# Desk use, critique and review · 30 September 2026

Four roles below are **author simulations**, not participant feedback. An independent agent separately reviewed teaching, source cells, every actual PDF and browser behaviour. Root used the running finder through the in-app browser. No classroom/child/family/local device/access trial or measured learning gain occurred.

## Teacher perspective

I followed three-thing guide → documented key → willing reader/private repair → own data → fresh E → later snapshots → unknown → optional lookup → own useful guide → fresh F. Prepare approved broad props/view outside the session and print one useful aid. Cards are item-indexed entries, not a spatial map. Ten starts allocate 3 + 4 + 5 + 7 + 4 + 2 minutes; actual pacing remains unpiloted. Several forms/check prompts may need further short sittings. Read one entry/prompt at a time, preserve first response before worked examples/key, distinguish actual handling/child direction/described plan and adult operation. Criteria identify observed scope/support/next moves without an ability score or full-standard verdict.

## Parent/carer perspective

I checked ten optional three-minute bridges: one approved thing, pretend mark, willing helper or private keeping, empty category, later record and gap-spotting question. No purchase, printer, extra screen time, returned work, home map, family disclosure, photo or upload needed. Stop after one entry or pass; home participation is not assessment credit. Actual school records remain private.

## Child perspective · author simulation

The practical question is “What does this guide tell us?” Choose a tiny collection/purpose/key, help a reader and spot a gap. Art/train/pretend repair/parcel/story contexts are invitations, not compulsory interests. Quiet/private work, exact words/sign/AAC, actual handling and adult-directed placement are choices. The finder changes representation while keeping data; it doesn't replace the child's own guide. Dense source/assessment language belongs to adults reading one short prompt. No speed, spelling, neatness, forced touch/public display or fixed learning style determines success. Interest/comprehension still need actual child use.

## Funder/reviewer perspective

Original freely reusable content includes ten starts, 30 routes, 20 worked domains, ten home bridges and ten extra invitations, two fresh checks with a key and criteria, and nine print pairs with exact alternatives and finder/paper route. Exact source/rights/asset receipts support review. They do not establish time savings, demand, gains, all-state adoption or a complete grade. QCAA's indexed code discrepancy is retained against canonical workbook evidence; no authority endorsement is implied.

## Independent findings and fixes

1. An invented practice reader quote sounded like an actual reply. It is now explicitly conditional; no real reply supplied.
2. Day44 heading assumed child authorship. It is now “Whose key is this?”; actual own key versus supplied model stays distinct.
3. Materials promised two views; final finder has three. All documentation matches Pictures/Words/Symbols.
4. Day41 assumed a supplied view marker. It now agrees/names or marks actual tabletop view and states item-indexed cards aren't a spatial map.
5. Narrow table fragmented CLOTH/MIDDLE/RECORDED. It now uses three explicit source/time cards; all nine entries retained. Targeted 320/390 JavaScript on/off recheck confirmed readable whole words, exact associations and fit.

Root additionally shortened source options after actual 320px native label clipping. “Guide D · incomplete” now fits visibly; full source/time/unknown result stays explicit. Only option labels changed, not records/renderer/key/paper data.

## Executed evidence

- Fresh official workbook HTTP 200 / 2,049,917 bytes / pinned SHA; three actual Foundation rows 20098 / 20105 / 20111 and import/crosswalk wording independently checked. Complete current official indexed QCAA 230317 and ACARA work-sample result compared; final direct-fetch/current PDF-byte limits recorded.
- All nine actual searchable single-page A4 PDFs rendered/inspected independently and as root contact sheet; fresh E/F at full size inspected. Whole SVG/PDF words, exact alternative reading order/diagrams, source positions/unknowns/key marks and measured margins pass. Untagged PDFs/tactile guidance are not PDF/UA or physical-access certification.
- Nineteen temporary regenerated outputs byte-identical: nine SVGs / nine PDFs/exact alternatives. Default source/content/print/JS/link/hash audit passes. Generator/verifier/probes formatted/linted.
- Ten temporary-copy defects rejected: source row, timing, fabricated reader wording, fresh answer leak, invented unknown location, mobile source mismatch, network API, printed unknown changed, wrong actual PDF and stale receipt. Original bytes/sizes/modification times unchanged.
- Independent browser review: 81 states, covering 27 source/item/view states at 1280/390/320. Actual data/key geometry/Words/no-picture/unknown, keyboard/reset/reload, no-JavaScript, zero errors/outside requests and empty storage passed. Targeted post-layout 320/390 with JavaScript on/off confirms all nine source/time/card associations and whole words; unchanged renderer/data did not need another full sweep.
- Root actual in-app use: D/BOX Words; D/BOX Symbols with triangle/meaning/NOT RECORDED; keyboard C/BOOK; reset and reload to A/BOOK/Pictures. At 320px page client/scroll both 305. Screenshots inspected, shorter option visibly fits. Review tab closed/viewport reset. Stopped local server caused one failed initial preview; restored fresh preview completed walkthrough. No learner details entered/submitted.

## Remaining actual-use work

Local educator programme/source/safety/access/device/privacy/permissions review; teacher/child/family trials and feedback; actual child representation/physical forms/own-data decisions; actual running-system purpose/input/output where offered; real willing reader response where chosen; appropriately reviewed translations. Missing opportunities remain not offered/not observed. Several dated samples/local interpretation needed for reporting. None is inferred from authored models or technical checks.