Governed Attention: A Transaction-Cost Theory of Repository Usage Statistics and the Measurement of Research Significance

This article argues that repository “views” and “downloads” are not direct measures of scholarly significance but institution-specific accounting outputs shaped by governance choices. Using transaction cost economics, it models archives as governance structures that manage uncertainty, opportunism, identity/versioning, and measurement error. Two platform primitives—routing (ρ(t)), how attention reaches a paper over time) and a measurement operator (Mp(⋅) what gets counted)—explain why some platforms show early “zero windows” while others register immediate baseline activity. The paper adds an object-level correction by defining information content measures—overlap/centrality (Oi), novelty (Ni), and brokerage (Bi)—so “significance” can be anchored in intellectual structure rather than clicks. It concludes with testable propositions and a matched multi-platform empirical strategy focused on time-to-first-count, early-zero probability, activation change-points, and attention dilution from identity fragmentation.