The Protocol Selection Problem: PACE-Plan Workload-Class Routing Under Disrupted, Disconnected, Intermittent, Limited-Bandwidth Conditions

P3 established DTN custody transfer as the transport primitive. P4 established WAL + CRDT as the state coherence primitive. P5 established the AI Supervisor as the out-of-path governance layer. What none of these papers specifies is how the substrate selects the correct transport mechanism for each operation in each link condition. The DTN custody layer offers multiple convergence layer adapters, multiple routing algorithms, multiple bundle prioritization schemes, and multiple delivery confirmation modes. The CRDT merge layer offers full-state merge and differential update modes. The contact scheduler offers predictive staging and opportunistic forwarding. Each of these choices has operational consequences: the wrong selection wastes scarce link capacity, violates latency requirements, or degrades state coherence guarantees. This paper argues that protocol selection at the tactical edge is not a configuration problem but a runtime governance problem. The correct protocol for any given operation depends on three variables that change continuously: the current link conditions (bandwidth, latency, error rate, predicted window duration), the workload class of the data being transmitted (latency-sensitive targeting data has different protocol requirements than bulk historical state synchronization), and the current governance posture (a classification conflict in the pending-custody queue changes the relative priority of the governance policy update bundle). Static protocol configuration — choosing a protocol at deployment and running it until the next maintenance cycle — cannot accommodate this three-variable runtime dependency. The paper develops a protocol selection architecture in which the AI Supervisor evaluates protocol options against runtime link conditions and workload class requirements, applies governance policy to constrain the selection space, and selects the mechanism that maximizes delivery effectiveness within the governance bounds. This selection is not ad hoc optimization; it is a governed decision recorded in the WAL with the same fidelity as any other Supervisor decision. The §6 Governor Application specifies which protocol selection decisions the Supervisor may make autonomously, which require human governance input, and how the AE² mechanism enforces protocol discipline when link conditions create pressure to use unauthorized shortcuts. This is Paper 6 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/protocol-selection-problem/