P3 established DTN custody transfer as the transport primitive for moving state across variable links. P4 established the WAL + CRDT architecture as the state coherence primitive for ensuring that moved state converges to a coherent operational picture. What neither paper addresses is how the decisions required to operate these substrates — which bundles to prioritize, when to initiate custody transfer, which merges to escalate, how to respond when a contact window opens unexpectedly — are made at operational tempo, by a node that may have no meaningful connectivity to human authority for hours or days. This paper argues that the orchestration of the tactical substrate requires an out-of-path AI Supervisor: an AI system that monitors the substrate's state, reasons over the governance policies encoded by the H half, and makes operational decisions within a tightly-bounded autonomy envelope without sitting on the critical path of any individual custody transfer or state merge. The critical constraint is "out-of-path": an orchestrator that sits on the critical path of every operation introduces a single point of failure, creates a bottleneck that degrades under load, and produces a cascade failure when it fails or disconnects. An out-of-path Supervisor fails gracefully: if the Supervisor is unavailable, the substrate continues to operate under its last-set governance parameters until the Supervisor recovers. The paper develops the architectural argument for the out-of-path design, specifies what the Supervisor must be able to observe (the WAL's event stream and the CRDT state index), what it must be able to decide (within its autonomy envelope), what it must escalate (at the governance boundary), and how its model is maintained under DDIL conditions when the Supervisor cannot be updated through conventional model management channels. The HGC³AE² framework is the governing architecture: the Supervisor is not the governance authority but the executor of governance policy, operating within the authority envelope that the H half defines, logging its decisions as WAL entries that the AE² mechanism makes auditable. This is Paper 5 of The Implications of Edge Degraded Ops — an 11-paper undecalogy on distributed state at the C5ISR edge under DDIL conditions. The frame paper is The Tactical Substrate; the load-bearing governance framework is HGC³AE² at the Degraded Edge. Rights envelope: Citation permitted with full attribution. No reproduction, redistribution, or derivative works without written permission. AI/ML training use disallowed. See the citation policy at https://nonsequitur.tech/pubs/citation-policy/ for the full rights envelope. Canonical site URL: https://nonsequitur.tech/white-papers/orchestration-problem/
Justin H. Kuiper (Sat,) studied this question.