P3 established that custody-based transport — DTN custody transfer across heterogeneous, intermittent links — is the transport primitive for the tactical edge. Bundles arrive. State is moved. But moving state between disconnected nodes does not guarantee that the state held across those nodes constitutes a coherent operational picture. Two ships that have been operating independently for seventy-two hours, each accumulating sensor fusion data, targeting updates, and planning revisions from their own local sources, will not automatically produce a coherent combined operational picture when they reconnect — even if their custody transfers have been technically correct. They will produce a collision of locally-valid states, each internally consistent, together inconsistent, with no agreed mechanism for resolution. This paper argues that state coherence at the tactical edge is a substrate-level problem, not an application-level concern, and that the two primitives required to address it — the Write-Ahead Log (WAL) for durable, ordered state recording and Conflict-Free Replicated Data Types (CRDTs) for deterministic, convergent state reconciliation — must be architectural first-class components of the tactical substrate, not optional add-ons that individual applications may or may not implement. Without these primitives, the operational picture assembled from state delivered by the P3 transport architecture is a probabilistic approximation of operational truth. With them, it is a deterministically convergent, governance-auditable representation of operational reality. The paper develops the WAL as the primary substrate mechanism for state integrity within a single node: every write is logged before it takes effect, the log is append-only and cryptographically verifiable, and the current node state is fully reconstructible from the log at any point in time. The CRDT layer is then developed as the multi-node convergence mechanism: a class of data structures whose merge operations are commutative, associative, and idempotent, such that merging any two CRDT instances from any two nodes produces the same result regardless of merge order, with no central authority required and no window of connectivity required beyond eventual delivery. Together, WAL and CRDT constitute the state coherence substrate on which the AI Supervisor's operational reasoning (P5), the protocol selection mechanism (P6), and the workload class governance (P7) all depend. This is Paper 4 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/state-coherence-problem/
Justin H. Kuiper (Sat,) studied this question.