# Author desk run-through · teacher, learner and potential funder

**Date:** 29 September 2026. This is an author simulation and local file/browser inspection, **not** a classroom trial, independent mathematics review, assistive-technology study, real measurement or funder evaluation. No learner outcomes or live site deployment are claimed.

## Teacher path

I opened [Day 2](LESSONS.md#day-2--extract-a-square-factor), [Card B](LEARNER-CARDS.md#card-b--find-a-square-hiding-inside-day-2), and the [square-factor A4 mat](print/square-factor.pdf). I could model `72=36×2`, ask learners to produce `6sqrt(2)`, then check by squaring before giving an optional new context. The five time blocks add to 25 minutes and identify a next move for the common false sum rule. On [Day 5](LESSONS.md#day-5--subtract-and-transfer-public-check-a), the new Check A values are separate from all card and swap values, and a 12-minute first-response window precedes the public key. In a real class I would adjust duration after seeing prior algebra knowledge.

## Learner path

I solved [Check A](STUDENT-CHECKS.md#check-a--day-5-fresh-surd-transfer) on paper before opening the key: `sqrt(243)=9sqrt(3)`, `sqrt(48)=4sqrt(3)`, so difference `5sqrt(3)≈8.7` cm. I tested the proposed `sqrt(195)` by squaring both positive expressions: 75 and 195 are unequal. I then used the labelled-point route on [Check B](STUDENT-CHECKS.md#check-b--day-10-fresh-graph-and-surd-transfer): zeros 2 and 8, midpoint axis 5, vertex `(5,9)`, y-intercept `(0,-16)`, graph positive for `2<x<8`; the independent ratio becomes `5sqrt(11)/11`. This confirmed the key values, not learner usability.

## Supporter/funder path

From [README](README.md), a supporter can find ten clean cards, public feedback checks, separate key, optional swaps, complete print text, and a paper route for the offline aid. The [source crosswalk](SOURCE-AND-CROSSWALK.md) exposes the exact v1.3 syllabus, page positions, sub-topic hours and unfinished quadratic work. A funder can inspect SHA-256 hashes and original asset rights. The pack does not claim a completed topic, actual workplace data, secure assessment, or measured learning gains.

## Visual, interactive and arithmetic inspection

I rendered all five PDFs to images at 700-pixel height and inspected a contact sheet. Titles, exact/approx notation, worked examples, tactile lines and the five-point parabola stayed on-page with no observed clipping. The PDF text is selectable, every sheet is A4, and the graph's coordinates and axis are reproduced in [full text](print/TEXT-ALTERNATIVES.md). This is visual inspection, not user testing with low-vision or blind learners.

I opened the [offline square-factor explorer](surd-explorer.html) in a headless browser with sample inputs `72`, `18`, `200`, `1`, `0` and `243`; outputs matched expected exact forms (`6sqrt(2)`, `3sqrt(2)`, `10sqrt(2)`, `1`, validation message, `9sqrt(3)`). I also inspected a full-page browser screenshot: controls, result, and paper route were visible at desktop size. The explorer uses no network request and makes no data claim beyond 1–5000 positive integers. The [verifier](verify_pack.py) checks lesson counts, links, source pin, known exact results, assets, interactive structure and file hashes.

## Critique and next work

- Eight surd sessions give 3h20, under the syllabus' 4 notional hours; the teacher should revisit division, rationalisation, fluency and error diagnosis rather than treat this as certified coverage.
- Two quadratic sessions introduce features of vertex and factor forms. The 7-hour sub-topic still needs general form, equation methods, completing the square, quadratic formula, discriminant, technology and genuine modelling with data.
- The three routes demand identical exact notation and explanation but were simulated by an adult author. Teachers should adapt access and pacing after observing their own learners; route choice should not be assigned by a fixed learner label.
- Contexts are invented to keep the materials freely usable and avoid false claims about measured objects. An actual modelling task needs real data, assumptions and review of model limits.