Data is a value made about something. The number 4 becomes data when it represents a temperature, sales total, age, or voltage; that domain meaning raises questions about units, fidelity, use, and valid ranges. Representation is the work of carrying those answers into code. section 2.3 section 2.8
Data orientation does not discard objects. Objects can encapsulate state and enforce boundaries, while data becomes the primary medium for describing domain facts, decisions, actions, and changing identities. The design pressure shifts from attaching behavior to an entity toward making the represented meaning explicit. section 1.1 section 2.3.1
Model data deliberately:
- Choose types that communicate distinctions a bare primitive or string would hide, then validate at construction so invalid data cannot enter the model. section 3.7
- Use records for compact immutable value and data classes; use sealed interfaces and records to express fixed alternatives and combinations. section 2.8 section 4.9
- Create a purpose-built view when an existing entity carries irrelevant baggage, accepting mapping code at the boundary as its cost. section 5.7
- Keep domain data at the core and adapt databases, cloud stores, HTTP, JSON, and other dependencies around the edges. section 10.4.2 section 10.4.8
Rules as inspectable data
Represent account-routing rules with data types such as Equals, And, Or, and Not rather than burying them in conditionals. A pattern-matching interpreter can evaluate the same rule data; serialization, documentation, visualizations, and a future rule-building UI can reuse it without parsing source code. section 8.5 section 8.10
Data-centric functions are easier to test when they take explicit input data and return output data: scenarios live in the test rather than in clocks, service mocks, or hidden setup. Even side-effects can be specified and asserted as data. section 11.2 section 11.3.3









