Track 3 · closure pilot · revision 2026-09-27

Ambiguous-Load Closure Pilot

Three cases selected from the registry's 141-facility genuinely ambiguous tier are taken through a bounded closure workflow: queue record → entity resolution → defensible location → public-record search → physical observation → transformer/electrical check → adjudication. This is the part of the project intended to demonstrate that the registry can produce a conclusion, not merely maintain a large schema.

What this proves: the workflow can move ambiguous public load records to documented, bounded dispositions. What it does not prove: detection completeness, facility-wide certification, covert-compute detection, or that these three cases are representative of the full 141-case population.
3
pilot cases
All drawn from the genuinely ambiguous tier.
12
derived observations
Six each for the two defensibly locatable sites.
2
bounded classifications
One manufacturing; one proposed data center.
1
inconclusive closure
Stopped at the physical-location gate rather than inventing a coordinate.

Case matrix

CaseQueueMWDispositionPhysical layer

Why the inconclusive case matters

Project IQ197 is not counted as a failed investigation. The public record supports a queue entry and a legal-entity/property link, but the review did not establish a defensible facility coordinate or the load's end use. The acquisition runner therefore returned a terminal PHYSICAL_TARGET_UNRESOLVED state instead of substituting a city or zone centroid. That is a useful closure outcome because it demonstrates the stopping rule under uncertainty.

Method boundary

Queue/interconnection data are planning and electrical-intent signals. Satellite and related physical observations are physical-realization signals. Neither channel independently establishes tenant, chip inventory, workload, or covert activity. Transformer searches record public evidence and public non-findings separately; a missing public transformer record is not evidence that no transformer exists.