The Setup
The temporal graph carries 43,834 COMMERCIAL_MANAGER edges, one for each of 43,834 distinct vessels, first appearing July 9, 2026 and most recently October 6, 2026 — 89 days of ingestion. Not one of the 43,834 has a valid_to timestamp. Closure rate: 0%.
The Chain
Set against the other ownership and management relation types tracked in the same graph over the same period: OWNED_BY carries 51,831 edges, with 4,629 closed — an 8.93% closure rate. SHIP_MANAGER carries 51,262 edges, with 2,171 closed — 4.24%. COMMERCIAL_MANAGER's 0% sits beside ISM_MANAGER's 0% on a comparably sized base of 43,621 edges. The graph splits cleanly along one line: relations describing who owns or who technically (ISM-code) manages a vessel show real closure activity over time, consistent with vessels being sold, reflagged, or re-registered under new management. Relations describing who commercially manages or who ISM-manages a vessel show none, across tens of thousands of edges and three months of data.
The Implication
If a vessel's commercial manager changes, the current graph has no observed mechanism for closing out the prior COMMERCIAL_MANAGER assignment — a new edge for the new manager would presumably get created, but the old one has no recorded end. For any analysis that depends on knowing who currently manages a vessel versus who managed it in the past — sanctions screening, counterparty risk work, ownership-chain reconstruction — a relation type with a 0% closure rate over 89 days can't distinguish "this commercial-management relationship is still active" from "it ended and the graph never recorded the end." The comparison to OWNED_BY's nonzero rate matters specifically because it rules out "this graph just doesn't do closures" as the explanation. It does, for that relation type — just not for this one, nor for ISM_MANAGER.
What to Watch
Whether a single COMMERCIAL_MANAGER edge closes in the next ingestion cycle. As of this snapshot, across 43,834 edges and 89 days of history, the answer is none. If ISM_MANAGER — also at 0% closure, also tracked here as a benchmark — begins showing nonzero closures in a future pipeline update before COMMERCIAL_MANAGER does, that would suggest the two relation types are being populated by genuinely different source logic rather than sharing one unfixed gap. Worth checking again at the 180-day mark, when a longer window will better separate "long-lived by nature" from "never closes."
Limitations
A 0% closure rate is consistent with several explanations this snapshot can't separate: a source-data limitation, where the underlying feed simply doesn't carry commercial-manager-change events the way it carries ownership-transfer events; a short observation window, where 89 days may be shorter than the typical tenure of a commercial management contract, making the true rate of change low rather than zero; or a pipeline gap specific to manager-type relations that OWNED_BY and SHIP_MANAGER don't share. No independent vessel registry was checked against this sample to confirm whether any of the 43,834 vessels have actually changed commercial manager within the window — this piece describes what's recorded in the graph, not what happened in the world. The 43,621-edge ISM_MANAGER figure is cited here as a benchmark from the dataset's current state, not as an independently re-verified number for this snapshot.
Data as of 2026-10-06. Source: temporal_edges. Axiom Overwatch.