchanges, and increasingly AI-based services, is being deployed globally at population scale. International definitions from the G20 and United Nations articulate desirable attributes (open, inclusive, accountable) without specifying structural conditions under which these attributes can be verified or enforced. This paper introduces a formal framework for evaluating DPI based on corrigibility: the structural capacity of those affected by a system to detect error, signal harm, and trigger correction without incurring material loss. We derive five necessary tests: EXIT (safe refusal), CODE (legible execution), AUDIT (independent verification), GOVERN (constitutive input), and FORK (credible replacement). These emerge from three independent theoretical traditions: cybernetics (Ashby's Law of Requisite Variety), commons governance (Ostrom's design principles), and free softwareprinciples. We prove that failure of any single test disqualifies a system from DPI status and extend the framework to learned systems (AI), where verification shifts from inspecting execution to bounding behavior. Empirical application to Aadhaar, UPI, Pix, and open-weights AI models demonstrates the framework's discriminative power. Positive examples (Linux, Let's Encrypt, Wikipedia) show corrigibility is achievable at scale. The framework is released under CC0 public domain to enable universal adoption. Framework homepage. | Theoretical analysis | Official definitions critique | JSON Schema. | Author