Highlights
- A controlled vocabulary is a shared-meaning contract — Lippell's practitioners treat the taxonomy as the agreement layer that sits upstream of any system or model; the same role the business glossary plays before a data model can conform to anything.
- Taxonomies are living artifacts that need governance — the collection stresses maintenance, ownership, and review cycles over one-off design projects; vocabularies are rules, and rules without an owner rot — the third-pillar argument in information-architecture form.
- Choose the structure depth deliberately — flat lists, hierarchies, faceted taxonomies, and ontologies each fit different retrieval needs; picking the lightest structure that works is a design decision, not a default.
- Tagging quality determines retrieval quality — metadata applied at capture time is what makes content findable later; the frontmatter-first, tags-as-data discipline of a knowledge vault is this principle at personal scale.
- Buy-in and iteration beat big-bang design — the case studies repeatedly show taxonomies succeeding through stakeholder involvement and incremental rollout, the same delivery lesson MDDE projects learn about models imposed versus models adopted.
- Structured vocabularies are what machines consume — taxonomies power search, navigation, recommendation, and increasingly AI retrieval; content structured against a vocabulary is content an LLM can actually use — the "knowledge, structured for AI" thesis with library-science credentials.
Highlights on this page are generated with the help of AI.