Orientation Profiles, Argument Maps, and Aporia Checkpoints: A Three-Layer Architecture for Substrate Governance of AI-Assisted Intelligence Analysis

Structured Analytic Techniques govern the operator’s cognition. They cannot govern the model’s implicit orientation function—the accumulated configuration of task stance, authority stance, and risk stance that shapes every output and drifts across turns and sessions without detection. This paper defines orientation drift as three coupled failure modes—task stance drift, authority stance drift, and risk stance drift—and proposes a three-layer architecture for governing it: a typed orientation profile owned by the substrate, a calibrated argument map maintained by the operator, and mandatory aporia checkpoints that force the system to surface named contradictions and generate structurally distinct alternative scenarios before any claim certainty can be upgraded. The architecture is instantiated against a concrete intelligence problem—a PIR on structural realignment along the Algiers–Tripoli–Sahel–Khartoum corridor—demonstrating how substrate governance rules for Blind nodes prevent false convergence, enforce scenario branching, and make reasoning chains auditable and reproducible across sessions.