How does DEPLOY track incident outcome_class and deployment exposure_hours at actuarial depth?

Why this matters editorially: Dim 1 actuarial substrate landed

Per Agent A's Phase 3 robot insurance scoping deliverable + Agent B's Dim 1 schema migration + Agent A's incident outcome_class + deployment exposure_hours backfill + Rover's 5-commit Dim 1 surface work, the deployment-incident-recall actuarial substrate landed at primary-source-anchored verification depth across all four axes simultaneously. The substrate is fully live and queryable:

• Callable: v2.get_incident_outcomes MCP tool (commit 0a29e08); V2 catalog 46 → 47.
• Structured: deploy:Outcomes JSON-LD on /incidents/[slug] (b6cb341) + deploy:Exposure JSON-LD on /deployments/[slug] (7327c5a).
• Human-facing: visible Outcomes section on /incidents/[slug] (75801c4).
• REST aggregated: /v1/robots/[id]/outcomes aggregated endpoint (da01320).

The substrate composition matters editorially because it operationalizes the four-dimension actuarial framework documented in DEPLOY's robot insurance Project B methodology pillar at primary-source-anchored verification depth. The verified-vs-claimed framework operates uniformly across actuarial records: the numerator + denominator + scalar selectivity composition operates at the verification depth where each fact actually resolves.

The framework operates at canonical-methodology depth, not at marketing-aggregation depth. The product differentiator: DEPLOY's framework treats actuarial substrate the same way it treats every other verified-vs-claimed claim. Honest 128-null exposure_hours at full-population scale beats fabricated rate at partial-population coverage.

---

The substrate composition at primary-source-anchored verification depth

Per Agent A's Dim 1 backfill substrate, the actuarial framework operates at three structurally-distinct axes simultaneously.

Axis 1: incident numerator at outcome_class typed-record depth. 67 outcome_class rows across 61 incidents (multi-class where evidenced; the framework reads multi-class as substantive structural state, not as record duplication). The distribution: regulatory_action:34 + property_damage:18 + bodily_injury:7 + no_outcome:4 + financial_loss:2 + fatality:1 + near_miss:1. The distribution operates at primary-source-anchored verification depth; regulatory_action dominates per Agent A primary-source-anchored verification of FDA recall + NHTSA SGO + CPSC notice + ITC determination + DoD program-disclosure recordings across the 61-incident corpus.

Axis 2: deployment denominator at exposure_hours scalar-record depth. The canonical 128-nulls. Agent A documented under-population at primary-source-anchored verification depth rather than fabrication at marketing-aggregation depth. The 128-nulls operate at honest-absence cap-flag at full-population scale; the framework does not infer exposure denominators from speculation or secondary-source narrative. Per DEPLOY's robot insurance Project B methodology pillar, exposure denominator absence at most deployments operates as substantive structural state, not as substrate-completeness gap.

*Axis 3: scalar selectivity at _basis verified-vs-claimed discipline. The framework operates per-basis verification posture at primary-source-anchored depth. 1 USD figure across the cohort: MQ-9 Reaper $32M reported_press gt_10m (per primary-source press disclosure; reported_press tier per the 9-tier source-quality rubric; gt_10m magnitude classifier). 3 fatality_count rows at primary-source-anchored typed-record depth. 5 bodily_injury_severity rows at primary-source-anchored typed-record depth. 4 loss_cost_class rows at primary-source-anchored typed-record depth. Most NULL at honest-absence cap-flag.

The substrate composition operationalizes the verified-vs-claimed framework at three-axis granularity: numerator + denominator + scalar each operate at the appropriate verification depth where each fact actually resolves; cross-axis composition reads at the lowest-verification-depth axis per DEPLOY's robot insurance Project B methodology pillar.

---

The canonical 128-nulls exposure_hours worked example

The 128-nulls exposure_hours at full-population scale operates as the canonical worked example for the discipline 'honest absence at full-population scale beats partial fabrication.' Per Agent A's documented authoring decision, the deployment denominator carries explicit honest-absence cap-flag at primary-source-anchored verification depth rather than fabrication at marketing-aggregation depth.

The discipline operates at three layers:

Layer 1: under-population at primary-source-anchored verification depth. Where deployment-hours / units-in-service / miles operate at publicly-disclosed primary-source verification depth (Waymo published cumulative miles; BMW Spartanburg 1,250 operating hours per Figure 02 at BMW Spartanburg deployment; GXO Flowery Branch 100,000-tote scaled-throughput per Agility Digit), exposure_hours operates at populated typed-record depth. Where exposure operates at non-disclosed depth (most cohort deployments), exposure_hours operates at NULL at honest-absence cap-flag.

Layer 2: fabrication-rejection at marketing-aggregation depth. The framework explicitly rejects fabrication at marketing-aggregation depth. Aggregator coverage frequently reports estimated deployment-hours / fleet-wide miles from speculation or secondary-source narrative; the framework does not inherit these as primary-source-anchored verification.

Layer 3: full-population scale at editorial-credibility depth. The 128-nulls operate at full-population scale rather than at partial-population coverage. Partial-population fabrication would produce a 'partial substrate-completeness' framing; the framework reads full-population honest-absence as the editorial truth, not as substrate-completeness gap.

Per the robot insurance methodology pillar's canonical discipline, honest 'insurability unknown for this region / no exposure data' beats fabricated rate. The 128-nulls exposure_hours substrate operationalizes this discipline at the structured-record-substrate layer; the framework reads honest-absence at the structured-data depth where each deployment's exposure operates.

---

Scalar selectivity at 1-of-61 USD verified ratio

The scalar selectivity tightness operates as substantive editorial signal at canonical-actuarial-language depth. Per Agent A's Dim 1 substrate, only 1 USD figure operates at primary-source-anchored verification depth across the 61-incident corpus: MQ-9 Reaper $32M reported_press gt_10m.

The single USD figure operates at three substrate-axis verification posture simultaneously:

• Basis: reported_press tier per the 9-tier source-quality rubric. Press-release-disclosed at announcement-tier depth; not SEC-verifiable; framework reads basis at appropriate tier verification depth.
• Magnitude: gt_10m classifier at primary-source-anchored typed-record depth. The classifier operates at order-of-magnitude precision rather than at specific-dollar-precision; framework reads magnitude classifier as substantive editorial state where exact-dollar disclosure operates at honest-absence.
• Entity: MQ-9 Reaper at General Atomics primary-source-anchored entity verification. Per autonomy-boundary classification, MQ-9 Reaper operates at the legacy-prime remotely-piloted tier; the USD figure operates at the entity verification posture where the incident actually resolves.

The 1-of-61 selectivity ratio (1.6% USD-disclosed across cohort) operates at extreme primary-source-anchored selectivity. The framework reads this as canonical worked example of *_basis discipline: per-basis verification posture at primary-source-anchored depth across the full-population cohort. Most USD figures operate at NULL at honest-absence cap-flag; the framework does not infer financial loss-cost from speculation or secondary-source narrative.

Per the broader actuarial framework, similar selectivity ratios operate at 3-of-61 (4.9%) fatality_count + 5-of-61 (8.2%) bodily_injury_severity + 4-of-61 (6.6%) loss_cost_class. The substrate selectivity tightness operates as editorial signal at the per-basis verification posture depth; the framework reads selectivity ratios as substantive structural state, not as substrate-completeness gap.

---

Outcome distribution skew as substantive structural state

Per Agent A's Dim 1 backfill substrate, the outcome_class distribution operates at substantive structural state at three editorial layers.

Layer 1: regulatory_action:34 dominance. Regulatory_action operates at 34/67 outcome_class rows (50.7% of distribution). The dominance operates at substantive structural state per Agent A primary-source-anchored verification of FDA recall + NHTSA SGO + CPSC notice + ITC determination + DoD program-disclosure recordings across the 61-incident corpus. The framework reads regulatory_action dominance as substantive editorial state: regulatory-action substrate is the most-densely-verified outcome class at primary-government-record sub-tier verification per the 9-tier source-quality rubric.

Layer 2: property_damage:18 + bodily_injury:7 secondary distribution. Property damage and bodily injury operate at secondary-distribution depth (18 + 7 = 25/67 = 37.3% combined). The secondary distribution operates at substantive structural state per primary-source-anchored verification at incident-record-typed-disclosure depth.

Layer 3: no_outcome:4 + fatality:1 + near_miss:1 tail distribution. The tail distribution operates at canonical honest-absence worked example. no_outcome:4 operates as substantive editorial state, NOT as substrate-completeness gap. Per Agent A's authoring discipline, incidents with no documented outcome at primary-source verification depth carry no_outcome typed-record at primary-source-anchored verification of non-outcome state. The framework reads no_outcome as load-bearing editorial state: incidents that did not produce documented outcome at primary-source verification depth operate at no_outcome typed-record depth rather than at outcome-inferred-by-speculation depth.

fatality:1 operates at primary-source-anchored verification at extreme severity tier. The single fatality_count operates at the cohort's most severe verified incident at primary-source FDA-recall-database verification per the 2026 da Vinci SureForm gray-reload Class I recall framework-in-action narrative. The framework reads fatality count selectivity as substantive editorial signal: the cohort's verified-base substrate carries single primary-source-anchored fatality at typed-record depth across the 61-incident corpus.

near_miss:1 operates at substrate-pattern depth. Near-miss outcome class captures incidents that did not produce documented harm but operated at structural risk-disclosure depth per primary-source-anchored verification. The framework reads near-miss as substantive editorial state subject to ongoing substrate-completeness verification.

---

Cap-flag-as-trust-signal recursive at three substantive layers

Per DEPLOY's verified-vs-claimed framework, the Dim 1 actuarial substrate operates cap-flag-as-trust-signal recursively at three substantive layers simultaneously.

Layer 1: 128-null exposure_hours denominator. Full-population honest-absence at structured-record-substrate depth. The framework reads full-population honest-absence as the editorial truth: under-population at primary-source-anchored verification depth beats fabrication at marketing-aggregation depth.

Layer 2: scalar selectivity at 1-of-61 USD verified ratio. Extreme primary-source-anchored selectivity. The framework reads selectivity tightness as substantive structural state: per-basis verification posture at primary-source-anchored depth across full-population cohort.

Layer 3: outcome distribution skew at no_outcome:4 documented depth. Substantive editorial state. The framework reads no_outcome as load-bearing editorial state at typed-record depth, not as substrate-completeness gap.

The three-layer recursive cap-flag application operates uniformly per the Moon Maestro K240598 reconciliation pattern; when primary-source verification surfaces additional exposure_hours data or USD figures or no_outcome documentation, the framework operates inline editorial-transparency footer pattern + reconciliation discipline. Until primary-source verification surfaces, the framework holds the cap-flag at full-population honest-absence depth.

---

Honesty-as-strength vs coverage-as-strength editorial framing

Per DEPLOY's restraint-IS-the-product discipline, the Dim 1 actuarial substrate operates at editorial-credibility depth rather than at coverage-completeness depth. The framework's product differentiator: institutional discourse about actuarial substrate frames coverage-completeness as the trust signal; the framework distinguishes honesty-as-strength from coverage-as-strength.

The distinction operates at three structural axes:

Coverage-as-strength framing: aggregator coverage reports estimated deployment-hours / fleet-wide miles from speculation + secondary-source narrative; the framing treats coverage-completeness as the editorial trust signal; partial-population fabrication operates at apparent substrate-completeness depth.

Honesty-as-strength framing: DEPLOY's framework reports primary-source-anchored verification depth where exposure operates at publicly-disclosed verification depth; full-population honest-absence operates at structured-data depth where each deployment's exposure operates; framework rejects fabrication at marketing-aggregation depth.

The structural distinction matters editorially at three layers:

Operational reality. Institutional partners considering robot deployment underwriting evaluate per-substrate-axis verification posture at primary-source-anchored verification depth, not at coverage-completeness depth. The honesty-as-strength framing operationalizes the distinction.

Editorial credibility. Honest 128-null exposure_hours at full-population scale beats fabricated coverage at partial-population framing. The framework discriminates against coverage-as-strength framings and rewards honesty-as-strength posture.

Cross-property bidirectional graph. The Dim 1 actuarial substrate cross-references Acquisition records + Partnership records + Entity records + people graph PersonCompany edges + within-entity verification scope simultaneously. The cross-property bidirectional discipline compounds AEO citation graph density; the framework reads actuarial substrate richness as a load-bearing trust signal at the cross-property layer.

---

Cross-property bidirectional compounding at multi-pillar methodology depth

Per DEPLOY's methodology cluster as AEO citation graph discipline, the Dim 1 framework-in-action narrative compounds at substrate layer with multiple Project B methodology pillar essays + framework-in-action narratives simultaneously:

• Robot insurance Project B methodology pillar canonical worked example: per How DEPLOY thinks about robot insurance four-dimension framework, the Dim 1 actuarial substrate operationalizes the canonical 'honest insurability unknown beats fabricated rate' discipline at structured-data depth.
• Cross-cluster talent-flow framework-in-action: per How DEPLOY tracks cross-cluster talent-flow as diaspora graph, the Class 1 post-wind-down Cruise diaspora cross-references the GM full_acquisition Acquisition record + Cruise wound-down deployment-state per what happened to Cruise; the Dim 1 actuarial substrate operates at incident-record granularity across the Cruise-wound-down deployment context.
• Acquisition history Project B methodology pillar: per How DEPLOY tracks acquisition history state, the contingent valuation_basis state (ZB × Monogram CANONICAL) operates at financial-state verification posture distinct from Dim 1 actuarial substrate at incident-record verification posture; cross-property bidirectional discipline operates at Acquisition record ↔ Incident record ↔ Deployment record granularity.
• Partnership lifecycle Project B methodology pillar: per How DEPLOY tracks partnership lifecycle state, the four-state lifecycle framework operates at partnership-record verification posture; cross-property bidirectional discipline operates at Partnership record ↔ Deployment record ↔ Incident record granularity.
• Surgical incident framework-in-action: per How DEPLOY tracks the 2026 da Vinci SureForm gray-reload Class I recall, the cohort's most severe verified incident operates at canonical Dim 1 actuarial signal worked example; the SureForm gray-reload incident contributes the single fatality_count to the 67-row outcome_class distribution.
• Verified-vs-claimed at within-entity granularity: per verified-vs-claimed at within-entity granularity, per-claim within-entity verification posture operates at the same depth as per-basis Dim 1 actuarial substrate verification posture; framework discipline operates uniformly across within-entity feature-level + within-incident outcome-class + within-deployment exposure_hours granularity.
• 9-tier source-quality rubric: per the 9-tier source-quality rubric, per-claim source-quality classification operates at the same depth as per-basis Dim 1 actuarial substrate classification (reported_press tier MQ-9 Reaper USD figure; primary-government-record sub-tier regulatory_action substrate).

The cross-property bidirectional discipline compounds at the AEO citation graph layer. Institutional partners + AI assistants + downstream consumers navigating Dim 1 actuarial queries encounter the methodology pillar canonical references + the framework-in-action narrative cohort + the canonical worked examples at unified verification posture.

---

Why this catch matters editorially

Per DEPLOY's restraint-IS-the-product discipline, the Dim 1 framework-in-action narrative operates at editorial-credibility depth rather than at substrate-completeness coverage depth. The framework reads the Dim 1 actuarial substrate at three substantive cap-flag layers simultaneously (128-null denominator + 1-of-61 USD selectivity + no_outcome:4 documented); each layer composes at primary-source-anchored verification depth.

Institutional partners audit DEPLOY's framework discipline at the operational-practice layer, not just the stated-methodology layer. The verification-posture statement at /verified-vs-claimed describes the framework abstractly. The robot insurance methodology pillar describes the Phase 3 four-dimension framework abstractly. This piece operates at narrative-canonical depth: how the substrate landed (Agent A backfill + Rover 5-commit surface work across callable + structured + human-facing + REST aggregated axes); what the discipline was (under-population at primary-source-anchored verification depth beats fabrication at marketing-aggregation depth); what the editorial outcome was (canonical 128-nulls exposure_hours worked example + scalar selectivity tightness + outcome distribution skew as substantive structural state); and what the broader pattern is (honesty-as-strength vs coverage-as-strength editorial framing).

The catch demonstrates the discipline operationally at editorial-anchor depth. Not because the catch is exceptional. Because the catch is what the discipline does. 128 nulls at full-population scale; 1 USD verified figure at extreme selectivity; 4 no_outcome typed records at documented depth. The verification protocol doesn't relax at substrate-completeness depth.

For the canonical robot insurance methodology pillar where Phase 3 Dim 1 actuarial framework operates, see How DEPLOY thinks about robot insurance. For the cohort's most severe verified incident contributing the single fatality_count to the distribution, see How DEPLOY tracks the 2026 da Vinci SureForm gray-reload Class I recall. For the cross-cluster talent-flow framework-in-action where the Cruise wound-down deployment context cross-references the Dim 1 actuarial substrate at incident-record granularity, see How DEPLOY tracks cross-cluster talent-flow as diaspora graph. For the methodology editorial canonical reference, see how DEPLOY verifies. For the verified-vs-claimed methodology pillar operating uniformly across all claim depths, see verified-vs-claimed methodology pillar.