docs: make M01 cohort executable - #9
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: 4768e0b36c
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".
| 2. Reset M01 validation data or use a fresh browser profile. | ||
| 3. Confirm that no previous baseline result is visible. |
There was a problem hiding this comment.
When P02 or a later participant reuses the facilitator's browser and follows the offered reset path, the reset button removes only pm01-validation-m01-v1; lesson completion, drill answers, feedback, and workbook text remain in pm01-state-v1. The next participant therefore opens already-completed Learning Labs with the prior participant's answers and feedback visible, contaminating the learning intervention and potentially invalidating most of the cohort. Require a fresh profile or explicitly clear both storage states before every session.
Useful? React with 👍 / 👎.
| 1. `project-system` Learning Lab; | ||
| 2. its decision drill; |
There was a problem hiding this comment.
Completing each current Learning Lab already requires the corresponding m01-drill-* as its cold drill and immediately reveals its feedback (m01-learning-lab-data.js reuses the objects from PM01.m01Validation.decisionDrills). Returning to the validation route then presents those same drills again and requires another answer before post-case unlock, so the recorded route answer is not a blind first choice and the requested feedback-change observation is contaminated for every participant following this sequence. The runbook needs a sequence that avoids or explicitly accounts for this duplicate exposure.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 397b4ed59f
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".
| 7. its exit drill; | ||
| 8. workbook/transfer fields required by the lesson. | ||
|
|
||
| The validation route does **not** ask the learner to repeat those cold drills. After both lessons satisfy their evidence-aware completion gates, return to validation and continue directly to post-case. |
There was a problem hiding this comment.
Following this instruction leaves the post-case locked for every participant: m01-validation-app.js computes postUnlocked using allDrillsAnswered(), and tests/m01-app-smoke.test.js:121-142 explicitly confirms that both validation-state drills are required in addition to completed lessons. The fresh evidence in the current tree is that this newly added sentence now tells facilitators to skip those controls, so they can proceed only by answering the same drills again after seeing their Learning Lab feedback, contaminating the choices as previously reported.
Useful? React with 👍 / 👎.
| Participant completes, in order: | ||
|
|
||
| 1. `project-system` Learning Lab; | ||
| 2. its **cold decision drill** — first choice is frozen when feedback opens; |
There was a problem hiding this comment.
If a participant changes a cold-drill selection after feedback appears, the claimed first choice is lost: the Learning Lab radios remain enabled and the handler at app.js:424-433 overwrites lessonState.drillAnswers[drillId] on every change. Because the runbook and evidence form rely on the first choice, either the facilitator must record it before feedback-driven changes or the product must actually freeze/persist it separately.
Useful? React with 👍 / 👎.
| 4. qualitative free-text/interview evidence does not materially contradict the score improvement; | ||
| 5. no unresolved blocking reliability/UX defect materially distorted more than one valid session; | ||
| 6. no repeated rubric ambiguity plausibly explains the observed gains; | ||
| 7. no unresolved learner-driven content-model exception blocks representing the validated M01 contract. |
There was a problem hiding this comment.
This Promote condition cannot be evaluated from the runbook's required evidence set: neither the facilitator closeout nor the minimal form records whether a learner exposed a content-model exception, although M01-VALIDATION-PROTOCOL.md:260-262 requires every such special case to be recorded. A cohort can therefore satisfy the documented checklist and be promoted without any evidence for this mandatory gate; add the content-model observation to each participant record.
Useful? React with 👍 / 👎.
|
Superseded by PR #11, which aligns the cohort operations pack to the published M01.1 release, adds alternative/falsifier/revision-condition evidence, pins exact release evidence, and carries the current RED→GREEN structural contract. Closing this older runbook PR to keep one canonical operations pack. No merge performed. |
1 participant
Purpose
Turn the existing M01 validation protocol into a short executable first-cohort runbook without changing product code.
Adds
Scope
Docs only:
docs/validation/M01-COHORT-RUNBOOK.md.No changes to app code, M02 activation,
main, Pages, storage, or telemetry.Product gate
This does not claim statistical effectiveness. It defines a deterministic product decision rule for the first exploratory cohort so the decision is not invented after seeing results.