NSTD.6 - Declared-Use Narrative Rendering Quality Evaluation

Type: DPF evaluation pattern body

Primary EntityOfConcern: NarrativeRenderingQualityEvaluationCharacteristicSpace@Context, an evaluation CharacteristicSpace for one evaluated narrative rendering kind and declared use.

NSTD.6:1 - Problem frame

Use this pattern when a team must decide whether one narrative rendering version, including an unaccepted candidate, is good enough for one declared reader or listener use. An editor may evaluate a directly authored or already existing product without knowing how its author produced it. State whether the present question concerns that product’s current quality, a promised source-to-result relation, or assurance about actual construction work; these questions need different evidence.

Evaluated object kind: NarrativeRenderingVersion@Context, an identifiable version of a narrative account in a form a reader can use. A seminar script, slide sequence or generated draft can express such an account. Select the narrative and its intended use within a larger source collection or teaching programme; evaluating that collection’s authority, the generation method or the whole programme is a different question. Missing source relations or poor quality are possible findings, not reasons to exclude a narrative candidate before evaluating it.

First useful move: identify the narrative version, reader and intended use, then select the relations and characteristics on which that use depends. Compare an adequate account with a deficient one. Distinguish those from a question about a different object, such as the generation method rather than its resulting narrative.

What goes wrong if missed: readability, elegance, engagement, expert approval, or generated fluency substitutes for epiplexity, source-return discipline, and bounded use.

What this buys: a repeatable evaluation that can feed E.23 improvement without confusing characteristics, measurements, eval programs, evidence, assurance, or gates.

Quality target: this pattern evaluates quality for one declared use under source-structure selection fit, NarrativeRenderingEpiplexity, ordering recoverability, temporal-posture and role fit, source-return readiness, bounded engagement, and owner-routed evidence, assurance, ethics, publication, and work claims.

NSTD.6:2 - Problem

Narrative rendering quality for declared use is not one property. A narrative can be fluent but structurally false, engaging but ethically unsafe, technically accurate but unusable for learners, or source-faithful but impossible to follow. A useful evaluation needs object-kind fit, characteristic slots, value meanings, evidence basis, missingness rules, floor, exceptional meaning, result-row shape, and repair actions.

NSTD.6:3 - Forces

NSTD.6:4 - Solution

Construct and use one narrative rendering quality evaluation characteristic space for one declared use. Reuse a sufficient current specification and result when their object, question and conditions still match.

For the structural-amount question, NarrativeRenderingEpiplexity specializes C.2.8 U.ExtractableStructuralInformation: the structure this reader or observer can extract. Its three bearers are the receiving narrative episteme, the publication form expressing it, and the specified reader or observer. The selected source structure, correctness criterion, usable prior knowledge, operations, help, access and budget qualify this tuple. An unambiguous rendering reference can identify the account and expression. Use the existing epiplexity basis, scale and evidence fields below to retain the qualifications needed for the comparison; they do not create another bearer or record kind.

Begin with named additional and missing relations when a qualitative comparison answers the question. If a number is useful, count correctly recovered selected semantic relations at a declared grain, optionally as n/N for a fixed nonempty selected set. Each unit preserves participants, predicate, polarity, modality and action-changing conditions. Repeated words, headings or arrows do not add units. Retain which relations were recovered: equal counts can differ in suitability for the next use. A different ordinal or nonnegative structural weighting needs its own explicit structural meaning; a utility weighting is another score. A formal bit estimate needs the mapping in C.2.8, not merely the name epiplexity.

State whether the basis is a conditional design walkthrough, an actual reading or a mapped formal estimate. One response establishes the recovery observed in that trial and can ground an extractability estimate; it does not establish a maximum, impossibility after failure, or reliability across a reader class. An exact computational amount requires the corresponding method and model. Keep the first response, actual help and source returns. Familiar structure can be recovered from an expression; an expert’s independent correction of false or absent instruction earns that expression no credit. Recovery of what a source asserts and recovery of warranted subject structure use different correctness questions.

Compare forms with claims and relevant conditions held constant. Added or changed substantive claims can identify another episteme through C.2.1; changed access, preparation or help qualifies another comparison. A transfer or perturbation result belongs to its tested condition or robustness question and adds nothing to the original amount by itself. Compare effort and usefulness separately through C.11.CRC and C.11; keeping the sufficient rendering is an available result.

NarrativeRenderingQualityEvaluationCharacteristicSpace@Context:
  evaluatedObjectKindRef: NarrativeRenderingVersion@Context
  declaredUseScope:
  objectKindFitRule:
  discriminatingCaseSet:
  characteristicSlotSet:
  epiplexityBasisRule:
  scaleBindingSet:
  valueMeaningSet:
  evidenceBasisRule:
  missingnessAndLoweringRule:
  resultRowShape:
  floorAndExceptionalMeaning:
  protectedTradeoffSet:
  stopOrReopenCondition:
  neighboringGoverningPatternRefs:

Object-kind fit:

Default value meanings for the other ordinal content-quality characteristics follow. NarrativeRenderingEpiplexity uses its own structural scale above; the following 0–5 labels do not assign its amount, missingness or use threshold:

For reliance-bearing or teaching use, the default for structural recovery is that every relation declared necessary for that use must be recoverable under the specified support and evidence conditions. State those necessary relations at the grain the decision needs; a short account of the required distinction and stop can suffice. There is no generic structural-amount floor of 4. Keep the amount and satisfaction of the use condition separate: unrelated extra relations cannot compensate for a missing necessary relation.

The other load-bearing quality characteristics retain default floor 4 on their own scales, including OrderingRecoverability and SourceReturnReadiness. When the selected source structure is a constraint-governed unfolding structure, DemonstrativeSliceRecoverability is load-bearing and may not be below 4. A local low-risk orientation use may set those quality floors to 3 only if non-admissible downstream use is explicit. Neither numeral translates into a structural amount. A narrower use fixes its structure selection before evaluation; removing a difficult relation after observing the answer changes the basis rather than improving the original amount.

Result-row shape:

NarrativeRenderingQualityResultRow@Context:
  narrativeRenderingVersionRef:
  declaredUseScopeRef:
  characteristicId:
  characteristicName:
  scaleRef:
  value:
  evidenceBasisRefs:
  missingnessClass:
  loweringReason:
  repairAction:
  directOwnerRefs:
  reopenCondition:

Package the rows before feeding an improvement loop:

NarrativeRenderingQualityEvaluationResult@Context:
  evaluatedNarrativeRenderingVersionRef:
  declaredUseScopeRef:
  evaluationCharacteristicSpaceRef: NarrativeRenderingQualityEvaluationCharacteristicSpace@Context
  evaluationPurpose:
  evidenceBasisRefs:
  resultRows:
  protectedTradeoffSet:
  belowFloorRows:
  candidateImprovementProposalRows?:
  nonUseBoundary:
  stopOrReopenCondition:

When repeated improvement is wanted, open E.22 first if the quality question is not already framed. Then use E.23 with NSTD.6 as the object-under-improvement evaluation. This DPF does not mint a local loop kind.

NarrativeRenderingImprovementLoopInput@Context:
  e22QuestionFrameRef?:
  objectUnderImprovementRef: NarrativeRenderingVersion@Context
  objectVersionBeforeRef:
  objectUnderImprovementEvaluationRef: NSTD.6
  improvementAim:
  protectedTradeoffSet:
  costAndRiskAccount:
  allowedChangeSlice:
    narrativeRenderingVersion | NSTD.1-intake | NSTD.2-ordering |
    NSTD.3-event-model | NSTD.4-viewpoint | NSTD.5-engagement |
    NSTD.7-generation-and-inspection | NSTD.8-learning-route |
    evaluationCharacteristicSpace
  returnedFindingOrProposalRows:
  expectedReEvaluationResultForm: NarrativeRenderingQualityEvaluationResult@Context
  neighboringGoverningPatternRefs:
  stopContinueSwitchOrHoldCondition:

E.23 may claim improvement only after the changed object version is re-evaluated through NSTD.6 or through a declared stronger evaluation. If the loop changes the source pack, source-currentness, generated-carrier admission, learning publication carrier, publication face, ethics claim, evidence claim, assurance claim, or evaluation characteristic space, the loop must name the neighboring governing pattern and either keep it as the allowed change slice or open separate work. Style edits, prompt retries, or additional drama are admissible loop operations only when their expected movement under NSTD.6 is stated and protected trade-offs are checked. B.4 is relevant only when the narrative episteme or learning route is claimed to evolve across use and renewed operation; G.11 handles refresh when source currentness, reader telemetry, teaching-test evidence, generated-narrative practice, or FPF edition changes.

Before assigning values, recover the source and use basis for the question being answered.

For current-product evaluation, name the selected subject structures, intended reader work and necessary prerequisites, then trace what the present rendering preserves, hides or loses. Use the task and qualified subject sources to establish the needed structure; the rendering can help refine that basis but cannot excuse its own omissions. Recover its ordering rationale, source-return condition, and the event or mechanism, viewpoint, agency, engagement and learning-route relations that matter to this use through NSTD.1 through NSTD.5 and NSTD.8. Point to actual source and rendering passages. If this account is reconstructed now, identify it as an evaluation of the current product, with its uncertainties, not a record of what the author did. Missing manufacturing notes do not by themselves lower current-product values; missing subject basis, inaccessible return or unsupported relation still does.

For a promised transformation, compare the named source and result versions against the promised preservation or change. This can establish their present fidelity relation without establishing the actual sequence of manufacturing actions. For a claim about actual construction or repair, obtain records or other admissible evidence of those actions and distinguish them from a plausible reconstruction. Trace the applicable NSTD.1 selection, NSTD.2 ordering, NSTD.3 mechanism, NSTD.4 viewpoint, NSTD.5 engagement and NSTD.8 learning-route decisions. If their history cannot be established, leave that historical or assurance claim unsupported and return it to its owner. A good product does not supply the missing history.

A candidate can be evaluated before acceptance; the result can inform acceptance, repair or rejection. Additional requirements follow from the proposed use and claims. When generated or discovered content is intended to inform architecture work, C.35 supplies that assessment. Reuse as evidence follows A.10; assurance, publication authority and actual human learning-effect claims require their own grounds. None of those conclusions follows merely from a favourable narrative-quality result.

Use this evaluation sequence:

1. Object-kind fit: which narrative account and version is being assessed? Identify it in its text, slide sequence or other form. A question about the source collection, generator or whole teaching programme returns to the evaluation for that different object.
2. Evidence fit for the question: for current-product quality, recover selected source structures, ordering, source return and relevant relations from the subject/use basis and present rendering; for promised fidelity, compare the named source and result; for actual construction or repair, recover evidence of the claimed actions. Keep these conclusions distinct.
3. Declared-use fit: is the reader or listener use narrow enough to evaluate, and are non-admissible downstream uses stated?
4. Load-bearing characteristics: assign values only to the characteristics needed for the declared use, but include every characteristic whose failure would make the use unsafe or useless.
5. Low-value repair: for every value below floor, name the smallest repair route before proposing style, drama, or generation retries.
6. Re-evaluation route: if any repair changes the object version or selected source basis, plan a new NSTD.6 evaluation before claiming improvement.

Missingness and lowering rules:

A defined trial that recovers none of the selected relations supports observed recovery 0 on that scale, not impossible extraction. An empty denominator supports no fraction. Wrong-kind handling precedes amount assignment. A qualified design estimate can remain available without an actual trial; obtain a trial when its possible outcomes can change the receiving decision.

Default narrative rendering quality characteristics:

LearningRouteReconstructionFit is an additional characteristic for teaching uses. It uses the ordinal quality meanings above and the adjacent calibration below; it is not a structural-amount scale.

NSTD.6:5 - Archetypal Grounding

Value bank: full result rows

Use this bank when a narrative rendering “sounds good” and therefore tempts the worker to skip evaluation. Each row illustrates evaluation of a narrative version for one declared use; prior acceptance is unnecessary. The same text may receive different values for a different use.

An existing product and an unknown manufacturing history

An editor receives a handout narrative with a declared onboarding use, qualified subject basis and source-return links, but no manufacturing or repair journal. The editor reconstructs the needed source relations from the intended work and subject sources, follows their use through the present explanation and exercises, and tests a meaningful changed case. A missing return or a false relation is a current-product defect. No journal is needed to identify or repair it, and a strong present boundary test can support a high product value without inventing an earlier repair.

If the same editor is instead asked whether the handout faithfully transforms a named seminar, the seminar and handout must be compared. If asked whether the author performed a prescribed review or made a particular correction, evidence of that action is also needed. Current fidelity or a successful reader attempt cannot answer that historical question. A generated candidate receives the same current-product evaluation; its acceptance or production-history claims remain separate questions.

Before and after evaluation repair

Before evaluation statement:

The narrative is strong because readers liked it and remembered the main point.

Failure: engagement and memory are treated as total quality. Source recovery, relation strength, owner routing, and use boundary are absent.

After evaluation statement:

For the declared onboarding use, the illustrative recovery result is that learners can reconstruct the pattern-use route; a numerical amount needs a specified relation set, conditions and evidence. Source-return readiness receives value 3 because two slogans lack pattern-body refs, and engagement boundary receives value 4 because the failure story is marked as archetypal. The first repair is to add source-return refs for the slogans before changing style.

Now E.23 has a concrete changed slice: add two source-return refs and re-evaluate source-return readiness. Revisit the amount only if the changed access or explanation can change structural recovery. An earlier amount label is reusable only through its recoverable selection, scale, conditions and evidence; an unsupported 4 does not become 4/5.

Adjacent-value calibration

For structural-amount calibration, fix five relations in a narrative about reusing a review after a display change: reuse requires the same claims, the same question, and the same qualification window; an unavailable required argument stops the new check and returns it to the source; that stop preserves unrelated review results. This adapts ME.22’s content example. The condition, stop and source return form one compound unit at this declared grain. Hold the reader, case facts, material, operations, help and budget constant. The counts below illustrate nested constructed recovery sets, not three observed trials or a universal five-point rubric. Another selected set uses its own count or denominator without conversion to five labels.

The other characteristics use their own ordinal quality meanings:

Evaluation-to-improvement repair input without process theatre

NSTD.6 does not create a big improvement program. It creates result rows. A repeated improvement loop needs only:

NarrativeRenderingImprovementLoopInput@Context:
  objectVersionRef: narrative rendering version, including a candidate
  evaluationResultRefs: selected `NSTD.6` rows
  improvementAim: raise one declared value without lowering protected trade-offs
  allowedChangedSlice: wording, source-return link, ordering marker, viewpoint repair, engagement device, generated source plan, or learning task
  protectedTradeoffSet: source fidelity, owner routing, engagement, cost, reader burden
  expectedReEvaluationForm: rerun the affected `NSTD.6` rows and any neighbor owner checks

If the change is “regenerate until better”, the object version and changed slice are gone. If the change is “add a source-return link and an analogy-stop task”, E.23 can operate and NSTD.6 can re-evaluate.

FPF owner teaching

C.2.8 governs structural amount for architecture and non-architecture narratives alike. NSTD.6 applies it to the narrative and the selected observer/use conditions. C.33 retains the separate architecture-description adequacy question. Source-selection fit, source-return readiness, effort and usefulness remain separately judged; a story need not be an architecture description for its structural amount to be compared.

Pass case: an FPF seminar handout narrates how a practitioner moves from problem frame to forces to solution and checks. It names FPF pattern source sections, ordering rule, learner reconstruction task, and source-return points. Its necessary source relations are recoverable under the declared support; the other selected quality characteristics can reach 4 or 5 on their own scales. The handout is not evidence that FPF is correct, and those quality labels do not assign its structural amount.

Fail despite fluency: a polished architecture story says one chosen architecture “won” because it felt coherent, hides rejected candidates, omits architectural characteristics, and gives no source-return path. The needed architecture relations are missing, so the structural-use condition fails; assign an amount only when its selection and conditions are established. OrderingRecoverability and SourceReturnReadiness can fall below their own floors even if engagement is high.

Different object: an evaluator is asked whether a text generator is reliable across unseen topics. One narrative’s NSTD.6 result does not answer that question; the generator and its proposed operating range need their own evaluation. By contrast, a fluent generated story with an unknown source is still a narrative candidate: visible defects can be diagnosed, while source fidelity and structural amount remain unassigned where their required basis is unavailable.

NSTD.6:6 - Bias-Annotation

This pattern blocks proxy-as-quality drift: readability, fluency, liking, engagement, expert approval, or generated-text benchmark value replaces object-kind fit and declared-use rendering quality. It also blocks the opposite drift where a test program is treated as the characteristic itself. Repair by selecting the evaluated object kind, scales, value meanings, evidence basis, missingness and lowering rules, floor, result rows, and repair actions. Scope: DPF-local for evaluating narrative rendering versions; it does not govern evidence, assurance, gate, decision, or publication authority.

NSTD.6:7 - Conformance Checklist

NSTD.6:8 - Common Anti-Patterns and How to Avoid Them

NSTD.6:9 - Consequences

The benefit is a usable improvement target: E.23 can improve narrative versions because values, floors, evidence, and repairs are declared. The cost is heavier evaluation before reliance-bearing use.

NSTD.6:10 - Rationale

A.19.ECS says improvement cannot be better than its evaluation. NSTD.6 specializes that lesson for narrative renderings: first recover object kind and use, then choose characteristics that discriminate narrative rendering quality for that declared use.

NSTD.6:11 - SoTA-Echoing

FPF A.19.ECS and C.16 supply the evaluation and measurement distinctions. C.2.8 supplies reader-relative structural amount and its qualified formal epiplexity specialization; C.33 concerns architecture-description adequacy. Amount, extraction effort, learning and task usefulness can change independently. Language-state, coarsening and explanation patterns support the particular repairs named in the scale; they are not extra tests triggered by every stylistic change.

Castricato et al.’s model-theoretic account distinguishes narrator-to-reader information and evolving story models. Mengelkamp et al.’s reading-goal experiments found condition-dependent comprehension effects and persistent overestimation. Georgiou et al.’s narrative-memory study separates recall, recognition and reconstruction of order. These results support asking what a reader actually recovered, separately from fluency, confidence or enjoyment; they do not validate the DPF’s ordinal scale.

The competing shortcut is to rate a narrative by fluency or engagement alone. A fluent account that turns temporal precedence into cause demonstrates why that shortcut fails for a causal explanation. This pattern therefore evaluates the selected relations and the intended use separately from engagement, accepting the cost of a focused source comparison or reader task when it can change the judgement. For an expressive use without that causal claim, do not invent the same obligation.

Ma et al.’s story-generation survey identifies continuing coherence, consistency, diversity, control and evaluation problems. Rahman et al.’s G-KMS adds structured, semantic and engine-execution checks for its RPG setting. Select these checks for the claim being assessed: an observable continuity defect can be judged from the product, while a claim about reader experience needs an appropriate reader basis. Neither source makes a human study a prerequisite for inspecting an ordinary draft.

NSTD.6:12 - Relations

Uses A.19.ECS, A.17, A.18, C.2.8, C.16, C.16.Q, C.2.LS, A.16.1, A.16.2, A.6.3.CSC, E.17.EFP, E.10, A.6.P, C.33, E.22, E.23, B.4, A.10, B.3, E.17, C.35, and G.11. C.2.8 defines structural amount for all admitted source domains; C.33 is used when architecture-description adequacy is current. E.23 consumes NSTD.6 result rows only after object version, allowed change slice, protected trade-offs, cost and risk, and re-evaluation form are explicit. B.4 is used only for an evolution claim over a narrative episteme, learning route, or other holon under repeated use; G.11 handles currentness and refresh. Reopen when evaluated object kind, declared use, source pack, characteristic set, language-state profile, cue or backoff status, coarsening or explanation relation, precision-restoration result, epiplexity basis, value meanings, floor, evidence basis, allowed improvement slice, or low-value repair route changes. Support-map entry: open Architecture and Narrative Work Bridge when NarrativeRenderingEpiplexity is architecture-relevant or tied to C.33; open Semiotic And Language-Precision Bridge for language-state, coarsening, explanation, relation, or quality-word repairs; open Source Use And Refresh Map when evidence basis or source-currentness supports a value; open DPF Precision Restoration And Owner Map when a characteristic name risks becoming a new ontology.

NSTD.6:End