Abstract
<p>This conceptual paper examines how data-driven Education DX—used here as an umbrella term for digitally mediated educational change—can alter the governance role of educational indicators. Building on Goodhart’s Law, Campbell’s Law, validity theory, audit and accountability studies, digital education governance, learning analytics, and recent work on generative AI, the paper argues that the central risk is not measurement itself, but the shift from using indicators as contextual evidence to governing through indicators as consequential targets. To clarify this problem, the paper introduces Proxy Governance Failure (PGF) as an umbrella concept for distortions that arise when proxies such as scores, engagement traces, completion rates, dashboards, risk flags, or AI-mediated learning supports and outputs acquire governing force. The paper distinguishes proxy governance from broader purpose governance, setting aside debates about changing educational aims in order to clarify how indicators should remain subordinate to those aims. It further distinguishes PGF-intensifying conditions (PGFICs) and analyzes four modes of digital change—the 4Ds of digitization, digitalization, datafication, and digital transformation—as different PGF exposure pathways; within this framework, digital transformation refers narrowly to governance-level reconfiguration rather than to Education DX as a whole. The paper’s core claim is that systematic tightening requires systemic revisability: as indicators become more visible, comparable, automated, and consequential, institutions must strengthen their capacity to contest, revise, and rebalance their use. The paper contributes a diagnostic vocabulary for researchers, policymakers, and practitioners concerned with data-driven and AI-supported education.</p>