Verification surveyor
TopicFrom the PointSav Documentation
The Verification Surveyor is the human-in-the-loop component of service-people that presents extracted identity fragments to an operator for manual verification before they are permanently committed to the verified ledger.
The Verification Surveyor is the architectural checkpoint in service-people that prevents automated extraction errors from compounding by requiring a human operator to confirm each identity fragment against an off-network source before it is committed to the verified ledger.
[edit]Key Takeaways
- Every identity fragment extracted by
[[service-people]]is held as unverified until a human operator confirms it against an external directory source using their own personal browser and personal account. The platform never initiates the external lookup itself. - Hard limit of 10 verifications per operator per day — a deliberate quality-control constraint, not a capacity ceiling. High-volume mechanical approval at speed produces habitual confirmation; the limit forces deliberate attention on each record.
- The air-gapped external-lookup design avoids three operational risks: no persistent foreign API tokens required, no per-query costs, and no exposure to rate-limiting or IP-ban from external directory services.
- At 10 verifications per day, roughly 3,650 confirmed institutional relationships accumulate per operator per year with negligible error rate — the throughput constraint is part of the data-quality guarantee.
Unsupervised extraction algorithms accumulate errors. A fully automated ingestion pipeline processing large volumes of email will inevitably produce false positives — an "Unsubscribe" link parsed as a person's name, a role title extracted from a footer rather than a bio. The Verification Surveyor is the deliberate architectural bottleneck that forces all extracted identity fragments through a human cognitive filter before they are permanently written into the verified ledger. The design accepts the throughput cost in exchange for high-fidelity, long-term institutional data.
[edit]Human-in-the-loop philosophy
Every identity fragment extracted by `service-people` is held as unverified until an operator confirms it. The Surveyor presents the fragment — the extracted text, the inferred entity type, the source context — to the operator. The operator then looks up the individual using their own personal browser and their own personal account on an external directory (such as LinkedIn), confirms the entity, and pastes the verified URL back into the terminal. The platform never initiates the external lookup itself.
This design is deliberate. API-based lookups against external directories would require persistent foreign tokens, incur per-query costs, and expose the platform to rate-limiting or IP-ban risk. The air-gapped approach avoids all three.
[edit]Daily throughput limit
The Surveyor enforces a hard limit of ten verifications per operator per day. The limit is not a capacity constraint; it is a quality control mechanism. High-volume data entry at speed produces habitual approvals rather than genuine verification. Ten careful verifications per day produce roughly 3,650 confirmed institutional relationships per year with negligible error rate. The scarcity is structural: it transforms what would otherwise be a mechanical clearing task into a deliberate, high-attention operational step.