How DEPLOY assigns verification posture

What posture is

Posture is the structural classification we assign to each substantive claim. It says where the evidence lives, what confidence level we hold, and what specifically anchors the assertion.

Posture is per-claim, not per-record. A company record can carry one claim at verified posture (the founding date per incorporation filing) and another at reported-not-verified posture (the current employee count per maker press release). The two claims live on the same record under different posture markers.

Posture flows through to every surface. A claim that resolves at verified posture on the registry surfaces at verified posture on the consumer property and at verified posture in news editorial coverage. A claim that resolves at reported- not-verified posture surfaces with the cap-flag visible at every property.

The eight source-quality tiers

The source-quality tier classifies the evidence type.

Tier 1: primary-government-record. FDA databases, NHTSA Part 573 records, ITC dockets, SEC filings (10-K, 10-Q, 8-K, S-1, Form 4), federal court records, state-level corporate registry filings, FAA certification records. The government record is the anchor; the government is the source.

Tier 2: counterparty-disclosed. A customer of record confirming a deployment on the customer's own surface; an independent third-party auditor's record; a partner disclosing a partnership in the partner's own filings. The counterparty is not the maker; the counterparty has its own motivation to surface the claim accurately.

Tier 3: independent-auditor. A regulator-commissioned inquiry, an independent safety auditor, an independent court- ordered investigation. The independent auditor is a structural-third-party verification surface.

Tier 4: maker-disclosed at IR depth. SEC-filed maker statements, audited investor-relations disclosures, prepared investor calls. The maker is the source; IR-depth disclosures carry audit weight that maker press releases do not.

Tier 5: maker-disclosed at press-release depth. The maker's own press release, blog post, executive talk, demo video. The maker is the source; the assertion is not audit-weighted.

Tier 6: tier-1 news independent verification. Bloomberg, Reuters, the Wall Street Journal, the Financial Times, the New York Times, IEEE Spectrum, FreightWaves and equivalent outlets with documented editorial verification practices that confirm the underlying claim against a primary source. The news outlet's source-discipline anchors the claim.

Tier 7: trade press. Trade-press outlets that cover the space but do not consistently confirm against primary sources. The trade-press outlet's claim is editorially meaningful; it does not anchor verification on its own.

Tier 8: aggregator press. Aggregator outlets that re-publish trade-press coverage or maker assertions without their own verification practice. Aggregator content is the lowest tier; we do not treat aggregator-only sourcing as verified.

The four confidence tiers

The confidence tier classifies the overall posture of a claim across whatever source-quality tier the evidence lives at.

Verified. The claim resolves at Tier 1, 2, or 3 source- quality depth, OR resolves at Tier 4 plus Tier 6 dual corroboration, OR at Tier 5 plus Tier 6 plus a structural independent corroboration. Multiple paths reach verified; the path is preserved in the source history.

Verified cap-flagged. The claim resolves at verified posture but a specific aspect of the claim is honest-absence. Stryker Q3 nine-month revenue is verified at SEC 10-Q depth; the full-year revenue is honest-absence pending the 10-K filing. The claim carries verified posture plus an explicit cap-flag on the absent aspect.

Claimed. The claim resolves at Tier 4 or Tier 5 source- quality only, without independent corroboration. The maker says the thing; no counterparty, regulator, or independent source confirms. The claim is surfaced as claim with the source attribution; it does not surface as verified.

Unverified. The claim resolves only at Tier 7 or Tier 8 source-quality depth, or no primary source has surfaced. The claim is surfaced as claim with the source attribution and a cap-flag at evidence-absent.

Posture markers per claim

A registry record's posture is asserted through a small set of markers that live alongside the substantive field.

verified_secondary. The claim resolves at verified posture via Tier 6 independent verification (a tier-1 news outlet confirming the underlying primary source). The marker carries the citation; the reader can pivot to the underlying primary source from the marker.

verified_ir. The claim resolves at verified posture via Tier 1 SEC filings or Tier 4 IR-depth disclosure. The marker carries the filing-form identifier and the filing date.

reported_not_verified. The claim resolves at Tier 5 maker- disclosed source-quality only. No independent corroboration. The marker carries the maker-source citation and surfaces the cap-flag at evidence-absent on independent confirmation.

The marker is per-claim. A deployment record carries verified_ir on the partnership announcement field and reported_not_verified on the per-unit throughput field if the partnership is SEC-disclosed but the throughput is maker- only. Both markers coexist on the same record.

Basis markers per posture

A basis marker carries the reasoning for the posture assignment. The basis says: this is verified because the following primary source resolves the claim; or this is claimed because no primary source has surfaced; or this is honest-absence because the maker has explicitly not disclosed.

revenue_basis on a manufacturer_financial_state record carries the source-form identifier (sec_10k, sec_10q, sec_8k, private_reported). Intuitive Surgical's FY2024 revenue carries revenue_basis sec_10k. Stryker's nine-month 2024 revenue carries revenue_basis sec_10q. Boston Dynamics carries revenue_basis null at honest-absence because no standalone revenue is disclosed; the Hyundai parent-backing is recorded separately.

role_basis on a people record carries the source classification for the role (sec_filed, corp_registry, court_record, operator_disclosed, linkedin_corroborating). A role with role_basis sec_filed plus corp_registry resolves at verified posture with dual primary-source corroboration. A role with role_basis operator_disclosed only resolves at verified posture only if the operator's own surface persists the claim and the claim does not contradict any other primary source.

incident_basis on an incident record carries the source-form identifier (fda_recall, nhtsa_recall, court_filing, regulator_inquiry). Incidents at incident_basis fda_recall plus tier_1_news_corroboration resolve at verified posture.

cash_runway_basis on a manufacturer_financial_state record carries the source classification for the runway estimate; null when cash_runway_months is also null per the validator- aware discipline. Honest-absence at both columns is the verified state when no primary-source disclosure exists.

Worked examples

The same fact resolves at different posture levels on different surfaces. The Apptronik Apollo enterprise pilot at Mercedes-Benz resolves at verified_secondary posture on the partnership-announcement field (multiple tier-1 outlets confirm); it resolves at reported_not_verified posture on the per-deployment throughput field (no customer-of-record throughput disclosure; the partnership is verified but the throughput surface stays cap-flagged).

Tesla Optimus pricing resolves at three distinct posture levels. The $20,000 to $30,000 at-scale projection resolves at Tier 5 maker-disclosed source-quality, claimed confidence tier. The pricing surfaces on the registry with the marker reported_not_verified and a basis pointing at Tesla's verbal framing. The same fact does not resolve at verified posture because no primary source confirms the pricing as transactable; it does not resolve at unverified posture because the maker source is verifiable. Claimed is the honest classification.

Stryker Q3 2024 revenue of $16,159M resolves at verified_ir posture with revenue_basis sec_10q. The full-year 2024 revenue carries cap-flag at evidence-absent because the 10-K had not yet been filed at the authoring date. The verified posture stays; the cap-flag is explicit on the absent aspect.

Boston Dynamics counterparty-risk classification resolves at verified posture via Hyundai's ~80% parent-backing per Hyundai's SEC-equivalent disclosures. The standalone Boston Dynamics revenue resolves at honest-absence (private subsidiary, no standalone disclosure). Two facts on the same record at different posture levels, both surfaced explicitly.

Intuitive Surgical's 11,395 da Vinci installed-base count resolves at verified_ir posture via SEC 10-Q disclosure. The per-procedure clearance scope per device class resolves at verified posture via FDA primary-government-record source. Two facts on the same Model record, both at Tier 1 primary- government-record source-quality but different specific source forms.

Cap-flag application

A claim with no source meeting Tier 1 through Tier 6 source-quality is cap-flagged at evidence-absent. The claim is recorded if it is editorially meaningful; the posture is unverified with the cap-flag explicit.

A claim with Tier 4 IR-depth disclosure but no Tier 6 corroboration is cap-flagged at single-source. The verified_ir posture is preserved; the cap-flag names the absent corroboration.

A claim that resolves at multiple source-quality tiers but where the tiers disagree is cap-flagged at evidence- contradicted. The verified posture does not apply; the claim surfaces with both sources visible and the cap-flag explicit.

The cap-flag-as-trust-signal discipline operates uniformly across deployment status, capability claims, pricing claims, regulatory coverage, and incident attribution. See /methodology/cap-flag-as-trust-signal for the canonical reference.

The framework applies to us

If we mis-assign verification posture on a claim, we correct it. A claim recorded at verified posture that resolves on closer examination at Tier 4 IR-depth without Tier 6 corroboration moves to verified_ir cap-flagged at single- source. A claim recorded at reported_not_verified that gets subsequent Tier 6 corroboration moves to verified_secondary.

The correction is logged at /corrections. The original posture is preserved in the source history. A reader can see what posture we held when, and what changed.

Where to go next

The methodology canon: /methodology/what-verified-means for the core verified-vs-claimed framework, /methodology/cap-flag-as-trust-signal for the standalone cap-flag pillar, /explainers/source-quality-rubric for the source-quality taxonomy at editorial canonical depth, /explainers/verified-vs-claimed-within-entity for the within-entity granularity discipline.

The applied work: registry.deploy.report for the canonical entity records with verification posture markers + basis markers per claim.