Defend and revise a thermal device design

1. Retain the initial plan

State the local device need and its connection to a wider resource problem without claiming that this model measures wider impacts. Predict a comparison, define the criteria and plan the evidence. Each field allows 2400 characters.

Unsaved plan text is local to this form until locked.

Choose a report round

Each round has its own one-time design freeze, stress receipt, assignment and response. Changing this selector clears unsaved round forms; retained records and the shared report draft remain. A revised round retains known exposure.

2. Freeze the first design and stress plan

The baseline is fixed at candidate 0: Gain 2 W/K; Standard enclosure; subtract 0 degC sensor correction. Choose two distinct alternatives. Inspect public development cases through the suite lesson if useful, and disclose that exposure. Freeze the selected candidate and prediction before this project shows its stress result.

Unsaved form choices are not retained until this action. The two alternatives must differ from each other and baseline; the selected candidate must belong to the three.

3. Freeze the assignment before its response

For independent review a human supplies a genuinely unfamiliar assignment, complete assumptions and evidence. Record assignment, prior prediction/method, assistance and exposure before saving the response. Shipped public cases and freeze-before-reveal remain practice; identity and unfamiliarity are self-recorded assertions, not authentication. A model run is not a validated physical prototype.

First freeze this round’s candidate comparison and reveal its stress receipt. Assignment controls become available after that evidence is retained.

The form starts with the public sensor-scale task. For a genuinely unfamiliar human assignment, provide its complete evidence, assumptions and identity. Describe a prediction and method before retaining the response. Do not relabel the public task as unfamiliar.

These form edits are not stored until frozen. One assignment is retained per round.

4. Write the first report

Explain the local stakeholder need and wider problem boundary, baseline and alternatives, calibration choice, numerical/graph evidence, failed criteria and a plausible alternative cause. Retain a worsening indicator and a next measurement. Repeating the model is not an independent physical trial. A revision explains exactly what changed and why.

Retained response to the frozen transfer assignment

Freeze an assignment and retain its response above.

Working report fields save locally, up to 2400 characters each. This draft is shared between rounds; submitted copies remain unchanged. Presence checks do not grade scientific quality.

Submission needs the frozen design, revealed stress, frozen assignment, retained response and complete report. It does not require a passing design. Revised work must differ substantively in report, assignment or plan.

5. Inspect submitted snapshots

No submitted report yet.

6. Record human review of a submitted snapshot

Use all seven criteria. Every label needs a specific evidence comment; the first two labels also require a targeted retry. Identity, date, assistance and unfamiliarity are local assertions, not authenticated certification. A public supplied assignment cannot earn an independent-case label. A materially unsupported conclusion needs revision even when arithmetic or presentation is polished.

Changing review target clears unsaved judgments and context. Pending review text is stored only when recorded.

The selected target is not submitted.

Keep a review handoff and restoration backup

The local handoff contains submitted rounds, full model evidence, assignments, responses and reviews. It excludes working drafts and incomplete rounds. Nothing is transmitted. JSON backup restores science work; retain both backups when an immutable conflict is reported.