Middle loop: project and paper¶
Experiments become an argument through review and revision. A capable review step is useful only when findings lead to verified changes.
Working notes¶
- Provenance audits connect paper claims to runs and datasets.
- Review findings do not reliably survive the transition into a revised draft.
- Judge calibration needs known outcomes and negative controls.
- Git and document-editor synchronisation remain operational work.
- Draft quality and reviewer scores need separate validation.
Source: course evidence synthesis, §5. These are selected observations, not general performance guarantees.
Maturity, in Oct 2026
Strong capability, immature workflow: audits and drafting work, but review-to-fix convergence still needs deliberate tracking and human-built support.
Before the course¶
- [TODO: select a compact worked example and its verification artifact.]
- [TODO: rehearse the live walkthrough and prepare a fallback.]