The Vessel Ownership Graph Has Produced Zero New Records in 34 Days. Every Active Temporal Edge Is Now an Obfuscation Event.

The Setup

Vessel identity records — ownership, management, flag state — flow through a temporal graph. Each relationship type gets its own edge class, timestamped with when it opened and closed. Six structural types anchor the graph in the current dataset: SHIP_MANAGER (8,139 edges), OWNED_BY (8,137), flagged_as (7,741), renamed_to (986), mmsi_changed (657), and shell_hop (406).

The graph has two modes. Structural edges describe stable relationships: who manages the vessel, who owns it, what flag it flies. Obfuscation edges describe changes to identity: the vessel changed its name, its MMSI, or its beneficial ownership entity. Both types are necessary for sanctions screening and cargo attribution. They come from different upstream sources and update at different cadences.

As of June 16, the graph is running on one mode only.

The Chain

SHIP_MANAGER, OWNED_BY, and flagged_as edges have produced zero new records since May 12 — 34 days of silence across all three structural types simultaneously. The 8,139 SHIP_MANAGER edges, 8,137 OWNED_BY edges, and 7,741 flagged_as edges are all dated between April 3 and May 12. Nothing after.

Identity-obfuscation edges have not stopped. Since May 12:

• renamed_to: 422 new edges (107 in the week of June 8 alone)
• mmsi_changed: 133 new edges
• shell_hop: 123 new edges
• Total: 678 obfuscation events across 34 days

The weekly cadence is not a spike — it is persistence. Week of May 18: 249 obfuscation events. Week of June 1: 146. Week of June 8: 170. The partial week of June 15 shows 14 more already posted. The obfuscation pipeline is running at roughly consistent volume since structural edges went dark.

The ratio is stark: in the last 30 days, 100% of new temporal edges are identity changes. Zero ownership updates. Zero flag updates. Zero management updates.

The two classes of edges behave like separate pipelines — because they are. Shell hops, MMSI changes, and renames can be detected from behavioral signals: entity linkage across IMO continuity, identifier-swap pattern matching. They do not require a registry push from an external source. Ownership and management updates typically do. That source, for all three structural types, appears to have stopped on the same day.

The Implication

The operational consequence is asymmetric opacity. Anyone querying the vessel graph today gets live data on every entity that has changed its identity in the last month — and gets stale data, frozen at May 12, on who actually owns, manages, or flags every vessel involved in those changes.

The graph shows you the mutations. It cannot currently tell you where the mutated entity landed in the ownership tree.

For the 406 total shell_hop edges in the dataset (123 in the last 30 days): each hop represents a vessel whose beneficial ownership entity changed. The prior entity is recorded. The destination entity is not confirmed via a structural edge — because structural edges have not updated. The shell hop is detected; the new owner is inferred, not registered.

What to Watch

The May 12 cutoff is the number to track. If SHIP_MANAGER and OWNED_BY edges resume with data dated after June 16, the gap was a registry ingestion lag and will backfill. If the cutoff remains at May 12 on the next observation, the gap is widening and the structural layer is not recovering passively.

Shell_hop rate is the secondary signal. The current rolling weekly cadence ranges from 10 to 72 hops per week. A sustained drop below 5 per week would indicate either systemic suppression across the entire obfuscation pipeline or a genuine slowdown in the underlying behavior — distinguishable by whether renamed_to and mmsi_changed also drop simultaneously.

Limitations

It is not possible from this dataset alone to determine whether the ownership graph freeze reflects a registry source disruption, a change in ingestion frequency, or a genuine halt in regulatory filings from the tracked fleet. The freeze affects all three structural edge types simultaneously — SHIP_MANAGER, OWNED_BY, and flagged_as — which is consistent with a single shared upstream registry source failing. It is equally consistent with a coordinated regulatory filing pause in the April–May window, with ingestion simply caught up to the available data.

The 678 obfuscation events are behavioral signal and are independent of the registry source. The structural freeze is a data-state observation, not a confirmed outage. The two facts coexist without one causing the other.

Data as of 2026-06-16. Sources: temporal_edges (SHIP_MANAGER, OWNED_BY, flagged_as, renamed_to, mmsi_changed, shell_hop relation types), full dataset and 30-day rolling window. Total edges: 25,967.