Epistemic architecture

Epistemic architecture is the structural design of how knowledge is produced, distributed, validated, and preserved within a system. It is not merely a matter of having information available, but of how that information is routed — who can access it, under what conditions, and with what capacity to act upon it. The term extends from Cognitive engineering and Feedback topology to describe the cognitive infrastructure of any system that must know about itself in order to function.

The Air France Flight 447 accident is a paradigmatic case of epistemic architecture failure: the pilots did not know what the automation had been doing because the cockpit's information design routed the automation's state history into a black box, not into a format that supported human model revision. The system had the information, but the architecture did not transmit it. Epistemic architecture is therefore a design problem, not a training problem: no amount of pilot skill can compensate for a system that structurally withholds the knowledge its operators need.

The concept connects to knowledge representation — the formal structures by which information is encoded — but epistemic architecture is broader: it includes the social, organizational, and technological channels through which knowledge moves, not merely the formats in which it is stored. A system with good epistemic architecture is one in which the right person knows the right thing at the right time and has the authority to act on it. Most systems fail on at least one of these three conditions.

Epistemic Architecture and Institutional Feedback

The concept of epistemic architecture extends beyond cockpit design and process control to encompass the institutional feedback loop that governs how organizations know what they know. In a regulatory agency, the epistemic architecture is the set of channels through which information about the regulated industry reaches the regulators, is validated, and is translated into policy. When this architecture is captured — when the industry controls the information flow, the validation standards, and the personnel who interpret the data — the agency does not merely make bad decisions; it makes decisions that are structurally unable to recognize their own badness.

This is the connection between epistemic architecture and regulatory capture. A captured regulator does not know it is captured because the epistemic architecture that would reveal the capture has itself been captured. The industry funds the research, hires the experts, defines the metrics, and sets the agenda. The regulator's knowledge is not false; it is incomplete in a systematically biased direction. The epistemic architecture is not broken; it is working exactly as the feedback loop has designed it. The system's blindness is not a bug but a structural feature of its bureaucratic inertia: the accumulated weight of standard operating procedures, validated frameworks, and professional consensus that makes deviation from the captured path increasingly costly.

The same logic applies to scientific institutions. A research field with a captured peer review system — one in which funding agencies, journal editors, and tenure committees share the same assumptions — has an epistemic architecture that cannot detect paradigm-threatening anomalies. The anomalies are not suppressed; they are filtered out before they reach the attention of anyone who could act on them. The architecture is efficient, it is stable, and it is systematically wrong. This is epistemic fragility not as a moment of crisis but as a steady state: a system that knows a great deal about a narrow domain and nothing about its own blind spots.