An operation combines or transforms values from a specified set to produce a value. In an algebraic design, operations are deterministic functions: data goes in, data comes out, without side effects or void methods. This makes their meaning open to equational reasoning rather than being hidden inside implementation steps. section 7.2
The useful design question is not how to execute an operation, but what it means. For conflicting records, that might mean defining what it means to “add” two records. The same move applies to business rules: equality checks, conjunction, disjunction, and negation can all be represented as data. section 7.2 section 8.4
Modeling operations as data enables several practical improvements:
- Represent an operation explicitly instead of burying it in conditional control flow. section 8.3
- Use enums for attributes instead of unrestricted strings, allowing the type model to reject invalid attribute names. section 8.3
- Unify related operations under a sealed interface, forming a closed algebra of equality, AND, OR, and NOT. section 8.4
- Add constructors and fluent methods after the model is correct, reducing the verbosity of composing operations. section 8.4
A composable rule operation
A rule can express “country is US and segment is Public, or region is not LATAM” as data: eq(COUNTRY, "US").and(eq(SEGMENT, "Public")).or(not(eq(REGION, "LATAM"))). Unlike a fixed conditional branch, this representation can be transformed as data.


