Plan — The LTP Explainer Trilogy

Three NYT-style scrolling visual explainers that make the case for thinking in trees, built in sequence. This file is the working ledger: decisions at the top, a checkbox pipeline below, and a position marker for resumable (loop-mode) work.

Position

Done, 2026-08-15. All three pieces shipped in both markdown and HTML forms — see changelog.md. Open questions and next steps now live in NEXT.md and issue #6.

Update this line at the end of every work unit.

Why this exists

story/draft.md ("What Counts") was the first attempt. It is too anchored to Second Renaissance's internal drama — it documents our work rather than answering the universal question. Start fresh.

The target, per the 2026-07-30 outflow (docs/plans/2026-07-30-outline-for-intro-narrative.md):

Whether you're running a business, a team, or any complex project, here's the problem you face, and here's a fundamentally different way of thinking about it.

Decisions locked

The three pieces

Piece 1 — The Bottleneck (working title)

Answers: why don't your fixes stick? Payload: a system's output is set by one constraint; optimising every part is not optimising the whole; local efficiency can actively harm global throughput. Hands off with: "So — find the constraint. But how do you find it? That turns out to be a thinking problem."

Source spine: Dettmer ch1; existing synthesis Parts 1–3.

Piece 2 — Thinking Made Visible (working title)

Answers: why would you ever draw your reasoning as a tree? Payload: reasoning is invisible, so it can't be checked, shared, or accumulated; necessity vs. sufficiency; cause-effect vs. correlation; the Categories of Legitimate Reservation as the etiquette that makes disagreement productive instead of exhausting. Hands off with: "One tree isn't enough — different questions need different trees."

Source spine: Dettmer ch2 (CLR); existing synthesis Part 4. This is the most original piece relative to Goldratt — it is where this project's own thesis lives (claims get addresses; debate accumulates against a stable scaffold; see motivation.md).

Piece 3 — The Six Trees (working title)

Answers: what the Logical Thinking Process actually is. Payload: IO map · Current Reality Tree · Evaporating Cloud · Future Reality Tree · Prerequisite Tree · Transition Tree — each with the question it answers, one beat of the clinic case, one diagram. Ends with: the live demo (claim-tree-annotation-demo/, dashboard).

Source spine: Dettmer ch3–7; existing synthesis Part 5.

Sourcing rules

• docs/The Theory of Constraints and the Logical Thinking Process - Learning to Think Better.md (449 lines) is pillaged and restructured, never merely extended. It is a good essay; a scroller is a different animal — beats, not paragraphs.
• Dettmer (library/logical-thinking-process-dettmer/) read just-in-time: ch1 + front matter for #1, ch2 for #2, ch3–7 per section of #3.
• The Goal — not required as a text. Its set pieces (Herbie on the hike, the dice game, the idle workers) are renderable without it. Revisit if a beat needs the actual prose.
• Every claim in a beat sheet carries a source tag ([D ch1 p12] or [syn Part 2]) so the drafting phase cannot invent.

Pipeline

Each piece runs the same five stages. Each stage is one loop-resumable work unit.

beat sheet → markdown draft → figure specs → SVG/HTML build → publish
             with [FIGURE n]   incl. interactive
             slots             beats

Piece 1 — The Bottleneck

• 1.1 Read Dettmer ch1 + front matter; read existing synthesis in full
• 1.2 Beat sheet — explainers/01-bottleneck/beats.md
• 1.3 Draft v1 (essay form) → draft-v1-prose.md
• 1.3c Converted to shot list — shotlist.md (57 blocks, ~1,020 body words, 0 interactions)
• 1.4 Rufus review gate — redirect here, not later
• 1.5 Drawing production — 57 drawings, hand-drawn/ink register, one shared paper ground
• 1.6 Build explainers/01-bottleneck/index.html — pinned-stage pattern, CSS position: sticky, IntersectionObserver steps
• 1.7 Verify (see below) and publish

Piece 2 — Thinking Made Visible

• 2.1 Read Dettmer ch2 in full
• 2.2 Beat sheet
• 2.3 Markdown draft
• 2.4 Rufus review gate
• 2.5 Figure specs
• 2.6 Build
• 2.7 Verify and publish

Piece 3 — The Six Trees

• 3.1 Read Dettmer ch3–7
• 3.2 Beat sheet
• 3.3 Markdown draft
• 3.4 Rufus review gate
• 3.5 Figure specs
• 3.6 Build + wire the demo handoff
• 3.7 Verify and publish

Cross-cutting (after #1 ships)

• Decide whether story/ "What Counts" is retired, kept, or relinked
• Nav + config.json entries for the trilogy
• Cross-links between the three pieces

Verification

A piece is done when all of these hold:

1. No unearned jargon. Every term is used only after the reader has felt the thing it names. Grep the draft for: throughput, constraint, system, local optimum, sufficiency, necessity, UDE — each first use must follow its felt example.
2. Source-traceable. Every factual/structural claim maps to a source tag in the beat sheet.
3. The prediction beats actually withhold. Each interactive beat must be unanswerable by scrolling past it — the reader commits first.
4. Legible at 375px. Every figure readable on a phone; no horizontal page scroll.
5. Theme-correct. All colour/type/spacing from Organic DS tokens; no hard-coded hex.
6. Reduced-motion respected. Every scroll animation has a prefers-reduced-motion path.
7. Cold read. A reader who knows nothing of TOC reaches the handoff line and wants the next piece.

Open questions

• Running case: clinic is proposed, not battle-tested. If it strains at Piece 3's six trees, the fallback is a school or a small product team — but switching after Piece 1 ships is expensive, so stress-test it during 1.2.
• Whether Piece 1 replaces /story/ on the site nav or sits alongside it.