Frozen Examination Identity & Governance Boundaries
Printable Books & Raw Evidence Downloads
Download the printable 59-page Examiner Marking Book, Candidate Answer Book, or complete unmodified JSON evidence:
Evaluation Tier Breakdown (58 Questions | 230 Marks)
| Evaluation Tier | Items | Max Marks | Offline Status | Model % Scope |
|---|---|---|---|---|
| 1. Historical development regression (T01–T28) | 28 | 84 | 10 Verified / 18 Abstained | Executable (42) |
| 2. Human-authored challenge cases (H01–H12; NOT independently blind) | 12 | 60 | 6 Verified / 6 Abstained | Executable (42) |
| 3. Multi-relation arbitration (A02–A03) | 2 | 10 | 0 Verified / 2 Abstained | Executable (42) |
| 4. Unsupported analytical questions (F01–F08, A01, A04–A06) | 12 | 52 | 12 NOT_RUN (must be 0 marks) | Model Capability (54) |
| 5. Human scientific audits (V1–V4) | 4 | 24 | 4 Human-Only Audits | Excluded from Model % |
Provisional AI Second-Marker Notes (Non-Authoritative Reference)
Rule reminder: On executed items, intermediate derivations are
NOT_RECORDED, so Dimension 4 (Mathematical derivation) is locked to 0. Enter marks across the other dimensions; the total updates automatically.
Part V — Independent Human Scientific Audits (V1–V4)
These 4 items are independent physicist audits. They are recorded separately and are never counted as model-earned scientific marks or included in model-performance percentages.
Final Examiner Sign-Off & One-Click Submission
No command-line tools or manual JSON editing needed. Click Submit Completed Review below: the server will validate your sheet using the authoritative scorer and save a tamper-evident SHA-256 submission receipt.