AI-Data Property ↔ Tier-3 Architecture Mapping

A four-property mapping framework that connects the data requirements of defensible AI development to the architectural properties of Tier 3 preservation, producing the structural argument that the same infrastructure investment serves both purposes.

Property 1 — Provenance. The ability to establish what data trained a model and where it came from is the structural product of content addressing and signed deposit. A content-addressed corpus is identified by the hash of its contents; signed deposit attests authorship. Both properties are byproducts of M-0002 (Three Architectural Principles).

Property 2 — Reproducibility. The ability to re-run a training pipeline against the original corpus requires that the corpus persist across the lifetime of any model trained on it. This is the preservation property that the Four-Tier Taxonomy (M-0001) shows is structurally produced only at Tier 3.

Property 3 — Federation. The ability to train across institutional boundaries without consolidating sensitive data into a single trust domain is the operational pattern of permissioned BitTorrent, federated Matrix, and permissioned IPFS clusters. It is the same architecture that HIPAA-covered, FERPA-covered, and export-controlled research already requires (C-0015).

Property 4 — Verification. The ability to demonstrate to a regulator, court, or peer reviewer that the training data was what the model card claims it was is the architectural property developed in C-0007 — a single cryptographic query across the distribution network produces evidence that any third party can independently re-verify.

The framework is the analytical instrument that turns a list of AI-data needs into a one-to-one architectural mapping onto Tier 3 properties already developed elsewhere in the paper. C-0008 and C-0039 invoke it to argue that the AI strategic position is captured as a structural byproduct of the same Tier 3 deployment that closes the §5 liability.