07-source-material/ — RDCO bookshelf

The canonical home for raw source material that grounds Ray’s reasoning + the founder’s research.

This directory is the answer to the founder’s 2026-04-30 question: “Where is our ‘bookshelf’ if we need to go back to the source to rediscover the details for a specific situation?”

See ../06-reference/2026-04-30-bookshelf-source-material-architecture-gap for the architectural concept and ../06-reference/2026-04-30-bookshelf-discovery-source-material-inventory for the discovery inventory that produced this initial migration.

Architecture

07-source-material/
├── books/                      ← canonical books (1 dir per title)
│   └── <slug>/
│       ├── source.pdf          (or source.epub, source.txt, source.docx)
│       ├── extracted.txt       (or extracted.md — text version for retrieval)
│       └── metadata.yaml       (title, author, ISBN, source, format, copyright)
├── masterclass/                ← purchased course materials (multi-file)
│   └── <slug>/
│       └── (mixed PDFs + xlsx + docx)
├── articles/                   ← long-form articles saved raw
├── papers/                     ← arxiv / academic papers
├── transcripts/                ← video/audio transcripts
│   └── youtube-and-podcast/    (symlink → 06-reference/transcripts)
├── newsletter-bodies/          ← raw newsletter HTML/text (extension to /process-newsletter)
└── web-archives/               ← saved web articles + blog posts
    └── <author-slug>-<title-slug>/
        ├── source.html         (or source.md if natively markdown)
        ├── extracted.md
        └── metadata.yaml

Per-source slugging

• Books: <author-lastname>-<short-title-slug> → wheeler-understanding-variation, boyd-vandenberghe-convex-optimization
• Masterclasses: <author-or-publisher>-<course-slug> → wheeler-spc-metrics-masterclass
• Web archives: <author-slug>-<title-slug> → cedric-chin-process-behaviour-charts-more-than-you-need-to-know
• Papers: arxiv-<id>-<title-slug> or <conf>-<year>-<title-slug>
• Newsletter bodies (when extended): <sender-slug>/<YYYY-MM-DD>-<topic-slug>.md

Per-source metadata.yaml schema

title: "Understanding Variation: The Key to Managing Chaos"
authors: ["Donald J. Wheeler"]
publisher: "SPC Press"
edition: "2nd"
year: 1993
isbn: "9780945320807"
pages: 161
source_format: "VitalSource ebook → screencap+OCR text extraction"
acquired: "2026-04-21"
acquired_by: "founder (purchased SPC Press → VitalSource)"
copyright: "Personal-use only. NEVER quote at length. NEVER index in QMD without explicit founder OK."
extraction_quality_notes: "Italics sometimes render with stray slashes. Page-number footers reliable."
canonical_citation_format: "Wheeler, *Understanding Variation*, Ch. N, p. M"

Copyright posture (inherited from private/ deprecation)

• Never commit. This entire directory is in .gitignore (line: 07-source-material/).
• Never quote at length in vault notes, public writing, social posts, or any context the founder didn’t approve. Standard fair-use limits apply: ≤15 words in quotation marks per excerpt, only for genuine commentary, never for displacive summary that would substitute for the source.
• Don’t index in QMD by default. The Phase-1 retrieval pattern is grep + Read for specific passages. QMD indexing into a separate source-material collection is a Phase-2 decision pending founder green-light.
• Don’t ship full source material to sub-agents unless the founder asks for a specific extraction. Reading these into context costs ~225KB-740KB per file. The right pattern is grep for the specific concept and read only the matching pages.

Retrieval patterns

Pattern 1 — grep for a concept

grep -n "operational definition" ~/rdco-vault/07-source-material/books/wheeler-understanding-spc/extracted.txt | head -5

Returns line numbers + matching lines. Use Read with offset/limit on those line numbers to fetch the surrounding context.

Pattern 2 — citation chain

When making a strong claim that benefits from grounding:

1. grep the relevant slug in 07-source-material/books/
2. Read the matching passage with surrounding context
3. Cite as: Wheeler, *Understanding SPC*, Ch. N, p. M ([07-source-material/books/wheeler-understanding-spc/extracted.txt:LINE](path))

Pattern 3 — Phase-2 (deferred): semantic search via separate QMD collection

Pending founder green-light: ingest 07-source-material/ into a SECOND QMD collection named source-material (distinct from the rdco-vault collection). Query via mcp__qmd__query with collection: "source-material". Keeps synthesis search and source search independent.

Current inventory (2026-04-30 initial migration)

books/ (17 total, ~488MB)

Founder-purchased / pre-existing:

• wheeler-understanding-variation/ — Wheeler, Understanding Variation (2nd ed) — OCR text
• wheeler-making-sense-of-data/ — Wheeler, Making Sense of Data: SPC for the Service Sector — OCR text
• wheeler-understanding-spc/ — Wheeler & Chambers, Understanding Statistical Process Control (3rd ed) — OCR text
• boyd-vandenberghe-convex-optimization/ — Boyd & Vandenberghe, Convex Optimization (free PDF, Cambridge)
• corr-stagnitto-agile-data-warehouse-design/ — Corr & Stagnitto, Agile Data Warehouse Design (BEAM* Modelling)

Free graduate ML curation (sub-agent download 2026-04-30, per ../06-reference/2026-04-30-dami-defi-12-graduate-ml-textbooks-curation):

• mohri-foundations-of-machine-learning/ — 505p, 5.9MB ✅
• prince-understanding-deep-learning/ — 541p, 21MB ✅
• mlsysbook-machine-learning-systems/ — Vol1+Vol2 = 2152p, 31MB ✅ (note: split into 2 volumes since curation note written; consolidated PDF no longer offered)
• kochenderfer-algorithms-for-decision-making/ — 700p, 12MB ✅
• goodfellow-deep-learning/ — ⚠️ PARTIAL: only front_matter.pdf (9p). Publisher contract (MIT Press) forbids PDF distribution; chapters HTML-only at deeplearningbook.org/contents/.html. On-demand HTML lookup is the workaround.
• sutton-barto-reinforcement-learning-introduction/ — 548p, 70MB ✅
• bellemare-distributional-reinforcement-learning/ — ❌ FAILED: no PDF available; publisher contract forbids; HTML-only at /contents/chapterN. Empty source.pdf slot; see metadata.yaml for retrieval path.
• albrecht-multi-agent-reinforcement-learning/ — 395p, 7.6MB ✅
• kochenderfer-decision-making-under-uncertainty/ — 349p, 9.2MB ✅
• barocas-hardt-narayanan-fairness-and-machine-learning/ — 294p, 2.1MB ✅
• murphy-probabilistic-ml-introduction/ — 860p, 88MB ✅
• murphy-probabilistic-ml-advanced-topics/ — 1370p, 144MB ✅

Free curation result: 10/12 success, 1 partial (Goodfellow front-matter only), 1 failed (Bellemare). ~408MB added.

masterclass/

• wheeler-spc-metrics-masterclass/ — 28 PDFs + xlsx + docx from Wheeler’s SPC Metrics Masterclass

web-archives/

• cedric-chin-process-behaviour-charts-more-than-you-need-to-know/ — Cedric Chin’s PBC explainer (commoncog.com)

transcripts/

• youtube-and-podcast/ — symlink → 06-reference/transcripts/ (367 transcript files: Moonshots, Acquired, Tim Ferriss, Lex Fridman, Dwarkesh, etc.)

Pending acquisitions (per ../06-reference/2026-04-30-bookshelf-discovery-source-material-inventory decision queue)

Free downloads (queued, sub-agent dispatched 2026-04-30):

• Mohri et al., Foundations of Machine Learning
• Prince, Understanding Deep Learning
• MLSysBook (community/MIT)
• Kochenderfer/Wheeler/Wray, Algorithms for Decision Making
• Goodfellow/Bengio/Courville, Deep Learning
• Sutton & Barto, Reinforcement Learning: An Introduction
• Bellemare et al., Distributional Reinforcement Learning
• Albrecht et al., Multi-Agent Reinforcement Learning
• Kochenderfer (long-game), Decision Making Under Uncertainty
• Barocas/Hardt/Narayanan, Fairness and Machine Learning
• Murphy Vol I, Probabilistic Machine Learning: An Introduction
• Murphy Vol II, Probabilistic Machine Learning: Advanced Topics

Paid acquisitions (founder yellow-light 2026-04-30, format-clarification pending):

• Reis & Housley, Fundamentals of Data Engineering (O’Reilly direct DRM-free PDF/EPUB ~$45)
• Deming, Out of the Crisis (MIT Press 2018, format TBD)
• Andy Grove, High Output Management (format TBD; likely Kindle DRM)

Web archive scrape candidates (deferred, founder call):

• Kingsbury Jepsen analyses (~50 articles, ~100MB)
• Commoncog source consolidation for top-cited posts

Skill integrations (Phase 2)

• /save-to-bookshelf <url-or-path> — new skill (in development)
• /process-newsletter — extend to also save raw body to newsletter-bodies/<sender>/
• /process-youtube — symlink already in place; future ingest writes directly under transcripts/
• /cite-from-bookshelf <claim> — Phase-2 retrieval skill (deferred until QMD-indexing decision)

Migration history

• 2026-04-30: initial migration. 5 books, 28 masterclass files, 1 web archive moved into structure. Wheeler material moved out of private/ (now deprecated; see private/README.md). Boyd & Vandenberghe Convex Optimization moved out of 06-reference/textbooks/ (folder now empty, can be removed). Founder approved scaffold-first approach, all 12 free downloads, yellow-light on 3 purchases pending format clarification.