Appendix B — Engineering Moves

B.1 Make One Source Authoritative

Problem. Consequential facts are duplicated across consumers. A service relationship may appear in deployment configuration, policy, documentation, and tests; a timeout may appear in callers, workers, and orchestration. Once those copies can change independently, disagreement becomes an ordinary state.

Move. Choose one machine-readable representation as the authoritative source of the fact. Downstream consumers should query it or derive from it rather than maintain independently editable copies.

Figure B.1-1 contrasts the two states.

C.1 Make one source authoritative: from many disagreeing copies to one representation others derive from Before: three copies of the same fact each feed a separate consumer, and because each copy can change independently, disagreement between them is an ordinary state. After: one authoritative representation carries the fact, and the three consumers derive or query it from that single source, so disagreement becomes detectable or impossible. BEFORE AFTER copy A copy B copy C consumer A consumer B consumer C AUTHORITATIVE REPRESENTATION A B C “same fact”, three times disagreement is an ordinary state derive / query disagreement becomes detectable or impossible Choose one representation to carry the fact, make it machine-readable, and derive the consumers from it. Downstream artifacts derive from or query the authoritative representation rather than maintaining independent copies.
Figure B.1-1. One authoritative representation. BEFORE — copies A, B, C each feed a consumer and can drift apart, so disagreement is possible. AFTER — one authoritative representation, with A, B, C derived or queried from it, so disagreement becomes detectable or impossible.

Example — Service policy. DocAble represents permitted service flows in a structured service model. Network policy and wiring derive from that model rather than maintaining independent accounts of permitted communication. A service relationship is authored once and propagated to its consumers.

Example — Timeout ordering. Scattered wall-clock budgets are represented in one timeout-budget model whose nesting relation is checked mechanically. The model makes the required ordering explicit: inner budgets must fit inside outer ones. Without it, that rule remains scattered across implementations.

Related mechanisms: Service-flow / API model · Executable Source of Truth · Model-driven codegen · Timeout-budget ordering model · Required-configuration-per-role manifest · Synchronization model.

© James C. Davis, 2026–present