6. Verification Object Schema and Qualification Semantics

6.1 Verification Object Schema

A Verification Object is the atomic output of Baseline. It is a content-addressed, immutable record of a single claim evaluation.

{
  "voId":           "string           // Keccak-256 hash of canonically encoded contents",
  "lineStatus":     "LINE_STATUS      // 'ABOVE' | 'BELOW'",
  "claim":          "Claim            // Full claim definition (Section 2.1)",
  "engine": {
    "version":      "string           // Engine version used"
  },
  "predicate": {
    "predicateId":  "string           // Predicate identifier (Section 2.4)",
    "version":      "string           // Predicate version used (from engine's Predicate Manifest, Section 2.5)"
  },
  "methods": [{
    "name":         "string           // Method identifier",
    "version":      "string           // Method version"
  }],
  "result": {
    "qualification":  "QualificationType",
    "confidence":     "number         // 0-100 (100 = fully deterministic)",
    "bounds": {
      "low":          "number | null",
      "high":         "number | null"
    },
    "verdictText":    "string         // Human-readable summary",
    "reasons":        "string[]       // Supporting/undermining factors"
  },
  "evidence":       "EvidenceReference[]  // References to evidence units used",
  "evidenceGraph": {
    "graphId":      "string",
    "summary":      "string           // Textual summary of graph structure",
    "previewUrl":   "string | null    // URL to visual graph rendering"
  },
  "attestations": {
    "count":        "number",
    "validatorIds": "string[]",
    "signatures": [{
      "validatorId":  "string",
      "sig":          "string         // Ed25519 or secp256k1 signature"
    }]
  },
  "baselineChain": {
    "anchoringStatus":  "ANCHORING_STATUS // 'PRE_ANCHORING' | 'PENDING' | 'ANCHORED' | 'FAILED'",
    "txHash":           "string | null    // FirmaChain transaction hash",
    "anchoredAt":       "string | null    // ISO 8601 timestamp",
    "anchoringPhase":   "string | null    // Migration phase: 'SHADOW' | 'TESTNET' | 'MAINNET' | null (full production)"
  },
  "timestamps": {
    "createdAt":    "string           // ISO 8601",
    "updatedAt":    "string           // ISO 8601"
  }
}

Content-Addressing:

voId = Keccak-256(BCE_ENCODE(VerificationObject excluding voId, attestations, baselineChain, timestamps))

This ensures:

• Any modification to the claim, result, evidence, or method produces a detectably different voId
• Attestations and anchoring metadata are excluded from the hash (they are added after initial creation)
• Timestamps are excluded to allow identical evaluations to produce identical voIds

Anchoring Status Semantics:

The anchoringPhase field provides additional context during the migration period (Section 8.8):

Trust transparency rule: API responses MUST include anchoringStatus and anchoringPhase for every VO. Consumers SHOULD treat PRE_ANCHORING VOs as operator-asserted (not independently verifiable) and display this distinction in their UIs.

6.2 Qualification State Determination

Qualification is determined mechanically based on evaluation results.

VERIFIED

Entry Conditions:

• evaluationClass = DETERMINISTIC for ALL predicates evaluated
• ALL relevant predicates were evaluated (none skipped)
• ALL results are conclusive and consistent
• No inference was required
• ALL evidence units are ONCHAIN class (independently re-fetchable). VOs that depend on any OFFCHAIN_DECLARED evidence cannot reach VERIFIED qualification, because the observational correctness of off-chain evidence cannot be independently confirmed (Section 3.3).

Confidence: 100

INFERRED

Entry Conditions:

• At least one inference method was used, OR at least one evidence unit is OFFCHAIN_DECLARED class
• Primary hypothesis confidence exceeds the method's reporting threshold (default: 50)
• Alternative hypotheses are attached

Confidence: As reported by inference method. When OFFCHAIN_DECLARED evidence is present, a confidence ceiling applies (see below).

Off-chain evidence confidence ceiling:

When a VO depends on OFFCHAIN_DECLARED evidence, the reported confidence is capped:

effective_confidence = min(method_confidence, offchain_ceiling)

This ceiling reflects that off-chain evidence is replay-verified (computational correctness) but not observation-verified (the snapshot was actually correct at collection time). The ceiling is applied BEFORE the confidence is stored in the VO, so it affects voId computation.

OBSERVED

Entry Conditions:

• Evidence graph was constructed successfully
• Structural properties were described
• The specific predicate could not be meaningfully evaluated within this context
• Structure was found and recorded, but the claim itself remains unevaluated

Confidence: N/A (structure was observed but claim not evaluated)

INCONCLUSIVE

Entry Conditions:

• Evidence supports multiple incompatible interpretations
• No primary hypothesis has confidence above the reporting threshold
• Scope was NOT exhausted (sufficient evidence was available, it was just ambiguous)

Confidence: Below threshold

UNQUALIFIED

Entry Conditions:

• Evidence is insufficient to construct a meaningful evaluation
• Evidence retrieval failed, subject was not found, or scope was entirely empty
• This is distinct from a negative result

Confidence: N/A

CONSTRUCTION_FAILED

Entry Conditions:

• Graph construction was aborted due to an operational timeout (Section 4.4) or infrastructure failure
• No deterministic graph was produced — the evidence graph does not exist for this attempt
• This is distinct from UNQUALIFIED: UNQUALIFIED means evidence was retrieved but insufficient; CONSTRUCTION_FAILED means the graph was never completed

Confidence: N/A (no evaluation was performed)

Replay: CONSTRUCTION_FAILED VOs are not submitted for attestation or replay. They are recorded for operational monitoring only and MUST NOT influence any deterministic output.

CONTESTED (extension)

Entry Conditions:

• Multiple qualified evaluations exist for the same claim under the same context
• These evaluations produce different qualifications or significantly different results
• Used by products to surface dispute status

Confidence: Varies per evaluation

6.3 Memecoin Evaluation Output (Product View)

The Memecoin Intel Portal renders multiple Verification Objects as a composite evaluation:

{
  "coinId":           "string",
  "observedAt":       "string         // ISO 8601",
  "overallScore":     "number         // 0-100",
  "categoryScores": {
    "onchain":        "number         // 0-100",
    "social":         "number         // 0-100",
    "project":        "number         // 0-100"
  },
  "scoringModel": {
    "type":           "string         // 'SIMPLE_AVG' | 'WEIGHTED'",
    "version":        "string"
  },
  "supportingVOs":    "string[]       // voIds that back this evaluation",
  "detail": [{
    "metricKey":      "string",
    "status":         "METRIC_STATUS  // 'PASS' | 'WARN' | 'FAIL' | 'NA'",
    "score":          "number         // 0-100",
    "linkedVOs":      "string[]       // voIds"
  }]
}

Scoring Categories (12 categories, 100 max points):

Score Tiers:

Critical FAIL Overrides (instant red flags regardless of score):

• Honeypot confirmed
• Hidden mint function active
• Single non-CEX holder > 20%
• LP unlocked with concentrated holder
• Known deployer rug history