Highlights
- Domain models recur across industries — Fowler's central observation is that businesses that look unrelated share deep conceptual structures (Party, Accountability, Measurement, Accounting); if models recur, they can be catalogued — and what can be catalogued can eventually be generated.
- The Party pattern dissolves the person/organization split — modeling the abstract actor once instead of duplicating structures per subtype is proto-hub thinking: identify the stable business identity, hang the variation off it.
- Knowledge level vs. operational level — Fowler separates the layer that defines what configurations are allowed from the layer that records what actually happened; this is a 1996 formulation of metadata-driven design, where the knowledge level is precisely what MDDE turns into generator input.
- Accountability abstracts every org-chart into one structure — hierarchies, matrix reporting, and contracts collapse into typed relationships between parties; a lesson in modeling the relationship as a first-class thing, which Data Vault later institutionalized as the link.
- Observations and measurements deserve their own model — separating the thing measured from the act of measuring (with units, protocols, and time) prefigures satellite-style context: descriptive facts change on their own clock, apart from identity.
- Patterns are starting points, not straitjackets — Fowler insists a pattern is adapted to context, never pasted; the same stance MDDE takes on templates: the pattern supplies the density, the model supplies the specifics.
- Conceptual models outlive the software built on them — the book's patterns transferred intact from Smalltalk-era systems to today's stacks; the model layer is the durable asset while implementations rotate underneath.
Highlights on this page are generated with the help of AI.