Highlights
- DRY is about knowledge, not code — every piece of knowledge should have a single, unambiguous, authoritative representation in a system; this is the philosophical core of MDDE (the model as the one representation, everything else derived) and of a vault where every concept has exactly one canonical home.
- Write code that writes code — Thomas and Hunt explicitly champion code generators for anything repetitive and pattern-shaped; two decades before generated warehouse pipelines were mainstream, the craft canon already said the generator is the professional response to repetition.
- Keep knowledge in plain text — plain text is durable, diffable, scriptable, and outlives every proprietary format; an entire markdown vault of tens of thousands of files can rest on this single tip, and the bet keeps paying.
- Orthogonality: change one thing without breaking another — designing components so concerns don't entangle is what makes systems evolvable; Data Vault's split of identity, relationship, and context is orthogonality applied to warehouse schemas.
- Fix broken windows before they define the neighborhood — one tolerated piece of rot licenses the next; equally true of codebases, data-quality debt, and knowledge vaults, and the reason hygiene runs on a cadence instead of on guilt.
- Tracer bullets: build a thin end-to-end slice first — getting a minimal working path through all layers beats perfecting components in isolation; the pilot-before-full-run instinct, whether the pipeline is code or content.
- Your knowledge portfolio needs regular, diversified investment — the authors treat learning like asset management: invest routinely, diversify, review; the compounding-brain mindset stated as an engineering discipline back in 1999.
- Ruthless, automated consistency — automation of builds, tests, and checks is presented as non-negotiable craft, because humans drift and scripts don't; rules enforced by machinery — the third pillar, avant la lettre.
Highlights on this page are generated with the help of AI.