Illustration 1.1 — Six answers the book produces for any problem¶
Referenced in Chapter 1. Six equally-weighted cards, arranged in two rows: two Tier-0 refusals (produced above the method) on top, four G3 placements (produced inside the method) below. Monochrome only — strokes and text in the theme's fg colour, fills in the theme's bg colour. No hue does any work. Equal weight across all six cards is deliberate: the book produces all six answers, not just the last two. Four of the six are non-AI outcomes.
Encoding without colour¶
- Equal cards, with row grouping — same border weight and typography across all six. The two top cards are wider (they are alone in their row); the four bottom cards share their row. No card is visually privileged by stroke or fill. This matches the book's claim that a method producing only the last two answers is a sales funnel, and one that produces only the first two is a manifesto.
- Small-caps labels — Off-limits to AI, Not a tech problem, Human-operated, Non-AI automation, AI as assistant, Autonomous AI rendered in the same letter-spacing and weight. Hierarchy is refused on purpose.
- Horizontal rule under each label — a quiet half-opacity line that separates the name from the description without introducing colour or shading.
- Row bands — two micro-labels ("Refusals — at the two Tier-0 checks" and "Placements — at G3 Route") tell the reader where in the method each row is produced. The bands carry the only hierarchical information in the illustration; they do it with letter-spacing and opacity, not with weight or hue.
- Caption strip — the bottom line ("Six answers — four of them non-AI — each available, each defensible, none presumed") states the thesis the illustration is there to support.
Dark-mode test¶
Toggle the theme (top right). Borders, strokes, rules, labels, captions — all flip to white on black via var(--md-default-fg-color) and var(--md-default-bg-color). No per-theme SVG, no asset duplication, no second export.
Where SVG + monochrome shines¶
- Printable as-is. Zero render cost to PDF or paper.
- Accessible:
<text>nodes searchable by Lunr;<title>gives screen-reader label. - Reviewable in git: a line edit is a visual change, diffable.
Where it pushes back¶
- Equal-weight cards give the reader no "look here first" cue. That is the whole point of the illustration — but if a reader skims the page, they may not feel the force of "all six are available". The chapter text around this illustration has to carry that weight.
- Two-line labels were introduced for the bottom row ("Human-operated", "Non-AI automation", "AI as assistant", "Autonomous AI") to keep the card widths narrow enough to fit four across. The label font is 12px at 0.06em letter-spacing; further expansion of the label vocabulary would need either a third row or a widening of the viewBox.