The ERQ contract

# IB DP Physics 2025 — Course Anatomy _Extracted from `Physics_Guide_2025_EN_Nov2024.pdf` + Nikhil's `IB_DP_Extended_Response_Questions.pdf`._ _This document is the contract any problem-generator / study-guide pipeline must respect._ ## 1. Course architecture | | SL | HL | |---|---|---| | Total teaching hours | 150 | 240 | | Syllabus content | 110 | 180 | | Experimental programme | 40 | 60 | | Duration | 2 years | 2 years | | Age | 16–19 | 16–19 | ## 2. Syllabus content — 5 themes, 22 sub-topics Bullet legend: `•` = all students · `••` = all + HL extras · `•••` = HL only. ### A. Space, time and motion — 27h SL / 42h HL | Sub-topic | SL hrs | HL extras | Level | Key equations | |---|---|---|---|---| | A.1 Kinematics | 9 | — | • | suvat (s=ut+½at², v²=u²+2as), projectile motion | | A.2 Forces and momentum | 10 | — | • | F=ma, J=FΔt, F=Δp/Δt, a=v²/r, v=ωr | | A.3 Work, energy and power | 8 | — | • | W=Fs cosθ, E_k=½mv², E_p=mgh, P=ΔW/Δt, η | | A.4 Rigid body mechanics | — | 7 | ••• | τ=Fr sinθ, I=Σmr², τ=Iα, L=Iω, E_k=½Iω² | | A.5 Galilean and special relativity | — | (HL only) | ••• | (relativistic kinematics) | ### B. The particulate nature of matter — 24h SL / 32h HL B.1 Thermal energy transfers, B.2 Greenhouse effect, B.3 Gas laws, B.4 Thermodynamics (HL only), B.5 Current and circuits. ### C. Wave behaviour — 17h SL / 29h HL C.1 SHM, C.2 Wave model, C.3 Wave phenomena, C.4 Standing waves and resonance, C.5 Doppler effect. ### D. Fields — 19h SL / 38h HL D.1 Gravitational fields, D.2 Electric and magnetic fields, D.3 Motion in EM fields, D.4 Induction (HL only — 6h). ### E. Nuclear and quantum physics — 23h SL / 39h HL E.1 Structure of the atom (6h SL + 3h HL extras), E.2 Quantum physics (HL only), E.3 Radioactive decay, E.4 Fission, E.5 Fusion and stars. Each sub-topic carries: **guiding questions** (frame the topic), **understandings** (split into SL+HL and HL-only), **guidance** (clarifications/limitations), **linking questions** (cross-topic + NOS). → Full syllabus tree mirror at `pdftotext` extract `/tmp/physics_guide.txt` lines 1325–3120 (we'll commit the tree as YAML once the structure stabilises). ## 3. Skills framework — 3 Tools + 3 Inquiry phases | Tool | Coverage | |---|---| | **Tool 1 — Experimental** | Safety/ethics, measuring (mass, time, length, volume, T, F, I, V, angle, sound/light intensity) | | **Tool 2 — Technology** | Sensors, databases, simulations, video analysis; spreadsheets, computer modelling | | **Tool 3 — Mathematics** | arithmetic/algebra, logs/exponents, rates of change, scientific notation, vectors (3 max, 2 components), units/SI, uncertainties, graphing/linearisation | Inquiry process: **1. Exploring & designing → 2. Collecting & processing → 3. Concluding & evaluating.** ## 4. Assessment objectives | AO | Description | Where assessed | Weight in Paper 1+2 | |---|---|---|---| | AO1 | Demonstrate knowledge (terminology + skills) | Paper 1, Paper 2, IA | (AO1+AO2 = 50%) | | AO2 | Understand and apply knowledge | Paper 1, Paper 2, IA | (AO1+AO2 = 50%) | | AO3 | Analyse, evaluate, synthesize | Paper 1, Paper 2, IA | 50% | | AO4 | Apply skills in insightful + ethical investigations | IA only | — | **The Paper-2 AO3 weighting (50%) is the lever the extended-response questions pull.** ## 5. Assessment outline ### SL — 3h external + 10h internal | Component | Time | Format | Marks | % | |---|---|---|---|---| | Paper 1A | (within 1h30m total Paper 1) | Multiple-choice | 25 | (Paper 1 = 36%) | | Paper 1B | (within 1h30m total Paper 1) | Data-based | 20 | | | Paper 2 | 1h30m | Short-answer + extended-response (SL material) | 50 | 44% | | Internal assessment | 10h | Scientific investigation | 24 | 20% | ### HL — 4h30m external + 10h internal | Component | Time | Format | Marks | % | |---|---|---|---|---| | Paper 1A | (within 2h total Paper 1) | Multiple-choice | 40 | (Paper 1 = 36%) | | Paper 1B | (within 2h total Paper 1) | Data-based | 20 | | | Paper 2 | 2h30m | Short-answer + extended-response (SL + additional HL material) | 90 | 44% | | Internal assessment | 10h | Scientific investigation | 24 | 20% | **Paper 2 is the ER battleground.** HL Paper 2: 90 marks / 150 min ≈ 36 marks/hour. An ER worth 10–16 marks ≈ 17–27 min — matches Nikhil's "~20 min" anchor. ## 6. Command-terms glossary (the generator's vocabulary) | AO | Command term | Definition (verbatim from IBO) | |---|---|---| | AO1 | **Draw** | Represent by means of a labelled, accurate diagram or graph, using a pencil. Ruler for straight lines. To scale. | | AO1 | **State** | Give a specific name, value or other brief answer without explanation or calculation. | | AO2 | **Annotate** | Add brief notes to a diagram or graph. | | AO2 | **Calculate** | Obtain a numerical answer showing the relevant stages in the working. | | AO2 | **Describe** | Give a detailed account. | | AO2 | **Estimate** | Obtain an approximate value. | | AO2 | **Identify** | Provide an answer from a number of possibilities. | | AO2 | **Outline** | Give a brief account or summary. | | AO3 | **Analyse** | Break down in order to bring out the essential elements or structure. | | AO3 | **Deduce** | Reach a conclusion from the information given. | | AO3 | **Determine** | Obtain the only possible answer. | | AO3 | **Discuss** | Considered + balanced review with range of arguments; opinions clearly stated + supported by evidence. | | AO3 | **Explain** | Give a detailed account including reasons or causes. | | AO3 | **Predict** | Give an expected result. | | AO3 | **Show** | Give the steps in a calculation or derivation. | | AO3 | **Sketch** | Diagram/graph (labelled). General shape or relationship with relevant features. | | AO3 | **Suggest** | Propose a solution, hypothesis or other possible answer. | ## 7. Extended Response Question (ERQ) anatomy — Nikhil's spec ### Core architecture - **Worth**: 10–16 marks per question - **Structure**: Stem (real-world / experimental / theoretical context with data, graphs, diagrams) → parts (a, b, c) → sub-parts (i, ii, iii) - **Strict cognitive ladder per question**: - Part (a) — AO1+AO2 — "State / Define / Outline / Draw" - Parts (b)(c) — AO2+AO3 — "Calculate / Determine / Derive / Show that" - Final parts — AO3 — "Explain / Evaluate / Suggest / Discuss" ### Five hard design specs 1. **Topic integration.** ERQ rarely stays inside one sub-topic; standard practice is to cross-pollinate two themes. Example: kinematics of charged particle in E-field (Theme A) ↔ quantum energy states (Theme E). 2. **Error Carried Forward (ECF).** A mathematical error in part (a) must not doom part (c). Mark scheme awards method marks if the student applies correct subsequent physics to a wrong earlier value. 3. **"Rule of One."** 1 mark = 1 distinct operation. A 3-mark "Calculate" implies: (i) correct formula/concept, (ii) correct substitution, (iii) correct final answer with **units + significant figures**. 4. **Assumptions & limitations.** The final discriminator part typically asks the student to critique the model: "Suggest one reason why the calculated value differs from the experimental value" — testing the gap between idealised physics and real-world physics (air resistance, ideal-gas assumption, internal resistance, etc.). 5. **Process over product.** Heavy weighting on the *communication* of physics — logical steps, clean diagrams, precise vocabulary. A correct final number with no working ≠ full marks. ### Why current AI generators miss this bar - They generate parts in a vacuum, not on a cognitive ladder - They keep questions inside one sub-topic (no cross-pollination) - They don't write a mark scheme that supports ECF - They don't ask the assumption-evaluation question at the end - They award "1 mark per part" rather than "1 mark per distinct operation" ## 8. Generator contract (what we'll build to) Any future Physics ER generator must, at a minimum: 1. Take input: `(theme₁, sub-topic₁, theme₂, sub-topic₂, total_marks ∈ [10..16], level ∈ {SL, HL})` 2. Generate a **stem** with: (a) novel real-world context, (b) at least one quantitative dataset or diagram, (c) all initial data needed 3. Generate parts (a)→(c) following the cognitive ladder + command-term distribution from §6 4. Generate a **mark scheme** that: - Allocates marks per the "Rule of One" - Explicitly notes which marks are ECF-eligible - Lists "alternative correct methods" for AO3 reasoning marks 5. Include at least one **assumptions/limitations** sub-part in the final discriminator 6. Tag every part with: AO level, sub-topic(s) tested, command term, mark weight 7. Reject any question where the cross-pollination test fails (≥2 distinct sub-topics from ≥2 different themes) ## 9. Math AA / AI mapping (for the math half of the project) The Math syllabus has its own structure (5 topics: Number/Algebra, Functions, Geometry/Trigonometry, Stats/Probability, Calculus) and its own Paper 1/2/3 split (Paper 3 is HL-only). The Christos Nikolaidis corpus in `sources/nikolaidis/` is already organised exactly along these 5 topics with explicit SL/HL flags per chapter — we'll mirror that taxonomy when we get to the Math ERQ generator. Math command-term glossary is published separately (Math AA/AI Guide); to be extracted in Phase 1 when we acquire those guides. ## 10. Open questions for Phase 1 - Do we have access to the **Physics Data Booklet** (the formula sheet students get in the exam)? It's referenced throughout and constrains what's "memorize" vs "look up". Likely on resources.ibo.org behind login. - Do we have access to **specimen papers** for first assessment 2025? IBO publishes these for the new syllabus. - Do we have access to **subject reports** (examiner commentary, distinct from mark schemes)? These describe how students actually performed and which common errors examiners saw — pure gold for generator critic-design.