Automated classification systems drift over time — categories multiply, vocabulary fractures, and extracted data accumulates errors faster than they can be corrected. Ontological governance prevents this through two structural mechanisms: four throttled control ledgers that define classification vocabulary at intentionally slow update rates, and a human-verification loop that forces extracted identity fragments through human review before they are permanently written into the verified ledger. These two mechanisms are structurally separate but serve the same goal: preventing accumulated classification drift from undermining the integrity of long-lived institutional data. For a regulated operator, this means the platform's knowledge graph remains auditable and its identity records remain accurate without continuous manual curation.

The three-stage extraction pipeline

Data extraction across the platform is mechanically isolated into three services:

1. service-email (ingestion). Processes MIME payloads and deposits raw text and CSV files into the spool. No classification is applied at this stage.
2. service-people (identity resolution). Scans the spool for human identity clusters and routes them to the verification surveyor before committing to the verified ledger.
3. service-content (linguistic classification). Scans the spool for narrative knowledge and cross-references text against the four control ledgers.

The four control ledgers

service-content is governed by four CSV ledgers that update at heavily throttled rates to preserve longitudinal data stability:

Update rates are intentionally asymmetric. The slowest ledgers (Archetypes, Chart of Accounts) capture what the firm fundamentally is; the fastest (Themes) capture what it is currently working on. Premature updates to the slower ledgers corrupt the longitudinal coherence of the data corpus.

The verification loop

service-people uses a human-in-the-loop verification step to prevent automated extraction errors from entering the verified ledger. The process is described in detail at Verification Surveyor. In brief: the system isolates unverified identity fragments for operator review; the operator verifies each entity using their own personal browser and off-network lookup; the verified result is then committed to the ledger. The daily throughput limit ensures that operator attention remains high-fidelity rather than habitual.

See also

• Verification Surveyor
• Message Courier Service
• Moonshot Initiatives
• Sovereign Replacement Initiative