# Dossier provenance and forecast audit

Reconstruct claims from dated sources and score forecasts without letting later knowledge leak backward.

## Learner edition

## 1. Rebuild a source-to-claim graph

**Task type:** derivation

Select one bounded DIRD-related claim and reconstruct the inferential path from primary sources to present wording. Formalize direct, inferred, contextual, and contradictory edges and identify the weakest dependency.

### Deliverables

- Dated claim graph with stable source labels
- Edge-by-edge warrant table
- A rewritten claim limited to the surviving evidence

### Scoring criteria

- Source chronology and provenance: 8 points
- Inference edges accurately classified: 7 points
- Final claim matches the graph: 5 points

## 2. Calibrate the forecasts

**Task type:** analysis

Create at least ten date-locked forecasts from dossier claims, assign probabilities before inspecting outcomes, then score calibration and discrimination. Compare results with a simple base-rate forecaster.

### Deliverables

- Forecast table with cutoff dates and resolution rules
- Brier scores and calibration bins
- Comparison with the base-rate benchmark

### Scoring criteria

- No temporal leakage: 7 points
- Scoring and benchmark calculation: 7 points
- Calibration interpretation: 6 points

## 3. Run an adversarial dossier review

**Task type:** design

Design a red-team/blue-team review in which both sides receive the same frozen corpus and must publish claim graphs, confidence changes, and decisive missing evidence.

### Deliverables

- Frozen-corpus and role-assignment protocol
- Symmetric scoring rubric and adjudication plan
- Public evidence packet with uncertainty log

### Scoring criteria

- Teams operate on identical evidence: 7 points
- Rubric rewards correction and calibrated uncertainty: 7 points
- Adjudication and release are auditable: 6 points

---

Evidence rule: distinguish calculation, model-dependent inference, experimental observation, and unresolved claim in every response.