docs: align M01.1 cohort operations and release evidence - #11
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: be610dad4b
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| 1. `project-system` Learning Lab; | ||
| 2. `m01-drill-system` cold choice — first choice is frozen after feedback; |
There was a problem hiding this comment.
In the runbook's “Exact session sequence,” completing the project-system Learning Lab is listed before taking m01-drill-system, even though app.js renders that drill inside the lab as step 01 and labReady() requires it for lesson completion; the same inversion occurs for system-diagnostic. A facilitator following this literally will either expose the worked example and technique before the supposedly cold choice or encounter an already-completed, frozen drill at the next step, invalidating the cold-decision evidence. Replace the lesson entries with an explicit open/start action and follow the rendered order through the cold drill before teaching content.
Useful? React with 👍 / 👎.
| - first choice; | ||
| - short reason; | ||
| - whether the learner chose or seriously considered a credible near-miss; | ||
| - feedback changed reasoning: `yes / no / unclear`; |
There was a problem hiding this comment.
For sessions without spontaneous think-aloud, this capture list does not tell the facilitator to obtain the reason and near-miss consideration before the learner selects an option. The inspected app.js radio-change handler renders option-specific feedback immediately and then freezes the cold choice, so asking for these observations afterward lets the explanation contaminate the purported pre-feedback rationale and makes “feedback changed reasoning” impossible to assess reliably. Require the learner to state and record the rationale/near-miss before clicking, then capture the post-feedback change separately.
Useful? React with 👍 / 👎.
| - [ ] P01 completed and valid. | ||
| - [ ] P02 completed and valid. | ||
| - [ ] P03 completed and valid. | ||
| - [ ] P04 completed and valid. | ||
| - [ ] P05 completed and valid. |
There was a problem hiding this comment.
When any original session is excluded, the runbook explicitly requires recruiting another participant until five valid sessions exist, but this checklist permanently requires every one of P01–P05 to be valid. For example, excluding P02 and completing a valid P06 leaves the cohort eligible under the decision rules while this evidence checklist can never be completed, and the fixed summary tables likewise have no replacement row. Track five valid participant IDs dynamically or add replacement slots instead of requiring the original five IDs.
Useful? React with 👍 / 👎.
1 participant
Status
Canonical M01.1 cohort/release operations pack. Supersedes PR #9. Issue #4 now points to this pack and contains the exact external blocker.
Scope
Docs/operations only. No product runtime changes.
What this adds
RED→GREEN
be610dad4b2aa2c1fbc13366a9387e02ac4c12b6— success.Exact production under test
c4f8ac2af6a66efdb79881baa58d4c091a0e6aed33b328d3ccc753cc3d0aee15da98bccd6913f92eGuardrails
mainchanges;Remaining empirical gate
At least five real eligible learner sessions P01–P05. Synthetic/AI sessions do not count. After ≥5 valid sessions, record exactly one outcome: PROMOTE / REVISE / REJECT. Until then Phase 2 remains blocked.