Status: author/desk simulation 2026-09-29. No children, classroom teachers, families, community custodians, real audience, outside assessor or physical printer took part. These checks are design reasoning, not observed learning outcomes.
| Lens | Walk-through | Finding / correction |
|---|---|---|
| Educator, Days 31–34 | Can a child work with invented signs without being taught false universal meanings? | Every script refers to the maker’s local legend and adult-stated purpose. A card’s image is never assumed to reveal intent by itself. |
| Educator, Day 35 | Is K a fresh source? | New THREE DOTS → SQUARE legend is absent from A–F solved practice; first response saved before After Check swaps. |
| Educator, Day 36 | What is made? | A completed child-chosen single-panel card is small paper media production. A blank template or adult-authored card is not credited as child work. |
| Educator, Day 38 | What if no willing reader is present or child passes? | Private rehearsal/later/pass remains valid participation, with no actual-audience claim or invented viewer words. |
| Learner, low reading or touch avoidance | Can sign–meaning decisions be made without colour, fluent print reading or hand movement? | Large cards, oral/tactile descriptions, AAC and neutral adult movement serve the same decision targets. No touching is compulsory. |
| Learner, Day 40 | Is GO/PAUSE too easily mistaken for real safety instruction? | Source and script explicitly call it a fictional paper game, not road or emergency signage. Answer accepts multiple non-colour revisions. |
| Funder / assessor | Are counts, source and rights claims reviewable? | Deterministic generator, read-only verifier, SHA256 manifest, exact QCAA code rows and scope holds are present. A desk pass cannot establish classroom impact. |
Self-critique: Many icon design tasks can invite overconfident claims about “clear for everyone.” This pack asks for local, stated meaning and actual reader words only when obtained. A real cross-cultural media encounter is essential later but cannot be fabricated here; the authentic-source slot remains permission and authority gated. A classroom trial should measure whether children can hold two symbol–meaning pairs, whether the large-card work is engaging, whether 25 minutes works with support, whether a real viewer question is neutral, and whether black-and-white copies still preserve distinctions. Adult transcription and material choices require careful attribution to child versus adult. The formative checks are public, so they are for teaching, not secure ranking.
Technical QA at freeze: run the generator, read-only verifier, Ruff, PDF size/font/searchability checks, exact SVG-text alternatives comparison, byte-identical regeneration, and rendered A4 visual inspection. These cannot verify physical print contrast in every setting, actual child accessibility, local source permission or audience interpretation.
Observed desk/technical result: The eight A4 pages were rendered and visually reviewed as one contact sheet. Each local legend and target has a distinct visible relationship; K and L show unsolved starting sources, with no answer panels. No text clipping or panel collision was seen. All eight SVG printed-text streams matched their verbatim alternatives. A second generation produced identical SVG/PDF bytes. The read-only verifier and Ruff checks passed after content review. No physical printer, child or reader trial occurred.