MATH40004 OVERANCHORING DIAGNOSIS REVISED 2

English passage 1 (6% down the source page) AUDITED_REVISED_2 revision statement

English passage 2 (11% down the source page) The body of this report comes from the previous revised version. Codex's independent review found that the previous main ZIP omitted the 2020 original question booklet and official solutions/marking materials, affecting the 2020 question-level evidence.

English passage 3 (12% down the source page) These question-level claims could not be independently verified within the package.

English passage 4 (15% down the source page) This version includes the original file “MATH40004MATH40011_Calculus and Applications_Q&S;_May_2020(3).pdf” unchanged under 06_SOURCE

English passage 5 (16% down the source page) _EVIDENCE/2020, registered in the original-evidence index and SHA256SUMS. This PDF has 22 pages. Its SHA-256 is:

English passage 6 (19% down the source page) 143a8d4fdaaf7f49f781e464c0794cc7a29cd1f96d25440bc26dafa79164520b Question-level conclusions for 2020–2026 must therefore be read with the original PDFs. Reports cannot replace original materials. Apart from evidence-chain completion and workbook

English passage 7 (22% down the source page) mechanical-format repairs, the report's conclusions have not been recalculated or altered. Version date: 2026-08-06.

English passage 1 (5% down the source page) diagnostic report on excessive anchoring

English passage 1 (5% down the source page) 1. review scope and evidence boundaries

English passage 2 (9% down the source page) The review actually checked the seven years of formal papers from 2020–2026, official solutions/marking annotations/examiner comments in the same files, the old analysis workbook, the old six practice papers and two later additions, all old solutions and the old coverage report. In the old practice compilation, Papers 1, 4, 6 and the later Paper 8 repeatedly use the 2026 method combination: Taylor/improper integrals – physical models/Fourier series – linear systems/integrating factors – bifurcations/Fourier methods for ODEs/implicit functions.

English passage 3 (14% down the source page) Materials not found: this session and the accessible library contained no lecture notes, slides, problem sheets, tutorial/assignment/formula sheets or Tony's answers/marking records confirmably belonging to MATH40004. The strongest accessible evidence for the course-method review therefore consists of official-solution labels such as seen/similar/from lectures, marking schemes and examiner comments. Tony's feedback from other courses was not mixed into this course.

English passage 4 (18% down the source page) [The source completes the preceding sentence here.]

English passage 5 (22% down the source page) 2. the substance of excessive anchoring in the old version

English passage 6 (26% down the source page) • The old version treated the four 2026 question slots as a fixed framework for six papers, instead of using 2026 only to calibrate duration, total marks and overall difficulty.

English passage 7 (28% down the source page) • Several papers changed functions, parameters or stories while keeping the same first key step and complete method sequence. This caused repeated practice rather than transfer of historical abilities.

English passage 8 (30% down the source page) • The old coverage matrix ticked off 50 knowledge-point names. It could not distinguish direct question-type entry points, transfer of methods to unfamiliar formulations, and merely sharing a term.

English passage 9 (32% down the source page) • The old eight-paper quality report explicitly described Paper 8 as “mirrors 2026 Q2 function”. This is precisely the anchoring logic now being abandoned.

English passage 10 (36% down the source page) 3. quantitative review

English passage 11 (40% down the source page) Recalculated across 210 formal-question ability units, the old six papers have A+B coverage of 185/210 (88.1

English passage 12 (42% down the source page) For the six revised papers, A+B coverage is 207/210 (98.6

English passage 13 (46% down the source page) 4. core correction

English passage 14 (49% down the source page) 1. Only Paper 1 retains a structural-reinforcement role close to 2026.

English passage 15 (52% down the source page) 2. Rebuild Paper 2 around high-mark 2025 entry points, Paper 3 around 2023–2024, and Paper 4 around 2020–2022.

English passage 16 (54% down the source page) 3. Paper 5 fills only gaps in the A/B matrix. Paper 6 mixes ability families from at least three years and changes information order.

English passage 17 (56% down the source page) 4. Every solution adds a recognition signal, first key step, course method, marking points, common errors and self-check.

English passage 18 (59% down the source page) 5. Keep old files as historical records, but include only revised files in the final ZIP and Codex import list.

English passage 19 (63% down the source page) 5. notes correcting earlier analytical conclusions

English passage 20 (66% down the source page) • “50/50 knowledge-point coverage” no longer means effective coverage; effective coverage counts only A+B.

English passage 21 (68% down the source page) • Withdraw the claim that eight papers are more comprehensive. The final training set has six papers with clear roles.

English passage 22 (71% down the source page) • The 2020 official centroid solution page has inconsistent numbers. For 0 ≤ y ≤ 1 − x² and 0 ≤ x ≤ 1, the correct x̄ is 3/8, not 3/4. The revised version uses 3/8, verified by direct moment calculation.

English passage 23 (74% down the source page) • Without course lecture notes, withdraw every claim of 100% complete-syllabus coverage; state only coverage of historical formal-examination abilities.