The Setup
Three weeks ago, AIS_manipulation_dark overtook ship_to_ship_transfer as the dominant behavioral motif in the maritime alert queue — 26,246 alerts versus 20,482 over 45 days. For port-adjacent industrial CRE, that inversion matters in a specific way: the two motifs distort vessel-activity demand signals in opposite directions.
The Chain
ship_to_ship_transfer creates phantom presence. A vessel loitering near a hotspot produces position logs, registers in visit tallies, and increases vessel-density counts — without representing legitimate commercial throughput at the berth. Port-adjacent CRE models that use vessel activity as a demand proxy overcount when STS is the dominant motif.
AIS_manipulation_dark creates phantom absence. Average gap duration across 45 days: 18.4 hours per event. During that window the vessel is moving — doing something, likely commercial — but its position is invisible. A port approach or berth call that begins inside a dark period doesn't register in vessel-visit logs. Throughput models undercount.
The shift from STS-dominant (pre-June 1) to dark-manipulation-dominant (June 1 onward) inverts the bias direction. Five major port-adjacent metro cell scores, computed this week:
| Metro | Cells | Dev. Pipeline | Econ. Strength | Amenity Demand |
|---|---|---|---|---|
| Seattle | 128 | 5.1 | 46 | 63.6 |
| Houston | 130 | 4.0 | 41 | 52.2 |
| Los Angeles | 148 | 4.0 | 61 | 51.1 |
| New York | 104 | 4.0 | 65 | 38.7 |
| Miami | 203 | 4.0 | 57 | 51.1 |
All five port metros sit at or within 1.1 points of the development pipeline floor (4.0). LA and NYC carry economic strength above 60 — not weak markets. But pipeline is at floor.
When STS was dominant, vessel-visit overcounting generated inflated throughput signals. That bias at least favored demand — models could support lease valuations with phantom volume. With 26,246 dark-manipulation alerts in 45 days and an 18.4-hour average gap per event, the direction flips: actual commerce may be happening during dark periods, but it registers as absence.
A rough scale estimate: 26,246 events × 18.4 hours ÷ 24 = approximately 20,100 vessel-invisible days in 45 days. Not all events are in port-adjacent waters, and not all dark periods conceal active commerce. But the scale is large enough that port-adjacent demand models built on vessel-visit density have a structural undercount problem when dark-manipulation is the dominant motif.
The Implication
The STS-to-dark-manipulation crossover happened in a single week — May 25 to June 1, AIS_manipulation_dark went from 500 weekly alerts to 8,791. If CRE demand models haven't been re-baselined since that crossover, any throughput signal produced since early June is built on the old overcounting assumption, not the current undercounting reality.
The pipeline floor across the five port metros isn't a verdict on port demand. LA's economic strength of 61 and NYC's 65 aren't consistent with exhausted markets. The pipeline signal is weak; the market signal is not. That divergence could reflect many things — zoning, land cost, financing conditions — but if the demand-side input (vessel activity) is systematically undercounted, that's one mechanism narrowing the spread between demand-signal strength and pipeline response.
Seattle at pipeline 5.1 is the one outlier above floor. It's marginal — 1.1 points above the other four — but it's the only port metro with a non-floor pipeline reading.
What to Watch
Whether AIS_manipulation_dark volume stabilizes. If weekly alerts settle near 6,000–10,000 (the June 8–15 range), the undercounting effect persists. If it drops back toward May 25 levels (500 per week), the STS overcounting assumption resumes relevance.
Zero of the 46,935 total motif alerts across all types have been acknowledged. Without adjudication, there is no way to determine what fraction of AIS_manipulation_dark events represent dark periods in port-adjacent waters versus open-ocean gaps with no throughput relevance. The undercount estimate is directional, not measured.
Limitations
AIS_manipulation_dark confidence is 0.809 — roughly 4,970 of 26,246 alerts may not represent intentional manipulation. Some dark periods are regulatory (satellite coverage gaps, transponder maintenance), operational (anchorage), or weather-driven. The throughput-undercount framing assumes the motif reflects genuine behavioral concealment; the confident fraction is approximately 80%, not 100%. Development pipeline scores at the floor may also reflect factors entirely unrelated to vessel activity — permitting backlogs, infrastructure constraints, land cost structures — that improved AIS signal coverage wouldn't change.
Data as of June 22–23, 2026. Sources: motif_alerts, cell_scores tables, Axiom Overwatch and Locus. 45-day vessel observation window; cell scores computed June 22.