C.32.ADA - Architecture Decision Adequacy Scales
Type: Architecture evaluation pattern under C.32 Status: Stable Normativity: Normative unless marked informative
C.32.ADA:1 - Problem frame
Use this pattern when a project architecture decision, its method docking, or its ADR-like publication projection must be evaluated for adequacy before use, review, handoff, governance, or improvement.
Primary working reader: an architect, reviewer, or architecture-responsible practitioner checking whether a project architecture decision is good enough for a declared use and which repair should happen next.
Typical entry phrases:
"The decision is written, but can developers use it?"
"The ADR looks complete; is the architecture decision itself adequate?"
"Which part is weak: candidate basis, trade-off, method instruction, work split, or publication projection?"
"We need a scale like E.21, but for architecture decisions rather than pattern quality."
"Do not average the decision; tell us what must be repaired."
First-minute use slice. An OrderFlow decision is to guide developer work, but its ADR says only “use events”. Read the decision and the available implementation instruction against the intended result: can developers recover the Method, inputs, readiness conditions and expected structural effect? If the instruction supplies part of that way but omits a needed readiness condition, assign the supported docking value and name the missing condition and its source. If the cited instruction is unavailable, leave that value unestablished and request the needed material; unavailable evidence is not established absence. Continue through the complete coordinate set, preserving supported results and the declared readiness floor. The evaluation yields bounded repairs or a use conclusion; approval and implementation commitment retain their own grounds.
The primary governed object is ArchitectureDecisionAdequacyEvaluation@Project: a C.32.ADA-local evaluation record over one ArchitectureDecisionRelation@Project, optional ArchitectureDecisionRecordProjection@Project, and declared use. It is not the evaluated decision, evaluation Work, or result episteme.
ArchitectureDecisionAdequacyEvaluation@Project is a local record form, not a new U.* kind, gate, evidence, assurance, pattern-quality evaluation, or replacement for C.32.PAD. Its coordinate table expresses the ADA result content; when that result must be a durable claim, one separately identified C.2.1 episteme states it. The dated evaluation remains separate U.Work, and any actual evaluation operation application remains with its subject pattern.
A decision can appear complete because it has a record, rationale, or diagram, while it is unusable for the declared work. Weak candidate basis, hidden trade-offs, missing method instructions, absent source-return, and vague supersession conditions remain invisible until implementation or review fails.
The project can evaluate architecture decisions by complete coordinate set, keep kinds distinct, and repair the weakest live coordinates without turning adequacy into a single score.
Ordinary working move: declare the evaluation use, assess every coordinate under :4.2, and give each supported value its rationale. Retain a grounded notTriggered disposition or an unestablished dependent value when appropriate. State the needed repair or missing basis at its smallest useful source locus.
Adoption test: after using C.32.ADA, another practitioner can recover the declared use, complete coordinate results with their rationales or missing bases, repair targets, and the supported use conclusion or reason it remains open.
Not this pattern when the current object is FPF pattern quality, measurement validity, evidence support, assurance, gate passage, candidate synthesis, comparison, selection, local choice, or ADR publication projection itself. Use the pattern for the next question named in Relations.
The first useful output is the complete coordinate evaluation and its use or repair conclusion. ArchitectureDecisionAdequacyEvaluation@Project can organize it when this structured record is useful:
ArchitectureDecisionAdequacyEvaluation@Project:
projectWorkOccurrenceRef?: U.EntityRef constrained to U.Work
architectureDecisionEvaluationProjectUseRelationRef?: U.RelationRef governed by the exact evaluation-use or work-use pattern
evaluationId:
claimScopeRef: U.EntityRef referencing one U.ClaimScope
selectedContextSliceRefs:
effectiveReferenceScheme:
referencePlane?: when interpretation depends on it
evaluationWindow?: when the judgement or its use depends on time
decisionQuestionInputProjectionRef?: when a selected projection is used
evaluatorSystemRef?: U.EntityRef constrained to U.System
evaluatorSystemRoleKindRef?: U.KindRef
evaluatorSystemRoleClassificationJudgmentRef?: U.RelationRef
evaluatorAssignmentSpeciesRef?: U.RelationKindRef constrained under U.SystemRoleAssignment
evaluatorAssignmentOccurrenceRef?: U.RelationRef constrained to U.SystemRoleAssignment
evaluationWorkRef?: U.EntityRef constrained to U.Work
evaluationPerformedUnderAssignmentRef?: U.RelationRef constrained to one obtaining F.6 performedUnderAssignment relation
evaluationOperationApplicationRefs?: subject-pattern relation or A.6.1 application references
adequacyResultEpistemeRef?: U.EpistemeRef
declaredUse:
architectureDecisionRelationRef:
architectureDecisionRecordProjectionRef?:
coordinateValues:
- coordinateRef:
triggerStatus: evaluated | notTriggered | unestablished
notTriggeredReason?: required only for notTriggered
missingBasis?: required only for unestablished; identify the needed source or judgement
value?: 0|1|2|3|4|5; required only for evaluated
valueLabel?: absent|namedOnly|partiallyExpressedForDeclaredUse|sufficientlyExpressedForDeclaredUse|wellExpressedForDeclaredUse|exceptionallyExpressedForDeclaredUse; required only for evaluated
adjacentValueRationale?: required only for evaluated
evidenceOrSourceRefs?
repairPatternRef?
repairInstruction:
strongestBlockingCoordinates:
noAveragePolicy: true
stopCondition:
reevaluationTrigger:
Here @Project is a compatibility and retrieval cue only. A project-local ADA record names both the composite U.Work in projectWorkOccurrenceRef and the obtaining record-use relation in architectureDecisionEvaluationProjectUseRelationRef; the evaluated decision’s own project relation, the suffix, or either field alone is insufficient. evaluatorSystemRoleKindRef and evaluatorSystemRoleClassificationJudgmentRef remain optional and separate. When a particular dated evaluation Work occurrence is claimed, the evaluator first has its A.13 core and evaluationWorkRef names Work independently admitted under A.15.1. An ordinary evaluation retains the operation, material conditions and support needed for its result. Assignment species, occurrence, and evaluationPerformedUnderAssignmentRef are optional and appear only when the record or receiving use represents precise assignment-bound attribution; any present F.6 ref uses the same obtaining A.13 assignment, and its absence or failure alone leaves independently admitted Work intact. If the finding questions a premise consumed by that admission, recheck it and dependent conclusions. Any operation application and result episteme remain separate. Kind, classification, assignment, Work, attribution, application, responsibility, and result do not substitute for one another.
C.32.ADA:2 - Problem
Architecture-decision adequacy concerns several distinct objects. A decision relation can be strong while its ADR projection is weak. A record can be readable while the decision lacks candidate traceability. A candidate basis can be strong while method docking is absent. A method instruction can be clear while the architecture-characteristic trade-off is hidden.
A single overall grade or average score hides these differences. Adequacy must be evaluated by coordinates tied to the declared use. “Ready for internal architecture review”, “ready for developer work”, “ready for ADR publication”, and “ready for governance enforcement” can require different stop conditions, but each use needs complete coordinate inspection.
C.32.ADA supplies an E.21-shaped ordinal evaluation pattern for architecture decisions. It uses the E.21 value domain and labels directly, then defines architecture-decision coordinates over PAD relation, method docking, publication projection, structural description, characteristic trade-off, and evolution. Weak coordinates point back to C.32.PAD, C.32.ADR, C.30.AD, A.15, C.32.ACS, C.32.ACE, C.16, C.25, or another subject pattern.
C.32.ADA:3 - Forces
| Force | Tension |
|---|---|
| Decision complexity | One architecture decision carries candidate, structure, characteristic, method, work, publication, and evolution content. |
| Declared use | A decision can be adequate for discussion and inadequate for developer work or governance. |
| Repair focus | Review must point to the smallest repair locus, not produce a general complaint. |
| No average | A strong rationale cannot compensate for absent work split or missing source-return. |
| Kind precision | Eval results, evidence, assurance, gates, and publication records must stay distinct from adequacy evaluation. |
| Evolution | The evaluation must expose reopen and supersession weakness before the decision becomes stale. |
C.32.ADA:4 - Solution
Evaluate the complete coordinate set for one declared use without averaging its values. Identify the blocking weaknesses and the repair that can make the intended use adequate. Retain a sufficient result; select further improvement when its useful contribution warrants the work.
C.32.ADA:4.1 - Shared value meanings
Use the same ordinal value domain and labels as E.21. ADA specializes what counts as expression for architecture-decision adequacy; it does not create a second scale.
| Value | Label | Architecture-decision adequacy meaning |
|---|---|---|
0 | absent | The coordinate is not expressed for the declared architecture-decision use. |
1 | namedOnly | The coordinate is named or implied, but cannot support reliance, action, evaluation, or repair. |
2 | partiallyExpressedForDeclaredUse | The coordinate is present but incomplete, fragile, misplaced, or too narrow for the declared use. |
3 | sufficientlyExpressedForDeclaredUse | The coordinate can support the declared use in the current project, with limits visible. |
4 | wellExpressedForDeclaredUse | The coordinate has clear refs, boundaries, source-return, and repair path for likely project changes. |
5 | exceptionallyExpressedForDeclaredUse | The coordinate is well expressed and transferable across another team, later slice, or adjacent holon kind with minimal recovery work and no hidden neighbor loss. |
Values are ordinal content evaluations. They are not measures, averages, votes, maturity ladder names, evidence weights, assurance levels, gate statuses, or implementation approval.
The result-bearing coordinate row uses the E.21 label domain with an architecture-decision coordinate:
| Coordinate | Value | Label | ShortRationale |
|---|---|---|---|
<ADA coordinate> | <0..5> | <E.21 label> | <for 1..4, why the lower adjacent value would understate the expressed content and the higher adjacent value would overstate it; for 0, why 1 would overstate it and what would raise or reopen; for 5, what makes 4 too weak and what would lower or reopen> |
5 is not required for every use. Stop conditions are declared before evaluation. A lower diagnostic floor may be used for exploration or internal discussion, but it does not make the decision ready for developer work, implementation commitment, or governance enforcement.
C.32.ADA:4.2 - Complete coordinate set
Evaluate every coordinate. If a coordinate is not live, mark it notTriggered only with a short reason grounded in the declared use. An established absence of required content can warrant a low value; an unperformed check or unavailable basis leaves the dependent value unestablished. Name that missing judgement, retain supported coordinate results and keep the complete evaluation as draft until its required basis is available. Do not use notTriggered to conceal that gap.
Assess each statement at the source needed for the decision. An omission in a memo can make the memo unusable for its audience. It leaves open whether the governing decision contains the missing provision. To judge transfer to another team, identify which information and help that team would have and which choices it must recover. Reuse adequate evidence for that coordinate and its stated use. Claim that another team actually succeeded only when its performance is evidenced.
| Coordinate | What is evaluated | Repair when weak |
|---|---|---|
BoundedDecisionQuestionRecoverability | Decision subject, described holon, exact U.ClaimScope, relevant A.2.6 U.ContextSlice membership, effective reference scheme, applicable plane, any time qualification required by the decision or use, status, and decision question can be recovered; a selected BoundedModelUseStructure is named only when it independently changes interpretation. | Repair the missing decision-subject, scope, slice, or model-use-structure assertion using C.32.PAD, A.2.6, or A.1.1. |
CandidateBasisAndSelectionTraceability | Candidate palette, residual frame, comparison, selection, selected set, or reason no candidate-set question is live is recoverable. | Repair the missing candidate, comparison, or selection assertion using the applicable content in C.32, C.32.MLAO, A.19.CPM, A.19.SelectorMechanism, G.5, or C.11. |
AffectedStructureAndDescriptionAdequacy | Affected selected structures, views, architecture descriptions, correspondence, structural-information lens uses, and source-return are recoverable. | Repair the missing structure, description, correspondence, or lens-use assertion using C.30, C.30.ASV, C.30.AD, A.6.F, A.6.M, or C.29. |
ArchitectureCharacteristicTradeoffAdequacy | Architecture characteristics, criteria rows, Q-Bundles, eval readings, accepted losses, and guardrails are explicit. | Repair the missing characteristic, criterion, reading, loss, or guardrail assertion using C.32.ACS, C.32.HCS, C.25, C.32.ACE, C.16, C.31, or C.31.ASAP. |
MethodAndWorkDockingAdequacy | Method-use instructions identify what a future performer must be able and authorized to do, the conditions to start and the expected effects on structure. Name the intended System when readiness depends on its identity. For performed work, identify the acting System and the extent of its Work. Actual Work retains its obtaining A.13 assignment basis even when the record omits separate assignment fields. Explicit assignment refs and F.6 are added for precise assignment-bound attribution; responsibility remains independently governed. | Repair the missing MethodDescription, Method, A.13 performer basis, A.15.1 Work fact, optional assignment species and F.6 attribution, responsibility predicate or missing governor, readiness, or expected-effect assertion using A.15, A.15.1, A.15.2, A.15.5, E.8, E.11.PUR, A.6.RCD, or C.24. |
ArchitectDeveloperSplitAdequacy | Structures fixed by the decision, refinement left open to developers, holon-transition or BOSC-triggered boundary refs, and source-return condition are explicit. | Repair the split or claim-kind assertion using C.32.PAD, A.15, or B.2.P; use B.2 only when whole reidentification is triggered. |
PublicationProjectionAdequacy | ADR-like or other publication projection carries the needed section functions for the declared readers. | Use C.32.ADR for the needed projection and E.24.PUB for the publication occurrence, form, carrier, audience and availability; add E.17 when a source-backed face and source return are needed. |
EvidenceEvalAndGateExitAdequacy | Eval, evidence, assurance, gate, or institutional-governance assertions are named only when live, with their exact predicates and subject-pattern locators. | Repair the assertion using C.32.ACE, C.16, A.10, B.3, A.21, or the named institutional-governance content. |
EvolutionAndReopenConditionAdequacy | Reopen, supersession, stronger-source return, and changed-context triggers are clear. | Repair the reopen, supersession, archive/front, improvement, or source-currentness assertion using C.32.PAD, C.32.FAIL, C.18, C.19, or E.23. |
TransformerTransformedCorrespondenceAdequacy | Required correspondence between a separately typed influence-source architecture and the transformed-side architecture is recoverable when the decision depends on it. Keep a modal or unresolved synthesis frame distinct from an exact row about an obtaining influence occurrence under C.32.CONWAY. | Use C.32.CONWAY to repair the correspondence frame or exact pair row. Use A.15, A.3.4, A.3.4.P, or E.18 only for a separately current Method/Work, actual-change, or flow-structure claim. |
NonOverreadAndSubjectAssertionAdequacy | The decision, description, publication, Method, eval, evidence, assurance and gate claims retain their distinct meanings and applicable definitions or tests. Exact predicate or ClaimGraph identity is needed when the claim or intended reliance depends on it. | Repair the overread, relation, wording, or name using A.7, A.6.P, E.10, F.18, or the subject pattern for the unresolved question. |
ConsequenceAndRepairGuidanceAdequacy | Consequences, accepted losses, weak coordinates, and next repair instructions are actionable for the declared use. | Repair the missing consequence, projection function, or coordinate assertion using PAD, ADR, or the coordinate-specific subject pattern. |
C.32.ADA:4.3 - Use-specific stop conditions
Declare the use before scoring. Common uses:
| Declared use | Ordinary stop condition |
|---|---|
| Internal architecture discussion | Every triggered coordinate is evaluated; 0 absent coordinates block reliance, and values below 3 sufficientlyExpressedForDeclaredUse name the patterns or sources that must be repaired. |
| Ready for architecture review | No triggered coordinate below 3 sufficientlyExpressedForDeclaredUse; candidate basis, trade-off, affected structures, work split, and reopen condition are strong enough for reviewers to inspect by value. |
| Ready for developer work or implementation commitment | Every triggered coordinate is at least 4 wellExpressedForDeclaredUse. A lower diagnostic floor belongs to a separately declared diagnostic use and is not implementation commitment. |
| Ready for ADR-like publication | Publication projection, section functions, status, source-return, and supersession are at least 4 wellExpressedForDeclaredUse; if the record will guide developer work, use the developer-work floor too. |
| Ready for governance enforcement | Every triggered coordinate is at least 4 wellExpressedForDeclaredUse; enforcement status requires a separate current gate, evidence-use, assurance, or governance result. |
Use these as ordinary defaults. A project can declare stricter stop conditions. It must not weaken a triggered coordinate by hiding it under an average, and it must not call a diagnostic result ready for developer work or governance enforcement.
C.32.ADA:4.4 - Small complete evaluation slice
This slice additionally records a dated evaluation and precise assignment-bound attribution. ArchitectureReviewService-4 has the A.13 performer basis for decision-adequacy evaluation. A.15.1 independently admits DecisionAdequacyEvaluationWork-12, performed from 10:00 to 10:20 on 2026-08-12, enacting DecisionAdequacyEvaluationMethod-2 within ProjectArchitectureReviewService-4 : U.System. The obtaining ArchitectureReviewerAssignment-6 is of the declared species ArchitectureReviewerAssignment, has that recovered performer as holder and covers the Work. The separate F.6 relation uses that same obtaining A.13 assignment. These are the grounds for the Work and attribution fields below; the coordinate result is a separate episteme.
ArchitectureDecisionAdequacyEvaluation@OrderFlow:
declaredUse: readyForDeveloperWork
claimScopeRef: OrderFlow developer-work readiness for the named decision and release slice
selectedContextSliceRefs: OrderFlow service, named product-family release, and current developer-work window slices
effectiveReferenceScheme: OrderFlow architecture decision scheme edition 4
referencePlane: selected architecture and developer-work commitment
evaluationWindow: review session 2026-08-12, 10:00–10:20
decisionQuestionInputProjectionRef: PAD decision relation plus its declared-use and source-return fields
evaluatorSystemRef: ArchitectureReviewService-4
evaluatorAssignmentSpeciesRef: ArchitectureReviewerAssignment
evaluatorAssignmentOccurrenceRef: ArchitectureReviewerAssignment-6
evaluationWorkRef: DecisionAdequacyEvaluationWork-12
evaluationPerformedUnderAssignmentRef: performedUnderAssignment(DecisionAdequacyEvaluationWork-12, ArchitectureReviewerAssignment-6)
adequacyResultEpistemeRef: DecisionAdequacyResult-12
architectureDecisionRelationRef: PAD:order-flow-event-integration
architectureDecisionRecordProjectionRef: ADR:order-flow-event-integration
noAveragePolicy: true
stopCondition: every triggered coordinate >= 4 wellExpressedForDeclaredUse
strongestBlockingCoordinates:
- MethodAndWorkDockingAdequacy
- ArchitectureCharacteristicTradeoffAdequacy
result: repairBeforeUse
| Coordinate | Value | Label | Short rationale and repair |
|---|---|---|---|
BoundedDecisionQuestionRecoverability | 4 | wellExpressedForDeclaredUse | Subject, holon, exact claim scope and selected slices, scheme and plane, window, status, and question are explicit for the named use, so 3 understates their clarity and boundaries; 5 would require grounds for the claimed reinforcement beyond this local reading, such as a working source return. |
CandidateBasisAndSelectionTraceability | 4 | wellExpressedForDeclaredUse | Candidate palette and selected option are cited in a recoverable local path, so 3 understates the explicit basis; 5 needs a supported reinforcing use. For instance, a prior independently recovered selection can already supply it under comparable access. |
AffectedStructureAndDescriptionAdequacy | 4 | wellExpressedForDeclaredUse | Module and information structures and the C.30.ASV refs make the affected structures and their local description use recoverable, so 3 understates that explicit fit; 5 needs evidence that the description sustains the stronger use, such as an established working source return. |
ArchitectureCharacteristicTradeoffAdequacy | 3 | sufficientlyExpressedForDeclaredUse | Substitutability gain and latency loss support a bounded reading of the trade-off, so 2 understates what is usable; incomplete guardrail eval rows prevent 4. Repair through C.32.ACS, C.32.ACE, C.25, and C.16. |
MethodAndWorkDockingAdequacy | 2 | partiallyExpressedForDeclaredUse | The directive to use events and the assignment-bound attribution requirement give more content than 1 namedOnly. The implementers still need the Method, required capabilities and readiness conditions for the intended change. This record also claims that implementation Work was performed under ServiceTeamAssignment. Recover that actual Work and its A.13/A.15.1 grounds. For the additional assignment-bound claim, establish the species, obtaining assignment and F.6 attribution. Future instructions may identify the assignment to be secured before implementation. |
ArchitectDeveloperSplitAdequacy | 3 | sufficientlyExpressedForDeclaredUse | The acting Systems and fixed-versus-open Work split are clear enough for a bounded local reading, so 2 understates them; the missing source-return threshold for schema refinement prevents 4. Repair that PAD boundary. If responsibility is claimed, cite its direct predicate, participants, applicability, and identity or return the missing governor; use B.2.P or B.2 only when the corresponding level or whole question is current. |
PublicationProjectionAdequacy | 4 | wellExpressedForDeclaredUse | The ADR maps section functions to the current reader use, so 3 understates the explicit local projection; 5 needs evidence of the stronger publication use, which an already documented update or supersession can supply. |
EvidenceEvalAndGateExitAdequacy | 3 | sufficientlyExpressedForDeclaredUse | Named evaluation and gate continuation conditions give a usable bounded next step, so 2 understates them; their instructions do not yet make the required commitment checks recoverable, preventing 4. Repair the triggered assertions through C.32.ACE, C.16, A.10, B.3, or A.21. |
EvolutionAndReopenConditionAdequacy | 4 | wellExpressedForDeclaredUse | Latency and schema-version-pressure triggers state local reopen conditions, so 3 understates their clarity; 5 needs grounds for the reinforcing use, such as a documented supersession under these conditions. |
TransformerTransformedCorrespondenceAdequacy | notTriggered | — | The OrderFlow decision in this slice does not rely on the locally mentioned toolchain/product-structure correspondence. Keep that observation as a cue; reopen this coordinate and use C.32.CONWAY if the correspondence becomes decision-relevant. |
NonOverreadAndSubjectAssertionAdequacy | 4 | wellExpressedForDeclaredUse | The decision, ADR, Method, eval, and gate claims retain their separate subject assertions and patterns, so 3 understates the explicit local distinctions; 5 needs evidence for the reinforcing use, such as a prior decision whose interpretation turned on this distinction. |
ConsequenceAndRepairGuidanceAdequacy | 4 | wellExpressedForDeclaredUse | Consequences and coordinate-specific repair loci give actionable instructions for this use, so 3 understates that guidance; 5 needs evidence for a reinforcing use, such as a repair decision in another supported setting. |
PAD adequate, ADR weak. A fixture architecture decision relation can reach 4 wellExpressedForDeclaredUse on every triggered PAD, Method, work-split, trade-off, and reopen coordinate while the trade-study memo omits status and supersession. ADA identifies only the missing publication-projection assertion and cites C.32.ADR; it does not rewrite the PAD relation.
ADR readable, PAD weak. A Markdown ADR can have clear headings, status, context, decision, and consequences while the project relation lacks candidate basis, affected selected structures, and Method/Work docking. ADA identifies those missing assertions and cites C.32.PAD, C.32, and A.15 as their subject-pattern locators; template completeness does not make the architecture decision adequate.
C.32.ADA:5 - Archetypal Grounding
Developer-work readiness. A service decision explains the candidate and trade-off, but the instruction “teams should use events” leaves the event boundary, Method, required capabilities and start conditions open. Its MethodAndWorkDockingAdequacy for the intended implementers is 2 partiallyExpressedForDeclaredUse. The decision can prescribe how a future shift should implement a change before its members are assigned. For an instruction to start now, establish who can and may perform it. If the report goes on to claim actual Work or its performance under a particular assignment, establish those occurrences under A.13/A.15.1 and F.6 as applicable. Any claim of responsibility has its own domain predicate.
ADR-publication readiness. The manufacturing trade-study memo omits status and supersession, so its readers cannot establish which instruction is in force from this copy. For that use the projection merits 2 partiallyExpressedForDeclaredUse. Open the decision or another authoritative record if the task also asks how adequately the underlying decision sets out its reopen conditions. A complete decision with an incomplete memo and a decision that itself lacks those conditions require different repairs. Until the relevant decision content is available, leave its coordinate unresolved. When it is available, assess that content and repair the memo through C.32.ADR.
Architecture review. A method-family architecture decision has candidate options and Method instructions, but no declared architecture characteristics. ADA gives ArchitectureCharacteristicTradeoffAdequacy = 0 absent. The repair states the missing characteristic assertions using C.32.ACS and C.25 before review can judge the decision.
Governance enforcement. A toolchain-product correspondence decision depends on team and tool structures. ADA evaluates TransformerTransformedCorrespondenceAdequacy; if the correspondence refs are absent, the repair states the missing correspondence assertion using C.32.CONWAY before institutional governance can constrain Method use.
C.32.ADA:6 - Bias-Annotation
| Risk handled | How C.32.ADA handles it |
|---|---|
| Record completeness as adequacy | ADR projection is one coordinate, not the whole evaluation. |
| Average-score drift | Coordinate values are not averaged; weak coordinates drive repair. |
| Generic review prose | Each weak coordinate names a governing repair pattern. |
| Gate confusion | ADA does not approve, certify, assure, or pass a gate. |
| Hidden method failure | Method and work docking has its own coordinate. |
| Evolution blindness | Reopen, supersession, and source-return have explicit coordinates. |
C.32.ADA:7 - Conformance Checklist
| Requirement | Required result |
|---|---|
CC-ADA-1 | Declared use and stop condition are written before evaluation. |
CC-ADA-2 | The evaluated decision relation and optional ADR projection are cited. |
CC-ADA-3 | Every coordinate has a supported 0–5 value, a grounded notTriggered disposition, or an explicitly unestablished dependent value with its missing basis. Keep the complete evaluation draft while a required judgement is unestablished, retaining supported results. |
CC-ADA-4 | Every value has adjacent-value rationale, not only a number. |
CC-ADA-5 | No coordinate values are averaged or converted into one global score. |
CC-ADA-6 | Weak coordinates name repair pattern refs and repair instructions. |
CC-ADA-7 | Evidence, assurance, gate, measurement, eval, publication, Method, Work and pattern-quality claims retain their distinct meanings and applicable definitions or tests; exact ClaimGraph identity is cited when the claim or reliance needs it. |
CC-ADA-8 | A project-local ADA record names both projectWorkOccurrenceRef and architectureDecisionEvaluationProjectUseRelationRef; the evaluated decision’s relation, the suffix, or either field alone asserts no locality. |
CC-ADA-9 | A local evaluator kind and a System-classification judgment use separate optional refs and neither requires an assignment. A claim about a particular dated evaluation Work occurrence first recovers the evaluator through A.13 and names independently admitted A.15.1 Work. Ordinary evaluation retains the operation, material conditions and support needed for its result. Assignment species, occurrence, and F.6 refs are optional and appear only when the ADA record or receiving use represents precise assignment-bound attribution; a missing or failed F.6 relation alone leaves independently admitted Work intact, with any questioned admission premise rechecked as above. An operation application and result episteme remain separate. Keep those objects and relations distinct, preserving the admission dependencies stated above; a label or record does not establish the other facts. |
CC-ADA-10 | Each evaluation binds its exact claim scope, relevant context slices and reference scheme, preserving every declared selector and any plane, time qualification or input projection required by the judgement or receiving use. The declared-use label and coordinate table do not replace those conditions. |
C.32.ADA:8 - Common Anti-Patterns and How to Avoid Them
| Anti-pattern | Symptom | Repair |
|---|---|---|
DecisionAdequacyAverage | Strong rationale and readable ADR produce a high average despite absent method docking. | Remove the average; use weakest triggered coordinates to choose repair. |
ADRCompletenessAsDecisionAdequacy | The record has all headings, so the decision is treated as adequate. | Evaluate PAD relation, method docking, trade-off, source-return, and reopen conditions separately. |
ReviewCommentWithoutRepairPattern | The reviewer says “unclear” or “not enough detail” without a target repair pattern. | Assign the weak coordinate to C.32.PAD, C.32.ADR, A.15, C.30.AD, C.32.ACS, or another subject pattern. |
GateByScale | A value of 4 or 5 is treated as approval or certification. | Keep ADA as evaluation; use A.21, A.10, B.3, or governance patterns for gate, evidence, assurance, and enforcement claims. |
NotTriggeredAsConvenience | A difficult coordinate is marked not triggered to close the review. | Require a reason grounded in the declared use. If the coordinate is live, score it only when its basis is available; otherwise identify the missing source or judgement and leave the dependent value unestablished. |
MethodDockingSkipped | The decision is adequate for architecture discussion but then used to direct developer work. | Re-declare use as developer-work readiness and evaluate method docking, work split, and publication handoff. |
SystemRoleLabelAsEvaluator | A reviewer system-role kind, assignment, or ADA record is treated as the performer of evaluation. | Recover the exact admitted evaluator System through A.13 and let A.15.1 independently admit the dated evaluation Work. Retain the required A.13 assignment basis; add explicit assignment species/occurrence refs and F.6 when precise assignment-bound attribution is represented. Keep any responsibility relation, result episteme, and operation application separate. |
ContextLabelAsEvaluationScope | A project, domain, or bounded-context label stands in for the evaluation boundary. | Bind the exact U.ClaimScope, selected context slices, reference scheme and every declared selector, with any plane, time qualification or decision-question projection required by the judgement. |
C.32.ADA:9 - Consequences
| Consequence | Benefit | Cost |
|---|---|---|
| Adequacy is coordinate-based. | Review can point to repairs instead of vague approval or rejection. | Evaluation requires inspecting every coordinate and explaining its supported result or missing basis. |
| Declared use controls stop condition. | A decision can be adequate for one use and inadequate for another without contradiction. | Teams must state intended use before scoring. |
| No average is allowed. | Weak but critical coordinates stay visible. | Some dashboards and summaries need redesign. |
| Explicit repair conditions and subject-pattern locators are mandatory. | Review results become actionable. | Reviewers must recover the missing assertion and the pattern description containing its definition or constraint. |
C.32.ADA:10 - Rationale
C.32.ADA exists because architecture-decision adequacy is not one property. It is a family of recoverability and use-readiness coordinates across decision relation, selected structures, architecture characteristics, method docking, work split, publication projection, evidence or eval exits, and evolution.
The pattern follows the same discipline as FPF quality evaluation: declared use first, complete coordinate set, ordinal values with rationale, no average, and targeted repair. It remains architecture-specific because the coordinates are tied to PAD, ADR projection, architecture descriptions, candidate synthesis, method-use instructions, and architecture-characteristic trade-offs.
C.32.ADA:11 - SoTA-Echoing
These sources inform the coordinates, value meanings, stop conditions, and repair exits used below.
| Source to inspect | Why this source is load-bearing here | Transfer into ADA | Blocked overread |
|---|---|---|---|
Current FPF E.21 scale discipline | Existing FPF pattern for declared-use evaluation, exact 0 absent through 5 exceptionallyExpressedForDeclaredUse labels, complete coordinates, adjacent-value rationale, and no averaging. | Reuse the value domain and evaluation discipline for architecture decisions without copying pattern-quality coordinates. | E.21 pattern-quality status and coordinates are not architecture-decision adequacy. |
C.32.PAD and C.32.ADR | PAD and ADR define the decision relation and publication projection being evaluated. | Make relation adequacy and projection adequacy separate coordinates. | A complete ADR does not make the decision relation adequate. |
ISO/IEC/IEEE 42010:2022 official standard (https://www.iso.org/standard/74393.html; IEEE page https://standards.ieee.org/ieee/42010/6846/) with the 42010 companion site as secondary reading (https://iso-architecture.org/42010/) | Current official source for architecture descriptions, viewpoints, views, correspondence, and rationale; the companion site is used only as secondary reading. | Add coordinates for affected structure, architecture-description adequacy, and source-return. | Architecture-description adequacy does not decide or approve the architecture. |
Ford, Parsons, Kua, and Sadalage, Building Evolutionary Architectures, 2nd ed. (https://www.oreilly.com/library/view/building-evolutionary-architectures/9781492097532/) | Current practitioner source for architecture feedback and incremental change under eval-like mechanisms. | Add evolution, reopen, guardrail, and confirmation coordinates. | Source-side fitness-function wording is not imported as ADA object naming. |
Ford, Richards, Sadalage, and Dehghani, Software Architecture: The Hard Parts (https://www.oreilly.com/library/view/software-architecture-the/9781492086888/) | Current practitioner source for trade-off analysis under competing characteristics. | Add architecture-characteristic trade-off adequacy and accepted-loss repair. | Trade-off discussion is not an assurance claim or governance approval. |
2026 ADR violation-detection research (https://arxiv.org/abs/2602.07609) | LLM detection in the study was more accurate for explicit decisions checkable from code than for decisions left implicit or requiring deployment settings or organizational context. | Add confirmation, source-return, method, deployment, and organization refs to adequacy checks when live. | Automated violation detection is not proof, evidence, assurance, or gate passage. |
Source-currentness boundary. Recheck a source row when FPF evaluation discipline, architecture-decision practice, ADR violation checking, evolutionary-architecture eval practice, or project governance changes a coordinate, value meaning, or repair exit.
C.32.ADA:12 - Relations
- Builds on:
C.32.PAD,C.32.ADR,C.32.P2S,C.32,C.32.MLAO,C.32.ACS,C.32.ACE,C.32.CONWAY,C.32.FAIL,C.30.AD,C.30.ASV,A.2.1,A.2.6,A.15,A.15.1,A.19,C.2.1,C.16,C.25,C.29,E.17, andE.21; uses A.1.1 only when a selectedBoundedModelUseStructurechanges evaluation interpretation. - Evaluation boundary: Use ADA only to evaluate architecture-decision adequacy for a declared use. It does not perform candidate synthesis, comparison, selection, selected-set result declaration, actual publication, local choice, evidence support, assurance, gate passage, governance enforcement, or pattern-quality evaluation.
- Decision and projection boundary: Use
C.32.PADto repair the decision relation andC.32.ADRto repair ADR-like publication projection. - Description and structure boundary: Use
C.30,C.30.AD, andC.30.ASVfor architecture claim, description, and view adequacy. - P2S docking: Use
C.32.P2Swhen a weak decision-adequacy row must reopen the connected architecturing flow rather than only repair the decision record. - Method and work boundary: Use
A.15,A.15.1,A.15.2,A.15.5,E.8,E.11.PUR, andC.24for method, work, readiness, pattern-use, and agentic tool-use claims. - Characteristic and eval boundary: Use
C.32.ACS,C.32.HCS,C.25,C.32.ACE,C.16,C.31, andC.31.ASAPfor characteristic rows, Q-Bundles, eval-program framing, typed measurement or evaluation results, modularity, and scale preference. ADA may inspect references to those objects as coordinate evidence; use their subject patterns for any definition change or actual operation. - Evidence, assurance, gate, and governance boundary: Use
A.10,B.3,A.21, and local governance patterns when those claims are live.