Trace Origin is a structural intelligence practice. It builds formal graph representations of relational data from public records, applies geometric and statistical methods to surface patterns that individual records do not carry, and produces evidence artifacts whose every claim traces back to its source.
The founding premise: concealment and pattern do not live in the labels a database records. They live in the structure the labels form. A leaked file names an individual; a properly measured ownership graph reveals a fourteen-company control loop that no single filing describes. A charge sheet lists an offense; a properly measured prosecutorial graph reveals which regime the defendant was assigned to before any evidence was heard. In both cases the method is the same: build the graph from records anyone can pull, measure its geometry, reason from the measurements with a formal ontology, and preserve provenance on every derived claim.
Method
The practice combines four disciplines that most firms use in isolation:
- Graph construction from public records. Structured extraction from company registries, land records, sanctions lists, court dockets, prisoner registries, and adjacent sources. No leaks required.
- Geometric analysis of relational structure. Hodge decomposition to separate hierarchy from circulation, Forman-Ricci curvature to locate structural bridges, persistent homology to detect layered loops across scales. Where standard tools query the graph, these methods measure it.
- Ontological reasoning with provenance. A formal vocabulary (OWL, SKOS, PROV-O) unifies heterogeneous sources; a reasoner derives new facts from stated ones by fixed, inspectable rules; every derivation is reproducible edge by edge.
- Statistical validation against null models. A finding earns the word "finding" only after clearing a permutation test or hypergeometric threshold that distinguishes signal from what chance alone would produce.
The output is not a black-box score. It is a defensible chain of reasoning, with confidence intervals, calibrated to ground truth, red-teamed before delivery.
Applications
The method applies wherever a body of public records forms a graph whose shape carries information the labels do not. Two current application areas:
- Beneficial ownership and financial-crime investigations. Concealment structures in corporate ownership networks. Nominee resolution across jurisdictions. Sanctions-adjacency detection. Engagements in this area are conducted under confidentiality.
- Human rights accountability. Systemic patterns in state prosecutorial and detention practice. Corpus-level evidence for treaty-body submissions.
Other application areas the method fits and the practice is open to engaging: procurement corruption networks, litigation-funding structures, enforced-disappearance registries, and supply-chain concealment. The underlying question is always the same: what does the shape of the network reveal that the individual records do not?
Founder
Eric Brattin runs the practice. Independent researcher, US-based, no institutional affiliation with any government, opposition organization, or aligned entity. Trained in ontology engineering; current work extends into algebraic topology as applied to concealment detection in relational data.
What the practice does not do
- Direct advocacy campaigns
- Analysis of specific individuals outside publicly documented cases
- Work that would compromise the anonymity of at-risk persons
- Work adverse to independent journalism or human rights documentation
Contact
Eric Brattin
ebrattin@traceoriginresearch.com
www.linkedin.com/in/ericbrattin
Response within two business days. For engagements involving confidential material, please indicate as much in your first message and we will move to a secure channel before discussing specifics.