Epistemic Failure in Scientific Knowledge Transfer: The Absence of End-to-End Responsibility

This paper examines a structural epistemic failure mode in scientific knowledge transfer that arises when locally valid knowledge claims are composed into decision-facing outputs without end-to-end epistemic ownership. Treating knowledge production as an epistemic pipeline coupling measurement, analysis, interpretation, and institutional handover, the paper shows that local methodological validity does not in general guarantee recoverability or interrogability of the composed claim under delayed or adversarial scrutiny. The failure identified is epistemic rather than methodological: it can occur without demonstrable error in any individual component of the pipeline. The vulnerability persists because epistemic pipelines are not treated as first-class epistemic objects with their own end-to-end failure modes. The paper is intentionally diagnostic rather than remedial and does not propose governance mechanisms, policy reforms, or procedural fixes. This work is intended as a foundational analysis of epistemic responsibility in knowledge transfer and serves as a conceptual anchor for subsequent formal and applied investigations.