Design-Rationale Record (DRR) Method
About this pattern
This is a generated FPF pattern page projected from the published FPF source. It is canonical FPF content for this ID; it is not a FPF Reference product feature page.
How to use this pattern
Read the ID, status, type, and normativity first. Use the content for exact wording, the relations for adjacent concepts, and citations to keep active work grounded without pasting the whole specification.
Type: Governance and authoring pattern Status: Stable Normativity: Normative
- one proposed normative change needs an explicit by-value account of what FPF should say, why this decision is preferred, and which neighboring patterns or selected non-pattern FPF kind-reference pairs it affects
- several patterns or selected non-pattern FPF kind-reference pairs must move together and one external decision record is needed to keep one bounded coordinated change set (one mutually dependent change set) semantically complete while enduring Core text is redistributed
- one bounded content decision question would otherwise force authors to decide the same load-bearing answer separately across several patterns or selected non-pattern FPF kind-reference pairs
- one deprecation, narrowing, or cross-pattern amendment must stay reviewable without reconstructing intent from patch history, chat memory, or scattered notes
Relations
Content
Use this when
- one proposed normative change needs an explicit by-value account of what FPF should say, why this decision is preferred, and which neighboring patterns or selected non-pattern FPF kind-reference pairs it affects
- several patterns or selected non-pattern FPF kind-reference pairs must move together and one external decision record is needed to keep one bounded coordinated change set (one mutually dependent change set) semantically complete while enduring Core text is redistributed
- one bounded content decision question would otherwise force authors to decide the same load-bearing answer separately across several patterns or selected non-pattern FPF kind-reference pairs
- one deprecation, narrowing, or cross-pattern amendment must stay reviewable without reconstructing intent from patch history, chat memory, or scattered notes
Not this pattern when. Do not use E.9 as the permanent location of normative Core law, as a campaign or process brief, or as the main vehicle for purely editorial Delta-0 or Delta-1 cleanup that fits the lightweight variant in CC-DRR.5. Use E.9.DA when one concrete DRR already exists and the question is whether its selected answer, selected-locus obligations, source use, lexical closure, and drafting actionability are adequate for a declared downstream authoring use.
What goes wrong if missed
- Core text changes without one explicit rationale account, so later readers cannot recover which alternatives were rejected or which exclusions were intentional
- coordinated multi-pattern amendments drift apart because the temporary selected-answer account survives only in patches, handoffs, or reviewer memory
- future repairs overfit to local wording and silently lose Pillar, taxonomy-lens, impact-graph, practical-use, or pattern-placement discipline
What this buys
- one external decision record that states the bounded FPF change by value before Core text is rewritten
- one minimum kernel that keeps Problem frame, Decision, Rationale, and Consequences recoverable for later review and replay
- one temporary convergence record for coordinated changes, while keeping enduring Core text in the selected patterns and selected non-pattern FPF kind-reference pairs rather than in the DRR
- one temporary convergence record that fixes the selected answer (the chosen content answer for the bounded content decision question) before later drafting fans out across several selected patterns or selected non-pattern FPF kind-reference pairs
First useful move. State the bounded FPF content decision question, the selected answer, the rationale for that answer, and the selected distribution across patterns or selected non-pattern FPF kind-reference pairs before drafting or landing the Core text.
Cheap stop. If the change is ordinary local wording repair, application of an already accepted pattern, or editorial cleanup that does not change FPF semantics, obligations, boundaries, names, admissible uses, or normative force, do not open a full DRR. Use the lighter governing pattern for the local repair: E.17.AUD.LHR for one overloaded local lexical head inside one publication unit, C.2.P for one episteme, publication, or source-use phrase requiring local epistemic precision restoration, E.10 for general lexical repair, F.18 only when a durable reusable name is being minted, and E.8 for authoring-form correction. Leave E.9 for bounded content decisions that need rationale by value.
Kind-or-boilerplate diagnostic. When a DRR proposes wording for selected patterns, apply F.19 to separate boilerplate apparatus from remaining content before any wording is treated as pasteable pattern prose. If the remaining content still hides wording-use, naming, relation, claim, admissible-use, selected-locus, user-move, or flow-role precision, the DRR names the applied E.10, E.10.ARCH, F.18, or governing pattern. Process, architecture, review, or reference apparatus belongs in its own carrier, not in pasteable pattern prose.
A DRR-proposed wording repair is not pasteable pattern prose until it carries a kind-restoration check. The DRR must show the pre-repair and post-repair object kind, relation or claim kind, slot or use-position, admissible use, and scope, or explicitly decide that the change is a semantic change rather than an editorial repair. A nicer head word, shorter phrase, or removed trigger word is not decision evidence when it narrows a graph into a sequence, turns a method into work, widens an evidence record into assurance, treats a use-position as a new kind, or otherwise changes the kind or use-position without an accepted decision. When the decision depends on slot, lens, role, method, work, evidence, assurance, gate, or decision ontology, the DRR cites the governing pattern rather than redefining that ontology locally.
Primary EntityOfConcern in plain terms. The primary EntityOfConcern here is one external decision-rationale record for one bounded FPF content decision or one bounded coordinated change set. The minimal lens is simple: the record must keep the problem frame, decision, rationale, consequences, and impact and boundary account recoverable enough that accepted content can be distributed into the selected Core patterns and selected non-pattern FPF kind-reference pairs without semantic invention.
Primary working reader. The first working reader is an FPF author, reviewer, or steward who must evaluate, challenge, or land one bounded content decision. Downstream pattern readers benefit from the landed Core text; they are not the primary reader of the DRR itself.
Problem frame
FPF is engineered for Pillar P‑10 Open‑Ended Evolution: its normative rules must adapt as new calculi and insights arrive. But change without a record of why leads to conceptual erosion and undermines auditability. Hence FPF requires an explicit Design‑Rationale Record (DRR)—a durable conceptual record that precedes every normative change.
Problem
Direct edits to the Core, absent a structured rationale, trigger three systemic hazards:
- Lost provenance – future authors cannot infer the reasoning behind a rule; intent decays.
- Implicit assumptions – discarded alternatives vanish from memory, so debates resurface and churn repeats.
- Conceptual drift – incremental tweaks slip past the Eleven Pillars and Principle Taxonomy lenses, blurring the framework’s foundations.
Forces
Solution — the DRR as a structured argument and temporary convergence record
Any proposal to add, modify or deprecate a NORM, A, D, or GOV
rule MUST be accompanied by a Design‑Rationale Record. By default,
a conforming DRR contains at least four conceptual components (below);
these form the minimum decision kernel recoverable by any conforming DRR.
A lightweight editorial variant is permitted by CC‑DRR.5.
In this pattern, a bounded coordinated change set means one bounded
group of mutually dependent content decisions whose enduring FPF
expression will be distributed across several patterns or selected non-pattern FPF kind-reference pairs.
In this pattern, the selected answer means the current set of chosen
content decisions for that bounded content decision question: what FPF should say, which
selected patterns or selected non-pattern FPF kind-reference pairs carry it, what stays outside, and which source-use row, evidence path, validation evidence obligation, or loss/recoverability regime applies.
In this pattern, selected non-pattern FPF kind-reference pair is a tuple-like instruction, not one new kind: when a DRR selects a non-pattern publication, view, record, or relation to carry durable content, it must name the FPF kind named by value and reference by value, for example pattern profile, U.View, source map, source-use note, authoritySourceRef target, evidence-path record, review-finding record, or architecture-decision record.
In this pattern, a temporary convergence record means one external
decision record that temporarily holds the selected answer while
the selected Core patterns and selected non-pattern FPF kind-reference pairs are still being updated.
A nontrivial DRR may therefore govern one bounded coordinated change set. In that case the DRR is the temporary convergence record for the selected answer until selected Core patterns and selected non-pattern FPF kind-reference pairs are updated; it is not a second permanent Core-law section.
Minimum decision-inspection content blocks
A conforming DRR must also make the following decision-inspection content blocks
recoverable. They may appear inside the four kernel components or inside one
dedicated Decision grounds used or decision-inspection block, but they are part of
substantive DRR adequacy rather than later review-only hardening.
These decision-inspection content blocks are not separate process paperwork. A DRR that keeps only the four labels while leaving decision grounds, first-minute use question, naming, selected content distribution, pattern or selected non-pattern FPF kind-reference pair sufficiency or necessity, overlap handling, impact, or unresolved uncertainty implicit is structurally labeled but still substantively immature.
Together these decision-inspection content blocks let the DRR act as one decision record for one bounded coordinated change set: enough semantic closure that later drafting distributes the selected answer into selected patterns and selected non-pattern FPF kind-reference pairs rather than inventing it for the first time pattern by pattern.
When one bounded decision coordinates several patterns or selected non-pattern FPF kind-reference pairs, or one cluster of mutually dependent pattern edits and selected non-pattern FPF kind-reference pair edits, the DRR MAY carry additional substantive sections beyond that minimum kernel. Typical substantive additions include obligations on selected patterns and selected non-pattern FPF kind-reference pairs, one explicit new-pattern vs existing-pattern decision, one impact or non-goal map across selected patterns and selected non-pattern FPF kind-reference pairs, coverage or agreement maps across selected patterns and selected non-pattern FPF kind-reference pairs, convergence classification, and one provisional decision-law account by value that keeps the bounded change account semantically complete until enduring Core text is distributed.
Such additions do not change the DRR’s kind. A DRR carrying them remains conforming only when it stays about the FPF content decision: what FPF should say, why, what is excluded, how selected patterns and selected non-pattern FPF kind-reference pairs are affected, and what practical use or authoring action improves. A DRR carrying richer convergence content MUST NOT become a campaign plan, process script, baton carrier, packet checklist, staging log, or other development-process brief.
When one selected answer could plausibly fit one already-existing pattern or selected non-pattern FPF kind-reference pair or require one newly proposed pattern or selected non-pattern FPF kind-reference pair, the DRR must decide that sufficiency/necessity question by value. It is not enough to list a tentative carrier list or leave downstream drafting to discover the selected pattern or selected non-pattern FPF kind-reference pair later.
When the accepted decision grounds or the DRR itself already names one pattern or
selected non-pattern FPF kind-reference pair as part of the distribution question, that
pattern or selected non-pattern FPF kind-reference pair is not a neutral future watch item. The DRR
must classify it now either as one selected pattern or selected non-pattern FPF kind-reference pair
with explicit obligation, one explicit boundary neighbor kept unchanged,
one inherited-unchanged neighbor, or one outside-current-decision item
with named pattern, selected non-pattern FPF kind-reference pair, or decision record. Conditional or
time-relative pattern prose or prose for one selected non-pattern FPF kind-reference pair such as most likely, may need local hardening, if later touched, watch later, or one equivalent
placeholder is non-conforming there because it marks one unmade current
decision rather than one explicit current disposition.
When accepted decision grounds expose one potentially reusable selected non-pattern FPF kind-reference pair or neighboring source-use, evidence, assurance, validation, or architecture-decision mechanism, the DRR must not merely note that such content already exists. It must decide whether that content is generalized now, kept local with a substantive reason, rejected, or marked outside the current decision with a named pattern, selected non-pattern FPF kind-reference pair, or decision record.
When one selected answer involves source-loss mode, simplification, redaction, summarization, or other declared loss, the DRR must make the admissible-use template explicit by value. Explanation alone is not enough; the decision must say what remains preserved, what is dropped, which branch reading is admissible and which selected non-pattern FPF kind-reference pair carries it, which uses lack an admissible carrier or evidence path, what recoverability class applies, and what reopen or stop rule governs cases that exceed the declared source-loss or scope-narrowing state.
A nontrivial DRR is mature enough for downstream authoring only when material selected-answer branch choices about the EntityOfConcern, selected patterns and selected non-pattern FPF kind-reference pairs, outside-current-decision boundary, reusable-content disposition, and loss/recoverability regime have already been selected, rejected, inherited unchanged, or placed outside the current decision with a named pattern, selected non-pattern FPF kind-reference pair, or decision record. If those choices are still missing, the DRR is still decision-grounding work rather than one accepted design-rationale record.
The DRR lives outside the normative Core. An accepted DRR SHALL be landed by applying its Decision account and any stabilized enduring content to the relevant pattern or selected non-pattern Core kind-reference pair as explicit normative or informative text (the change is "in the Core"; the DRR is not). A richer DRR MAY remain the temporary convergence record while redistribution into selected Core patterns and selected non-pattern FPF kind-reference pairs is still incomplete, but it SHALL NOT remain the permanent sole semantic carrier once landed Core text exists.
Authors drafting from an accepted DRR MAY elaborate examples, SoTA‑Echoing, recognition sections, local wording inside the selected patterns and selected non-pattern FPF kind-reference pairs, and neighboring fit. They SHALL NOT silently revise the selected answer, selected patterns and selected non-pattern FPF kind-reference pairs, outside-current-decision boundary, reusable-content disposition, or declared loss/recoverability regime. Any such revision SHALL be handled through one successor DRR or other named successor decision record.
A DRR may itself be improved through E.23, but the DRR remains the selected decision record, not a full pattern draft. When SoTA is load-bearing in that improvement, it must mutate the selected answer, selected-locus obligation, boundary, example, validation obligation, or reopen condition; otherwise it is rationale-only or lineage-only for the DRR.
To preserve P‑2 Didactic Primacy without duplicating meta‑text, authors landing an accepted DRR SHOULD distill stable and reusable parts of its Rationale, Consequences, and other valid convergence sections into the appropriate informative sections of the affected pattern(s) (Rationale, Consequences, SoTA‑Echoing, Archetypal Grounding; per the Pattern Template, E.8). The full DRR remains external as provenance.
A substantive DRR is one current content decision object. It may carry selected content obligations only when they are part of the Decision or Consequences. It MUST NOT carry next-gate state, handoff/packet state, process-order state, monolith status, future campaign planning, or one hidden promise that the same current content decision question will be decided later inside the same decision object. Any undecided remainder must be marked outside the current decision with a named pattern, selected non-pattern FPF kind-reference pair, or decision record.
Process-source method admission into FPF
When a DRR imports stable method from process-source document-carried method description into FPF, it must decide the admission by value rather than treating process prose as a second canon.
The DRR names:
- the process-source passage or accepted source named by value process-source decision-ground item being considered;
- the reusable FPF method recovered from that passage;
- the current FPF pattern, section, or accepted
DRRthat already carries the method, if any; - the remaining delta that current FPF does not yet carry;
- the selected FPF pattern chosen to carry that delta;
- process-control material excluded from FPF pattern prose, such as role dispatch, seam state, helper behavior, Git recovery, packet transport, review transport, chat cadence, and mutable release state;
- the source-use result for that passage or decision-ground item: quote named by value, narrowed scope, instantiated case, decision-bearing use, draft-guidance source, example-only use, or retired source use;
- any meaning loss or addition created by that source-use result: changed scope, relation, evidence path, admissible use, non-admissible use, reader move, or recoverability condition;
- the first improved FPF use that the admitted method gives to an author, reviewer, or downstream FPF user;
- the current disposition: selected now, inherited sufficient, rejected now, or outside the current decision with the named evaluation pattern, accepted
DRR, or accepted decision-ground item named by value.
Reusable process-source method is not limited to semio wording or pattern-authoring language. It may enter FPF only when it is separable from local process mechanics, improves FPF use, and has one exact evaluation pattern. After the method lands in FPF, process documents should cite the selected FPF pattern instead of keeping a parallel long-form rule.
Archetypal Grounding (System / Episteme)
Bias-Annotation
Scope: this bias annotation is universal for FPF semantic changes governed by E.9. It does not turn project-management state, helper state, or review logistics into DRR content.
Conformance Checklist
Common Anti-Patterns and How to Avoid Them
Consequences
Rationale
FPF evolves by explicit, reviewable deltas rather than silent edits. The DRR is the minimum structured argument—and, when several patterns or selected non-pattern FPF kind-reference pairs must move together, an allowed temporary convergence record that keeps P‑10 Open‑Ended Evolution compatible with P‑1 Cognitive Elegance and P‑2 Didactic Primacy.
E.9 sets a floor, not a ceiling: every conforming DRR must make Problem‑frame / Decision / Rationale / Consequences recoverable, but it may carry richer substantive coordination content when that prevents shadow documents or semantic invention during distribution into Core patterns and selected non-pattern FPF kind-reference pairs. The same floor also requires the decision-inspection content that later authoring and review otherwise reconstruct manually: exact decision grounds, use-value, first-minute working situation, scenario grounding, alternatives, current disposition map, naming/ontology obligation, selected content distribution, existing-pattern sufficiency/new-pattern necessity, overlap classification, selected-answer stability, impact/boundary graph, practical payoff, and any remaining uncertainty that materially shapes the decision.
Pointer-based DRRs (CC‑DRR.1a) prevent duplicated prose, and distribution into Core patterns and selected non-pattern FPF kind-reference pairs (CC‑DRR.4) keeps the specification itself learnable without turning the DRR into a permanent shadow canon. Process-law ordering, gate, and handoff records stay outside because they are not part of the content answer that FPF is selecting.
SoTA-Echoing
E.9 aligns with contemporary architecture-decision and rationale-capture practice, but its contribution is not the existence of a decision record. ADR practice already carries compact context, decision, and consequence records. FPF uses the DRR as a decision-rationale record for one bounded FPF content decision, with enough by-value rationale to distribute durable content into selected patterns and selected non-pattern FPF kind-reference pairs.
The practical gain is content-selection quality under semantic load: the DRR decides the selected answer, alternatives, losses, boundary, and selected loci before pattern drafting begins. Any durable rule, example, or content obligation that remains useful after acceptance belongs in the selected FPF pattern or selected non-pattern FPF kind-reference pair, not in the DRR as a permanent shadow canon.
When a DRR relies on a source document, workstream plan, campaign queue, external review packet, standard, article, ADR-like note, or prior accepted decision, it states how the source is used and the source adoption/adaptation/rejection decision, then carries the selected payload by value: adopt, adapt, reject, lineage-only, rationale-only, selected payload, rejected or non-carried payload, source loss, selected locus, non-use boundary, and reopen condition. A cited source is not FPF doctrine, child DRR, review result, gate, evidence sufficiency, or monolith landing source by citation alone.
Relations
-
Instantiates: P‑10 Open‑Ended Evolution, P‑2 Didactic Primacy
-
Template governed by:
pat:authoring/pattern‑template(E.8) -
Interacts with:
pat:guard/bias‑audit(E.5.4) via lens check -
Complemented by:
E.9.DAwhen one concreteDRRfollows E.9 form but its adequacy for downstream drafting, host amendment, accepted-decision carry-through, source-use carry-through, or selected-locus distribution is disputed or materially relevant.E.9.DAreads theDRRdecision-adequacy claim; it is not a second DRR form, review gate, or mandatory ordinary editorial step. Also complemented bypat:authoring/code-of-conduct(E.12) for etiquette in DRR debate. -
Coordinates with:
E.23when oneDRRis being improved through repeated quality-improvement passes.E.9keeps theDRRkind and decision-record form;E.9.DAsupplies the decision-adequacy object-under-improvement evaluation when adequacy is being improved;E.23governs the repeated method rather than turning the DRR into final pattern prose.
E.9:End
Last Updated: 2026-05-29 — this section last modified in upstream FPF commit 2e112078 (github.com/ailev/FPF)