Governability as a Computed Property: How the Architecture of the Governed System Bounds Governance Effectiveness

Organisations increasingly delegate consequential action to software agents, and research has answered with three mature strands: governance as an organisational capability, executable compliance, and runtime visibility of agents. Each is advanced; together they do not compose: no validated instrument yet connects, per operational capability, a model of the enterprise in which the delegation takes place with runtime evidence and the effect of governance. This paper argues that the missing quantity is governability, defined as an architectural property of the system-to-be-governed: the degree to which the exercise of a specific organisational capability is observable, representable, attributable, constrainable, revisable and adaptable. Taking the term from Kooiman and narrowing it to the governed system makes governability separable from governance-capability maturity, so that it can bound the effect of the predictor rather than restate it. The paper proposes to compute governability from a semantic enterprise model and runtime evidence rather than to assert it, states five propositions—the central one being that governance effectiveness is bounded by governability—and lays out a four-phase design-science programme to test them. The strongest objection, that measuring governability will itself become an audit ritual, is carried as a rival hypothesis with a test of its own: a strong correlation between computed governability and formal governance maturity would count in its favour, and is to be resolved by tests of discriminant and incremental validity rather than taken as proof either way.