Demonstration and Evaluation

A single Australian detached dwelling, followed across its building life-cycle, is the venue at which the artefact suite is exercised at full life-cycle scope. The demonstration traces one site from a minimum-viable baseline through six transformation events to a fork at the empty-nest threshold, and reports the resulting evidence against the four critical properties the thesis pre-registers as its falsifiable commitments. Those properties — interoperability (Proposition 1), transportability (Proposition 2), manipulability (Proposition 3), and transformability (Proposition 4) — are stated in Section 1.4 as the substantive content of the architectural claim, and the evaluation strategy declared in Chapter 4, Section 4.4 binds each property to specific evaluation questions, measures, and evidence objects. The demonstration is the venue at which that binding is discharged. The suite under exercise comprises the standardisation schema of Chapter 5; the Governed Kernel Architecture, the governed instance library of Chapter 6; the notation of Chapter 7; the empirical substrate of Chapter 8; and the generator, the generation-and-documentation pipeline of Chapter 9.

The site begins at a minimum-viable baseline — state S0: one bedroom, one bathroom, a combined living-and-kitchen zone, approximately fifty-eight square metres on a fifteen-by-thirty-metre suburban lot — and progresses through four trunk transformation events: an occupancy commitment on the same envelope, an expansion for family formation, a further expansion for family growth with an upgraded bathroom, and the re-purposing of a bedroom into a home office at the empty-nest threshold. At state S4 the trajectory forks. One branch applies a typical aging-in-place adaptation to the empty-nest envelope; the other applies a Specialist Disability Accommodation overlay to the primary dwelling and adds a secondary attached dwelling for an adult occupant under the SDA Robust design category. The fork is the chapter’s principal demonstration of Proposition 4 transformability: the same architectural commitments at S4 support two divergent regulatory-compliance pathways from a common substrate, neither of which requires the architecture to be re-authored and neither of which violates the interface contracts declared in earlier chapters. A single substrate produces two regulatory pathways because the substrate is not asked to do the divergent work — the absorption is borne at the governed instance library, and the governed kernel does not move. The trunk and the fork together span seven dwelling states (S0, S1, S2, S3, S4, S5a, S5b) and six transformation events (E0→1, E1→2, E2→3, E3→4, E4→5a, E4→5b).

The trajectory exercises the four properties at distinct scopes. Proposition 1 interoperability, Proposition 2 transportability, and Proposition 3 manipulability are exercised at every state: each of the seven states is serialised, parsed, and re-emitted under the notation’s round-trip discipline, and each state’s documentation packet is checked for interpretive trace completeness, interface-bounded verification scope, and replay determinism. Proposition 4 transformability is exercised across the six events and the fork at S4, where the architecture must absorb regulatory and occupant change without disturbing the shared modular core. The secondary measure, Proposition 5 burden, is exercised cumulatively: the per-event practitioner-burden proxies are tracked from S0 through both fork branches, and the resulting growth profile is reported in Section 10.17.

The contract these results answer to is the one pre-registered in Chapter 4, Section 4.4, recapped here so the findings are judged against criteria fixed before the results were known rather than against criteria that drift to fit them. The contract is requirement-linked and proposition-linked: each evaluation question traces to at least one environment requirement, at least one proposition, and at least one pre-declared measure, under the alignment EQ-01↔︎P1 (interpretive convergence under semantic continuity); EQ-02↔︎P2 (interface-bounded verification under bounded checking); EQ-03↔︎P3 (replay reliability under round-trip); EQ-04↔︎P4 (governed variation under shared rules); and EQ-05↔︎P5 (integrated utility under diachronic burden). Four summative measures discharge that contract in this chapter. EM-09-01 reports standards-interpretability trace completeness across the seven states (Section 10.4). EM-09-02 — modular-fit and bounded-check performance — reports interface conformance and bounded verification under each event (Section 10.10); EM-09-02 is the canonical pre-registered indicator for Proposition 2, and any concrete sub-metric coined within this chapter operationalises it rather than displacing it as the canonical name. EM-09-03 reports cumulative workflow burden across the trajectory’s events and the fork (Section 10.17). EM-09-04 reports exception-governance quality for the Rule-4 variants generated across the trajectory (Section 10.10, with the fork’s exception-budget table in Section 10.3). Each measure outcome is classified under the four-result adjudication scheme declared in Section 4.5 — pass, fail, indeterminate, or boundary-limited acceptance — and every claim promoted to the discussion carries at least one linked evidence object; the per-measure adjudications aggregate into the per-property verdicts tabulated in the conclusion (Section 10.23).

The trajectory produces two classes of evidence object, and the asymmetry between them is surfaced here so that downstream readers can distinguish the two without re-deriving the partition from the source files. Primary objects are those declared at the Section 4.4 traceability-matrix level; inline-synthesised objects are identifiers coined within the results sections’ own prose to record reasoning steps the matrix abstracts over. Table 10.1 records the register.

Table 10.1: Evidence-object register — primary (matrix-declared) and inline-synthesised identifiers exercised across the trajectory.

The seven primary objects carry the canonical register held in the requirements–design–evaluation traceability matrix appendix; the inline-synthesised identifiers record per-instance reasoning the matrix abstracts over, and the Proposition 3 objects are inherited from the round-trip evidence established in Chapter 7, Section 7.20 rather than re-run here.

Four scope-limits bound what the chapter’s verdicts can claim, and they are stated at the outset because they are the conditions under which the verdicts hold rather than concessions appended after the fact. First, the demonstration follows a single dwelling site rather than a sample of dwellings: it establishes feasibility and internal coherence, not population-scale generalisability, and cross-corpus validation is declared future work. Second, the trajectory’s authoring, rating, and interpretation are candidate-driven; external multi-rater evaluation is future work, and the declared evaluation ceiling is therefore efficacy under constructed demonstration rather than effectiveness under naturalistic field deployment, consistent with the Technical-Risk-and-Efficacy classification of Section 4.4. Third, the seven dwelling states are synthetic but grounded: their key dimensions are constrained to fall within the empirical distribution of the sealed CP-D5 corpus, which is a complete census rather than a probability sample, so the grounding is read descriptively and no inferential generalisation is drawn from it. Fourth, the SDA-Robust overlay applied at the fork is scope-limited to the Kitchen and Exterior modules per the governed-kernel declaration of Chapter 6, Section 6.4; the chapter does not claim full SDA-Robust coverage beyond that bound.

The chapter proceeds in four movements. Section 10.2 specifies the trajectory site — the baseline S0 dwelling and the canonical trunk S0→S4. Section 10.3 specifies the fork at S4 and tabulates the divergent pathways that distinguish its two terminal states. Section 10.4, Section 10.10, and Section 10.17 report results across interpretability, modular fit, and workflow efficiency, the last closing with the synthesis of the evidence, its disconfirming outcomes, and its bounds. Section 10.23 assigns a proposition-status verdict to each of Proposition 1 through Proposition 5 and routes the bounded findings into Chapter 11 for the thesis’s contribution synthesis.

The empirical site for the demonstration is a single Australian detached dwelling followed across its building life-cycle. The single-dwelling longitudinal design exercises Proposition 4 (transformability) at full life-cycle scope rather than at the more familiar single design instant: the substrate is held constant, and each transformation event is observed as a differential outcome on a common physical and modular envelope rather than as a confounded comparison across distinct sites.

The choice of one dwelling followed across decades, over a multi-case cross-section, is deliberate. A multi-case study evaluating several distinct dwellings at single design instants would conflate inter-site variability with the architecture’s response to change: the apparent absorption of a transformation event would be confounded with the differing baselines of the cases. A single dwelling followed through its life-cycle isolates the transformation response from substrate variability, and at the empty-nest decision point it presents a fork that exercises the architecture’s support for divergent regulatory-compliance pathways from a common substrate — a property the architecture must possess if the deontic separation between the Fully Accessible and Specialist Disability Accommodation categories established in earlier chapters¹ is to absorb realistic regulatory transitions without core revision.

The trajectory comprises seven dwelling states and six transformation events. The trunk runs from the baseline (S0) through couple-young (S1), small-family (S2), and teen-family (S3) to empty-nest-with-home-office (S4); at S4 it forks into a typical aging-in-place state (S5a) and a Specialist Disability Accommodation state with an attached secondary dwelling (S5b). This section addresses the baseline and the canonical trunk; the fork is the subject of Section 10.3.

Minimum-viable substrate (S0)

The baseline state is intentionally small. The dwelling occupies a fifty-eight-square-metre rectangular footprint at the front of the lot, oriented north–south, and comprises a twelve-square-metre bedroom (BR1, the first BED-class instance under the canonical nine-type taxonomy of Chapter 6, Section 6.2), a five-square-metre bathroom (BA1, the first SAN-class instance), an open-plan twenty-five-square-metre living-plus-kitchen zone, a six-square-metre central circulation hub, a three-square-metre service core, and a step-free entry approach. All Fully Accessible category dimensional and performance requirements are met at this baseline, including the fifteen-hundred-and-forty-millimetre circulation diameter at the bedroom centre, the nine-hundred-millimetre clear doorway widths throughout, and the fifteen-hundred-and-fifty-millimetre kitchen bench clearance.

S0’s position within the empirical record must be surfaced explicitly, and it is read descriptively rather than as a statistical test, because the reference corpus is a complete census rather than a probability sample. Against the sealed CP-D5 corpus of seven hundred and forty-five Australian residential floor plans,² S0 sits below the general-residential one-standard-deviation band on gross floor area, bedroom count, and bathroom count. The cohort records a median internal floor area of ninety-seven point eight-six square metres with a standard deviation of thirty-seven point two-four square metres — a one-standard-deviation band of roughly sixty-one to one hundred and thirty-five square metres — and S0 at fifty-eight square metres lies just below it, at about the fifth-to-tenth percentile of the observed distribution. Its single bedroom sits in the extreme low tail (one-bedroom dwellings are zero point seven-five per cent of the cohort), while its single bathroom sits in a more populated low tail (about twenty per cent of the cohort). These figures describe where S0 falls within the census’s observed spread; no inference beyond the corpus is drawn from them.

The triple deviation is a deliberate methodological choice rather than an ecological-grounding failure. Anchoring the trajectory at the empirical low tail maximises the demonstrative range of the architecture’s transformation response: had S0 begun at the cohort median — four bedrooms, two bathrooms, ninety-eight square metres — the subsequent life-stage events would already be present in the baseline, and the cumulative transformation would be impossible to observe. Fifty-eight-square-metre one-bedroom detached dwellings are an attested Australian category rather than a synthetic outlier,³ and the minimum-viable framing conforms to the National Construction Code Volume 2 minima for Class 1a residential dwellings.⁴ Two baseline dimensions are not corpus-testable and are anchored to public benchmarks instead: lot area, which the CP-D5 schema does not capture as structured metadata, is set against established-suburban lot benchmarks,⁵ and internal-circulation footprint, recorded for only a small fraction of corpus spaces, is anchored to the Code’s minimum corridor and accessway widths.⁶

Canonical trunk (S0 to S4)

Four transformation events carry the dwelling through the canonical Australian residential life-cycle, and each is recorded as a specification consumable by the procedural documentation pipeline.⁷ The first event, E0→1 (occupancy commitment), transitions the dwelling from a single-occupant baseline to a couple-young state on the same fifty-eight-square-metre envelope: the module composition and interface obligations are unchanged, only the occupancy declarations move, and the predicate count holds at eleven. The event is recorded explicitly so that occupancy-only changes remain legible in the trajectory’s burden accounting, where a zero-spatial-delta re-run establishes a per-event burden floor against which later events are measured.

The second event, E1→2 (family formation), is the first spatial expansion. A second bedroom (BR2, eleven square metres) is added at the south-east; the living zone is extended by six square metres; and the circulation hub grows from six to nine square metres to provide approach to BR2. Gross floor area grows from fifty-eight to ninety-eight square metres, and the predicate count from eleven to twelve. BR2 at eleven square metres engages the architecture’s secondary-bedroom interpretation: the Fully Accessible typology asserts a twelve-square-metre bedroom minimum, but this is read under Chapter 6, Section 6.4’s Rule 4 as applying to the participant-exclusive primary bedroom, with secondary bedrooms governed at the same Fully Accessible category level under a relaxed area floor consistent with multi-occupancy practice. The interpretation is surfaced here because it is the trajectory’s first concrete variation-governance event, and it recurs at the next event.

The third event, E2→3 (family growth), is the trunk’s largest transformation. A third and fourth bedroom are added (BR3 at eleven square metres, BR4 at ten square metres); the original bathroom becomes a master ensuite and a new family bathroom (BA2) is added; and the living zone is separated from a new dining zone by a partial-height screen. Gross floor area grows from ninety-eight to one hundred and sixty-five square metres, bedroom count from two to four, bathroom count from one to two, and predicate count from twelve to thirteen. BR4 at ten square metres is the trajectory’s smallest bedroom and the case against which the secondary-bedroom interpretation is most exposed. The fourth event, E3→4 (empty-nest re-purposing), is structurally lightweight: the envelope is preserved exactly at one hundred and sixty-five square metres, and the change is functional. BR2 is re-classed as a home office by removing bedroom furniture and installing a desk-and-shelving fit-out, task lighting, a six-outlet power cluster, and a data outlet. The office is treated as a Living-variant under Chapter 6, Section 6.4’s Rule 4 — its office-specific qualities are encoded at the variant level and governed at the governed instance library without modifying the governed-kernel nine-type taxonomy — and the predicate count grows from thirteen to fourteen.

Module-class variants introduced across the trunk

Four module-class variants surface across the trunk and the fork: the home office (OFC, a Living-variant), the dining zone (DIN, a Living-variant), the helper or carer room (HLP, a Bedroom-variant introduced at the fork), and the connecting link (LNK, an Exterior-variant introduced at the fork). All four are governed under Chapter 6, Section 6.4’s Rule 4 on the canonical nine-type space taxonomy.⁸ The trajectory therefore exercises Rule 4 at four distinct points without revising the governed-kernel taxonomy itself, which is the form the architecture’s claim takes at the level of modularity: life-cycle functional variation is absorbed at the governed instance library.

Corpus position across the trunk

The trunk’s progression traverses the census distribution in step with the Australian family-life-stage cohort structure, and the per-state positions are read descriptively against the relevant cohorts. S0 and S1 share the low-tail position by design; S2 moves to the two-bedroom small-detached cohort at its median; S3 lands in the modal four-bedroom-with-two-bathroom cohort at its median; and S4 preserves the S3 envelope within a fifth of a standard deviation of the four-bedroom-with-office cohort median. The progression covers more than ninety-five per cent of the corpus’s bedroom-count mass, from the one-bedroom low tail through the three- and four-bedroom cohorts that together hold most of the distribution. Table 10.2 records the per-state position; the verdict column reports where each state falls within the census’s observed spread, with the deliberate low-tail anchoring at S0 surfaced as a methodological choice rather than glossed.

Table 10.2: Per-state corpus position against the general-residential cohort distribution (descriptive; the corpus is a census, not a sample).

Structured-output evidence

The trunk’s states and events produce five documentation packets from the generator (DWL-FA-01 through DWL-FA-05, for S0 through S4), preserved in the Chapter 10 evidence appendix bundle. Each packet records the module specification, the PlaniSyn predicate list extracted from the state’s provenance field, the interface-obligation hooks detected in the body, and the raw provenance triples preserved verbatim as the auditable link to the SDA Design Standard. These packets are the structured-output evidence of the trajectory’s instantiation, and their fields are read into the per-event measure tables of Section 10.4 and Section 10.10. The trunk concludes at S4 — the fork junction — whose one-hundred-and-sixty-five-square-metre envelope, module composition, and interface obligations are inherited by both branches of the fork that Section 10.3 develops.

The trunk concludes at S4 — the empty-nest state at one hundred and sixty-five square metres, four bedrooms (one of them a home office), two bathrooms, separated living and dining zones, and a cross-shaped circulation hub — and at this junction the trajectory branches into two divergent pathways. The fork is the chapter’s central demonstration, and the architectural claim it bears out is a claim about restraint: a single substrate produces two regulatory pathways because the substrate is not asked to do the divergent work. The same architectural commitments at S4 — the nine-type module taxonomy, the governed-kernel composition constraints, and the variant-inheritance rule of Chapter 6, Section 6.4 — support both branches; the divergence is absorbed at the governed instance library, and the governed kernel does not move. The fork exercises Proposition 4 (transformability) under simultaneous regulatory and occupant change, and Chapter 11 reads it as the thesis’s decisive empirical instance.

Path A: typical aging-in-place (E4→5a, S5a)

The left branch applies a set of accessibility-hardware retrofits that preserve the S4 envelope exactly: no walls move, no doors are repositioned, and no module is added in volumetric terms. Both bathrooms receive grab-rails, fold-down shower seats, hand-held showers, and non-slip floor finishes to the relevant standards;⁹ the master ensuite shower is enlarged for attendant access; non-slip flooring is fitted in the circulation hub and kitchen; and the front-door hardware is upgraded to an easy-grasp lever. The most consequential change is functional rather than dimensional: BR3 (BED-class) is re-classed as a helper or carer room (HLP-1) for episodic carer occupancy, a Bedroom-variant under Chapter 6, Section 6.4’s Rule 4 with the participant-exclusive fifteen-hundred-and-forty-millimetre circulation diameter relaxed to a single-bed access pattern appropriate to non-participant-exclusive use.¹⁰ The envelope at S5a is preserved at one hundred and sixty-five square metres; the predicate count drops from fourteen to thirteen; the dwelling hosts a couple in late life with episodic carer support; and the design category remains Fully Accessible. The state is recorded as carrying an aging-in-place overlay rather than as a Specialist Disability Accommodation declaration — a distinction the framework keeps explicit because practitioners frequently conflate accessibility-aware design with SDA-eligible design, and the architecture’s treatment of S5a versus S5b is precise about the regulatory boundary between the two.

Path B: SDA-overlay with attached secondary (E4→5b, S5b)

The right branch applies a full Specialist Disability Accommodation Robust-category overlay to the primary dwelling¹¹ and adds an attached secondary dwelling for an adult occupant under SDA Robust. The transformation is the trajectory’s most extensive: eight new module instances, nine reconfigured modules, nine new bilateral interface obligations, one design-category transition (Fully Accessible to SDA Robust on the primary dwelling), and approximately fifty new Robust quality assertions across the primary and secondary dwellings. The primary dwelling’s covered porch is enclosed as a zero-step entry vestibule with full Robust hardware; both bathrooms receive the Robust accessible-bathroom upgrade (reinforced walls, impact-resistant lining, four grab-rails each, emergency-release locks, walk-in showers); corridors and accessible windows receive impact-resistant lining and laminated glazing; and the existing five-fixture kitchen is retained at its bench clearance. The secondary dwelling (DWL-2) is a forty-five-square-metre self-contained one-bedroom unit on the rear yard, connected to the primary dwelling by a covered, step-free passageway (LNK-1, an Exterior-variant under Rule 4). The combined gross floor area is two hundred and fifteen square metres — one hundred and seventy primary plus forty-five secondary — and the predicate count on the primary grows from fourteen to fifteen, with the secondary contributing an independent set of approximately twenty-five Robust assertions.

Fork comparison against the four properties

The fork is the trajectory’s clearest test of the four critical properties.¹² The two paths discharge each property differently, and the divergence is itself the empirical content the fork delivers.

The two paths share the same S4 substrate and the same nine-type taxonomy, but they discharge the four properties at materially different scales, and the difference is worth stating precisely because the chapter does not let the larger figure inflate the smaller claim. The measured artefact-delta is descriptive: S5b adds eight modules and nine bilateral interface obligations against S5a’s single reclass and zero new interfaces, and its emitted documentation packet is about one and a half times the size of S5a’s in bytes, as the per-event figures of Section 10.17 record. The practitioner-effort difference is larger again — reasoned at roughly three to four times S5a’s under the matched-task baseline of Section 10.17 — but it is an analytic estimate rather than a measured quantity, projected from the documentation labour each pathway would demand under unserialised regulatory text and carrying no practitioner cohort behind it. The two figures are kept distinct throughout: the measured artefact-delta is the chapter’s empirical evidence of burden-proportionality, and the practitioner-day multiple is an analytic projection that Section 10.17 declares as such, with neither presented as a measured warrant for a generalised practitioner-burden claim. Neither path requires modification of the governed-kernel taxonomy or the Chapter 6, Section 6.4 composition contracts; both are absorbed at the governed instance library.

Exception-budget tracking

The trajectory introduces five exceptions to the canonical taxonomy, all governed under Chapter 6, Section 6.4’s Rule 4 and tracked against the exception-budget mechanism of Chapter 4, Section 4.4.¹³ The exceptions are surfaced explicitly here rather than buried in the per-state specifications.

The five-exception count sits within the single-digit budget the architecture declares before a governed-kernel revision is warranted. The trajectory therefore exercises the architecture’s variation governance at meaningful scope without exhausting the budget — a positive outcome under EM-09-04 that Section 10.10 adjudicates.

The SDA-Robust governed-kernel scope-limit

A documented scope-limit applies to S5b’s overlay, and it is acknowledged honestly because it is itself the architecture’s claim under test. Per Chapter 6, Section 6.4, only the Kitchen and Exterior modules carry authored SDA-Robust governed-kernel entries; the other module classes’ Robust-equivalent qualities are inferred via Rule 4 from their Fully Accessible counterparts, substituting Robust-equivalent dimensional values where they differ. The trajectory’s Robust-overlay assertions for the Bedroom, Bathroom, Living, Dining, Service, Entry, and Circulation modules are therefore inferred rather than directly authored. A future expansion of the Chapter 6 baseline-library to add explicit Robust entries for these classes would tighten the assertion provenance, but the inferred-via-Rule-4 path is precisely what the architecture claims — that the Fully-Accessible-to-Robust transition is absorbable at the governed instance library — and S5b is the case-in-point. The depth of the inference, and its most exposed substitutions, are reported under the disconfirming outcomes of Section 10.17.

Multi-dwelling site composition

A second novel test at S5b is the architecture’s interface-obligation discipline under multi-dwelling composition. The connecting link introduces a shared service envelope — the primary dwelling’s electrical, water, and drainage services extend to the secondary dwelling along the link — which requires the service-core module to declare cross-dwelling boundaries that the Section 6.3 boundary contracts must accommodate. The egress provisions are equally novel: the Robust category requires at least two egress paths per dwelling, and the secondary dwelling’s paths run through the link to the primary circulation hub and directly from its living zone to the rear yard. The framework’s handling of these cross-dwelling egress declarations — bilateral on each interior boundary, with the egress-path quality propagated through the link — is a direct test of the architecture’s interface discipline at the multi-dwelling scope, and Section 10.10 reads it into the bounded-check evidence.

Corpus position across the fork

Both fork states are read descriptively against the relevant CP-D5 cohorts. S5a sits at the cohort median for aging-in-place four-bedroom Fully Accessible dwellings. S5b requires the SDA-Robust-with-attached-secondary cohort, which is small — four records in the census — but exemplary: the audit shortlist identifies a two-hundred-square-metre, four-bedroom, five-bathroom register-listed Robust house with on-site overnight assistance as the pattern S5b is modelled on,¹⁴ and S5b at two hundred and fifteen square metres falls within that cohort’s observed range. As with the trunk, these are descriptions of where the fork states sit within the census’s observed spread, not inferential generalisations from it.

Structured-output evidence

The fork’s two terminal states each emit the generator’s documentation packet. The S5a packet (DWL-FA-06) preserves the aging-in-place overlay’s predicate set and the HLP reclass with its relaxed circulation assertion; the S5b packet (DWL-RB-01) preserves the Robust overlay’s full predicate set across approximately fifty assertion lines spanning the SDA Design Standard’s organisational domains. The two packets, with the prototype run-log, constitute the structured-output evidence that the fork has been instantiated end-to-end through the procedural pipeline of Chapter 9. The fork is the trajectory’s central demonstration: the architecture absorbs not only spatial expansion and functional re-purposing but also divergent regulatory-compliance overlays from a common substrate, with no S4 module reconfigured to support either branch. Section 10.4 reads the per-state and per-event results from the seven packets into the standards-interpretability measure; Section 10.10 reads them into the modular-fit measure; and Section 10.17 reads them into the workflow-burden measure and synthesises the chapter’s verdicts.

The seven dwelling states of Sections 10.2 and 10.3 produce seven documentation packets when fed through the procedural-generation pipeline of Chapter 9, and each packet is a self-contained representation of a dwelling state’s compliance status against the regulatory surface engaged at that state. This section reports how interpretable those packets are when read against the standards they discharge — whether a reviewer holding the National Construction Code Volume 2¹⁵ and, for the SDA-Robust state, the SDA Design Standard 2019¹⁶ can locate, verify, and audit each clause’s discharge by reading the packet. Interpretability is the subject of evaluation question EQ-01 and is measured here through the standards-interpretability measure (EM-09-01), pre-registered in Chapter 4, Section 4.4.

10.5 Operational definition

A documentation packet is interpretable under a regulatory standard when it satisfies four sub-criteria at once. First, mapping: every regulatory commitment in scope at the packet’s declared state corresponds to at least one packet field that surfaces its adherence status. Second, bi-directionality: a reviewer can traverse the mapping in both directions — from packet field to originating clause, and from clause to surfacing field. Third, soundness: no clause is misattributed to a field whose scope does not discharge it, so a field cannot claim a clause it does not in fact address. Fourth, completeness at declared scope: the mapping covers the full set of clauses applicable at the packet’s declared scope rather than a convenient subset. The four-sub-criterion construction follows the trace-completeness conception developed in Chapter 5’s resolution rate methodology, under which an artefact’s interpretability gain is the reduction in interpretive divergence between matched readers operating on the same regulatory commitment.¹⁷ The four sub-criteria are evaluated for each of the seven trajectory states over a regulatory surface that expands rather than replaces as the trajectory progresses — National Construction Code only across the trunk, with the AS-standard retrofit overlay added at S5a and the SDA Design Standard’s Robust surface added at S5b — which is itself a property the chapter tests.

10.6 Per-state interpretability profile

Table 10.3 reports each state’s predicate-list coverage against the four sub-criteria. The clause counts are taken from the typology-domain lists carried in each state’s provenance frontmatter and rendered into the packet’s PlaniSyn predicate list during emission, per Chapter 7, Section 7.13’s grammar.

The headline figure of one hundred per cent coverage at every state is strong on its face and requires careful framing before it is read. The figure is the proportion of clauses carried in the packet’s PlaniSyn predicate list whose commitment can be traced to a surfacing field — not the proportion of clauses in the entirety of the relevant standards. The packets are scoped: each emits the predicates the source state declares applicable, and the table answers whether those declared predicates are internally mapped to packet fields. For the S5b packet, a reviewer can traverse from each of the fifteen Robust predicates in the predicate list to the corresponding entry in the constraints section and onward to the configuration notes where the clause’s discharge is described; the typology-domain annotation on each predicate supplies the soundness anchor, and the predicate-list’s completeness against the declared scope supplies the completeness anchor. The four sub-criteria are therefore satisfied at the predicate level for the seven states at the scope each state declares — the scope the chapter pre-registered as the test object — and the coverage figure makes no claim beyond that scope.

The predicate progression along the trunk is legible in a way that itself supports interpretability, because the count grows in step with the dwelling’s spatial complexity rather than arbitrarily. S0 and S1 share an identical eleven-predicate set, since the occupancy event has no spatial delta; S2 adds one predicate as the second bedroom makes the circulation-diameter requirement explicit at the dwelling level rather than implicit at the single-bedroom level; S3 adds the dining-domain entry corresponding to the living–dining separation; and S4 adds the office variant’s task-lighting and power-cluster qualities. A reviewer reading the seven packets in order therefore sees the dwelling’s growing complexity surfaced one predicate at a time, with each new predicate attached to the spatial element that warrants it, so that the trunk’s interpretability is a function of two structural properties of each packet — the predicate-list completeness against the dwelling’s spatial composition, and the interface-obligation field’s completeness against the neighbouring-module contracts of Chapter 6, Section 6.3. The first property is satisfied across the trunk by construction; the second is the subject of the limitation reported at Section 10.7.

Three packet fields collectively encode each state’s regulatory compliance status, and a reviewer reads all three together to audit a state. The design-category field carries one of the four enumerated values and signals which standards apply: the Fully Accessible value invokes the National Construction Code and the typology graph, while the Robust value invokes the SDA Design Standard’s Robust surface in addition to the residential code. Of the seven trajectory packets, six carry the Fully Accessible value and one the Robust value. The module-type field carries the module-class identifier — uniformly the dwelling-aggregate class for this trajectory — and its governed-kernel composition contract governs which child modules the dwelling may contain. The third compliance-bearing field is the predicate list itself, rendered as subject–verb–object–domain tuples extracted from the provenance frontmatter: these tuples are the auditable link between the packet and the standard, and the domain annotation on each routes the predicate to a regulatory subsection, so that a reviewer who wishes to check a single clause’s discharge reads one tuple rather than the whole packet.

10.7 The interface-obligation limitation

The chapter’s first explicitly declared limitation is stated here in full; the sections that follow back-reference this statement rather than restate it. The packet’s interface-obligations field reads zero for all seven states, and the packets’ interface-obligations section carries the message no cross-module interface hooks detected. The interface obligations are nonetheless present and richly developed in each state file’s body prose, bilaterally declared at the level of detail Chapter 6, Section 6.3’s boundary-ownership table specifies; they are simply not surfaced into the structured packet field. The diagnosis is parser-coverage rather than architectural: the minimum-viable PlaniSyn subset implemented in the prototype recognises module-level adjacency hooks expressed in baseline-library provenance notation but does not yet parse the dwelling-aggregate’s prose-level interface section. The underlying state specifications are intact, the interface-contract content is preserved verbatim in the configuration-notes and provenance-trace sections of every packet, and the remediation is a parser extension — mechanical rather than architectural — documented in Chapter 9, Section 9.6 and identified as future work in Chapter 9, Section 9.27. The consequence for the EM-09-01 verdict is that per-state interpretability is bounded above by the predicate-list completeness, currently one hundred per cent of declared predicates mapped, and bounded below by the interface-obligation surfacing rate, currently zero per cent of bilateral declarations lifted into structured form: a reviewer using the packets alone finds every commitment the state file declares in its provenance block but must read the configuration-notes prose to trace the inter-module interface contracts.

10.8 Diagnostic locality and cross-event traceability

EM-09-01’s interpretability claim is paired in the contract with the modular-fit and bounded-check measure (EM-09-02), under which a reviewer’s ability to locate a violation matters as much as the ability to read the mapping. The trajectory’s six transformation events test that locality. The bedroom-area boundary case at S2 — BR2 at eleven square metres, one square metre below the Fully Accessible typology’s nominal minimum — surfaces in the S2 packet at exactly one predicate row, which carries both the assertion and its boundary case in a single line, and is resolved at one paragraph in the state file’s constraints section; the reviewer does not scan beyond that row to locate the diagnostic. The same holds at S3 (BR4 at ten square metres), at S4 (the office variant’s task-lighting and power-cluster qualities), and at the fork, where each overlay’s discharge is decomposed across the SDA-constraint typology domains so that every predicate’s domain annotation routes the reviewer to the correct subsection of the standard. The general result is that a regulatory diagnostic surfaces at the module-and-interface scope and does not require a global re-walk to be located, supporting EVID-P2-CHECK-SCOPE; the qualification, already stated at Section 10.7, is that interface-level diagnostics localise to the state file’s prose rather than to a structured packet field.

Cross-event traceability is reconstructable from the packets’ predecessor and successor frontmatter and the run-log’s ordered emission timestamps: a reviewer can read the seven packets in order and follow the predecessor relation at each step to reconstruct the trajectory’s transformation trace. The trace is sufficient for examiner-grade review at the state-and-event level but not for a full audit at the predicate-delta level, because the pipeline does not yet emit a structured per-event delta — a second declared limitation, mechanical in the same way as the first (a set-difference between consecutive packets’ predicate lists) and scheduled for the same parser extension.

10.9 Verdict against the threshold

EM-09-01 fixes a three-term relationship to forestall metric drift: trace completeness is the clause-level input; the schema-mediated resolution rate is the derived quantity computed against the baseline; and ambiguity-reduction is the framing-level effect. The numerical aggregate resolution rate threshold of at least 0.25 is reported at the Chapter 5 measurement surfaces; this chapter reports the trace-completeness input directionally, against the unserialised-text baseline under which a building surveyor or accessibility consultant must identify activated clauses from drawings by hand. Against that baseline, each of the seven packets presents a structured predicate list with typology-domain annotations, a verbatim constraints section preserving the source provenance, and a configuration-notes section binding the predicates to the dwelling’s spatial composition — a material improvement over reading the standards documents alongside the drawings, because the predicate list compresses the regulatory surface into a tractable enumeration and the domain annotations route each predicate to a known subsection.

The verdict for EM-09-01 against its directional threshold is therefore a qualified pass subject to declared scope constraints — a boundary-limited acceptance under the four-result scheme of Chapter 4, Section 4.5 — supported by EVID-P1-INTERPRETABILITY (surfaced at all seven packets) and EVID-P1-SDA-TRACE (surfaced at the S5b Robust-category packet). The two scope constraints are those already declared: interface-obligation surfacing at the structured-field level is currently zero per cent, a parser-coverage gap whose content is carried at the prose level; and the cross-event predicate delta is reconstructable by hand but not structurally emitted. Both are documented as scoped extensions in Chapter 9, Section 9.6 and carried into Chapter 11’s limitations register. No threshold-near-miss is observed at the predicate-list level: every predicate declared in scope at every state maps to a packet-field discharge with bi-directional traceability.

Standards interpretability across the seven-state trajectory is therefore passed at the directional threshold under EM-09-01, with two declared scope constraints — interface-obligation surfacing into structured packet fields, and per-event predicate-delta emission — both attributable to the current packet-schema parser’s coverage rather than to the underlying architecture. Predicate-list completeness against declared scope is complete at each state; the packet structure scales gracefully as the regulatory surface expands at the fork; and diagnostic locality is preserved across the six events. The interpretability evidence carries forward to Section 10.10 and Section 10.17 as the substrate against which the modular-fit and burden claims are quantified.

The interpretability evidence of Section 10.4 establishes that the seven packets surface their declared commitments at a granularity a reviewer can audit. The next question — the subject of evaluation question EQ-02 — is whether the dwellings reported in those packets fit the modular architecture of Chapter 6, and it is measured through EM-09-02, the modular-fit and bounded-check measure pre-registered in Chapter 4, Section 4.4.

10.11 Operational definition and the module-inheritance ratio

Modular fit is decomposed into three operationally distinct components. The first is interface-obligation conformance: every shared opening, boundary, or path declared between two modules must appear bilaterally in both modules’ interface registers, with the boundary-ownership rule resolved per Chapter 6, Section 6.3. The second is module-class taxonomy fidelity: every module instance must belong to the canonical nine-type taxonomy of Chapter 6, Section 6.2 or to a documented Rule-4 variant. The third is compositional stability across transformation events — whether modules persist across events rather than churning into new identities at every change.

The third component is operationalised by the module-inheritance ratio — the proportion of a dwelling’s modules carried over unchanged from the immediately preceding state (formerly reported as the Composition-Boundary-Stability Index). The ratio is the proportion of a state’s modules that are inherited from its immediate predecessor without spatial change: the count of modules whose declared area-and-position is identical to the predecessor’s, divided by the total module count at that state. A state at which no module’s spatial boundary moved scores one; a state composed almost entirely of new or reconfigured modules approaches zero. It is a generic compositional-stability quantity — a set-cardinality ratio under a spatial-equality test against the immediate-predecessor module set — whose definition is independent of the trajectory’s specific module identifiers, events, or fork structure, and which neither inherits parameters from the per-state data nor selects features observed after the fact. The module-inheritance ratio is a concrete sub-metric coined within this chapter that operationalises EM-09-02 and does not displace it as the canonical name: EM-09-02 — modular-fit and bounded-check performance — remains the pre-registered summative indicator for Proposition 2 per Chapter 4, Section 4.4, and the ratio is defined here rather than in Chapter 6 only because Chapter 6’s governed-kernel specification does not pre-commit to a single quantitative stability metric.

The ratio’s location of definition within this chapter invites a hostile examiner’s question — was the measure designed to fit the data, or the data fitted to the measure? The honest answer is neither, and the ratio is falsifiable in two directions. It would have falsified the architecture’s compositional-persistence claim had the three trajectory-wide stable anchors (the primary bedroom, the kitchen wet-wall, and the service core) been forced to reconfigure across any event, since the trunk-mean would then have collapsed below one half; and its calibration would have failed had the three non-spatial events produced values comparable to the two largest spatial events, since it would then not have distinguished the trajectory’s transformation classes. Neither obtains: the per-state values track the engineering content of each event independently of the metric’s authorship, which is the discriminative pattern reported below.

10.12 Per-state modular-fit profile

Table 10.4 reports each state’s modular-fit profile across the three components. The interface-obligation conformance status is qualified per the parser-coverage limitation of Section 10.7: bilateral declarations are prose-resident, and structured-field surfacing is the parser-extension item. The conformance verdicts also inherit the empirically-grounded coupling commitments transmitted from Chapter 8, Section 8.33 as the HC-8C handoff contract.¹⁸

Three observations follow. First, the canonical nine-type taxonomy carries seventeen of the twenty-two module instances at S5b — about seventy-seven per cent of the dwelling’s modular surface — without variant escalation, the variants accounting for the remaining twenty-three per cent. Second, the variant share rises with the dwelling’s architectural age, from zero across the first three states to twenty-three per cent at the fork, tracking the engineering content of each event. Third, the interface-obligation count grows from seven at S0 to twenty-seven at S5b — almost a fourfold expansion — and the table marks every state’s conformance as prose-level, reflecting the Section 10.7 parser-coverage limitation: independent verification of bilateral consistency at any of the twenty-seven S5b interfaces currently reads the state file’s body rather than the structured packet field.

10.13 Per-state module-inheritance ratio

Table 10.5 reports the ratio per state, computed against the immediate predecessor. A module that is reconfigured — resized, repositioned, or repurposed spatially — counts as not-inherited at that transition; a module that is reclassed without spatial change (a bedroom to an office, or to a helper room) counts as inherited, because its spatial vector is preserved and the reclass is a governed-instance-library governance event rather than a spatial one.

The values discriminate the trajectory’s transformation classes sharply. The two highest — one at S1 and zero point nine-two-three at S4 and S5a — correspond to the three non-spatial events (occupancy commitment, empty-nest reclassification, aging-in-place retrofit); the two lowest — zero point five-three-eight at S3 and zero point five-nine-one at S5b — correspond to the two largest spatial events (family-growth expansion and the SDA-overlay-with-secondary); and the intermediate value at S2 corresponds to the family-formation expansion. The trajectory-level aggregate, the unweighted mean of the six per-step values, is zero point seven-nine-two: on average, about seventy-nine per cent of each state’s modular composition is inherited unchanged from its predecessor. Computed across the trunk-only sequence the value rises to zero point eight-three-two; across the right-fork branch it falls to zero point seven-six-six, because the SDA-overlay event drives the largest single perturbation. Both branches sit comfortably above one half — the level at which more than half of each state’s modules would be new or reconfigured at every step — and the trunk-versus-fork divergence is the trajectory’s reading of the architecture’s burden-proportional-to-scope property.

10.14 Stable anchors

Three modules are inherited unchanged across the entire trajectory: the primary bedroom, the kitchen’s wet-wall position, and the service core. The primary bedroom holds its dimensions and position from S0 through to the S5b primary dwelling; the kitchen wet-wall anchors the plumbing routing across every state; and the service core holds the laundry stack, hot-water unit, and switchboard in their baseline positions throughout. These persistent anchors are the architectural core of the trajectory’s coherence: the same physical envelope hosts seven distinct life-stages without losing the modules that establish its identity, and the ratio’s aggregate value is the reading of an architectural commitment — that the governed kernel admits compositional persistence — exercised over a realistic life-cycle rather than a coincidence of the trajectory’s authoring.

10.15 Variant introduction and its structured-field limitation

Four module-class variants and one site-level exception are introduced across the trajectory, and each is absorbed at the governed instance library without modifying the governed-kernel taxonomy. The dining zone and the home office are Rule-4 variants of Living, varying optional elements (partition arrangement; task-lighting and power provision) without weakening any required element; the helper room is a Rule-4 variant of Bedroom with a relaxed circulation assertion for non-participant-exclusive use; the connecting link is a Rule-4 variant of Exterior within its declared degrees of freedom; and the multi-dwelling site composition at S5b is the only case approaching a governed-kernel boundary.

The multi-dwelling case is worth developing, because it is the trajectory’s only test of the architecture at a governed-kernel boundary. Two responses were available, and each would have crossed the core-protection boundary if treated as a schema-level change: extending the Dwelling module’s composition contract to permit multi-dwelling-instance lot composition, which would touch the governed kernel’s composition layer; or introducing a new site-level module class governing lot composition, which would touch the taxonomy layer. The trajectory’s authoring takes neither. It treats the case as a Rule-4 exception class — the secondary dwelling is instantiated as an additional Dwelling instance bound to the same lot identifier, and the shared-service and cross-dwelling-egress obligations are declared as a typed exception with documented rationale referencing the SDA scheme’s secondary-dwelling provisions — so the architecture absorbs the multi-dwelling composition at the governed instance library rather than by amending its core. The exception register holds two such cases at the trajectory’s end, the office class and the multi-dwelling composition, both within the pre-registered budget; the variant share that rises to twenty-three per cent at the fork is therefore carried entirely by typed, justified governed-instance-library extensions, and the governed-kernel taxonomy is unamended across all seven states. A second instance of the Section 10.7 parser-coverage limitation applies to these variants: the packets carry the dwelling-aggregate’s module type without a dedicated field surfacing the embedded variants at the child-module level, capturing the variant declarations in the configuration-notes prose rather than in the structured predicate or interface lists. The remediation is the same parser extension scoped at Chapter 9, Section 9.6.

10.16 Hard-constraint cases and the verdicts

Two known boundary cases occur on the trunk. BR2 at S2 is authored at eleven square metres and BR4 at S3 at ten square metres, each below the Fully Accessible typology’s nominal twelve-square-metre bedroom minimum. Both are documented Rule-4 relaxations on a non-required-element axis — the area floor for non-participant-exclusive bedrooms — and both remain above the National Construction Code’s seven-square-metre habitable-room minimum, so the regulatory floor is not crossed. They are reported under EVID-P4-VARIATION as legitimate variants rather than constraint violations.

The directional threshold for EM-09-02 reads “improved bounded verification relative to baseline”, the baseline being verification scope under monolithic representation, where a structural addition would trigger a global re-walk of every clause across the dwelling. The trajectory’s verification scopes are bounded at each event to the modules touched by the transformation, with cross-module checks bounded by the Section 6.3 interaction rules, and no event triggers a global re-walk: the family-growth event bounds verification to its four instantiations, four reconfigurations, and seven new interfaces; the SDA-overlay event to its eight new modules and nine new interfaces; and the non-spatial events to their reclasses alone. The verdict for EM-09-02 is therefore a qualified pass subject to declared scope constraints — a boundary-limited acceptance — supported by EVID-P2-CHECK-SCOPE and EVID-P3-INVARIANTS, the qualification being the Section 10.7 interface-obligation surfacing constraint. The directional threshold for EM-09-04 reads “all exceptions typed and justified; rate within declared budget”; the trajectory’s five typed exceptions sit within the single-digit budget, each typed and justified against its governance basis, and the verdict is likewise a qualified pass supported by EVID-P4-VARIATION and EVID-P4-EXCEPTION-BUDGET.

Modular fit across the seven-state trajectory is therefore passed at the directional thresholds for both EM-09-02 and EM-09-04, with the same boundary-limited-acceptance constraint on interface-obligation surfacing reported at Section 10.4. The canonical nine-type taxonomy carries seventy-seven per cent of the modular surface at the largest state; the remaining share is absorbed by Rule-4 variants and one exception case, all typed and justified within budget; the ratio reports an aggregate of zero point seven-nine-two with the three stable anchors inherited unchanged throughout; and the two bedroom-area boundary cases are documented Rule-4 relaxations rather than violations. The modular-fit evidence carries forward to Section 10.17 as the structural reading paired with Section 10.4’s interpretability evidence in support of the chapter’s central Proposition 4 demonstration: divergent regulatory-compliance pathways absorbed at the governed instance library from a common substrate without modification of the governed-kernel architecture.

The third evaluation question on the chapter’s docket — EQ-05 — asks whether the integrated suite reduces practical workflow burden while maintaining or improving assurance reliability, measured through EM-09-03, a multi-dimensional delta across time, interpretive decisions, and re-authoring rate under matched-task comparison.¹⁹ The baseline is the unserialised-text condition of Chapter 9, Section 9.3: documentation produced manually, version-controlled poorly, synchronised only at formal handovers, and re-authored substantially whenever a clause, commitment, or built component shifts. This section reports the burden evidence and then synthesises the chapter’s three result domains into an aggregate verdict against the four critical properties.

10.18 Per-event workflow burden

Table 10.6 reports the per-event burden profile drawn from the prototype’s run-log and the emitted packets. Each event terminates in one documentation packet, and the figures are the emission cost of producing that packet given the substrate stack already authored.

Total emitted output across the seven packets is eighty-nine predicates and fifty-two thousand six hundred and ninety bytes — under fifty-two kilobytes of compliance-bearing material for a complete single-dwelling life-cycle — in under seven seconds of wall-clock execution under the deterministic-Python runtime of Chapter 9, Section 9.19.²⁰ The run-time is dominated by interpreter start-up and file input–output rather than by transformation work, so the marginal cost of one additional packet is bounded by reading and writing one file.

The architecture’s efficiency claim is a redistribution claim rather than a free-lunch claim, and the section is explicit about what it does not assert. The pre-vetted-library precondition of Chapter 9, Section 9.12 places the principal authoring effort on the substrate curator — the baseline-library, the predicate grammar, the schema validator, and the verification protocol must be specified before the per-design-instance pipeline can be invoked at near-zero recurring cost. Comparing the seven-second emission to the fourteen-day governed extraction effort for the empirical corpus would invite a six-orders-of-magnitude reduction claim; that comparison is not made, because the two activities are not commensurable — the corpus extraction is interpretation work performed against unstandardised plans, while the trajectory’s emission is post-curated transcription of pre-authored, schema-conformant specifications whose validation passes by construction. A defensible burden claim keeps the comparison within commensurable categories: per-design-instance emission, given an authored substrate, against per-design-instance documentation production under unserialised text. The matched-task baseline for each event is reasoned analytically — roughly one to two practitioner-days at the family-formation event, three to five at family growth, one to three at the aging-in-place fork, and ten to fifteen at the SDA-overlay fork — and these figures are explicitly analytic projections, reasoned from the documentation labour each event would demand rather than executed against a practitioner cohort, with no practitioners behind them.

The two largest events make the comparator concrete. At the family-growth event the baseline workflow would redraw the dwelling to incorporate two new bedrooms and a second bathroom, apply the residential code’s bedroom, ventilation, and circulation provisions against the redrawn plans, regenerate the council compliance documentation, and produce trade-level construction drawings — the documentation-regeneration component alone consuming on the order of three to five practitioner-days, against which the architecture emits one packet of thirteen predicates in under a second. At the SDA-overlay fork the gap is widest: the baseline workflow confronts the Standard’s eight Robust clause categories, each requiring explicit clause-level assessment against the redrawn plans, a cross-dwelling services contract that councils represent inconsistently across jurisdictions, and a multi-dwelling-site composition that triggers planning-permit as well as construction-certificate revisions — on the order of ten to fifteen practitioner-days — against which the architecture emits one fifteen-predicate packet through a single composite Robust-category overlay declaration. At the office and exception events the architecture’s contribution is strongest at the cognitive-load dimension rather than the time dimension: the office’s exception-class status is declared once, in a typed exception register, rather than re-justified each time a comparable conversion is requested elsewhere. These remain the analytic projections the matched-task baseline supplies; none rests on observed practitioner behaviour, and the chapter treats them as the reasoned comparator that gives the measured emission cost its evidential weight, not as field measurements.

The verdict for EM-09-03 against its directional pre-registration is a qualified pass subject to declared scope constraints — a boundary-limited acceptance. Per-design-instance emission burden, given the substrate stack, is bounded by sub-second run-time and tens-of-kilobytes output across a complete life-cycle, and it grows sub-linearly in fabric-modification volume: the predicate count grows by four from S0 to S5b over a regime change that introduces eight Standard clause categories and a second dwelling. The non-negotiable falsifiable floor of Section 4.5 — that the suite must not increase verification burden for equivalent assurance quality — is not breached: every packet validates against the schema. The chapter cannot and does not claim a generalised practitioner-burden reduction without a practitioner-cohort study, which it declares as future work.

10.19 Synthesis matrix

Table 10.7 collects the chapter’s three result domains, the summative measure anchoring each, the evidence objects, the propositions addressed, and the adjudication category.

Every Chapter 4 evaluation question has a corresponding row carrying at least one named evidence object, so the evidence-balance rule is satisfied within the chapter’s scope; three rows adjudicate as boundary-limited acceptance, and the fourth as a pass within budget. The four measures admit a single cross-state visual reading: the composite of Figure 10.7 lays out interpretability, modular fit, workflow burden, and exception governance, and its joint reading is that the trajectory yields uniform predicate-list interpretability, a modular-fit aggregate above one half at every state, sub-linear predicate-count growth alongside an approximately one-and-a-half-times byte increase at the S5b overlay, and five typed variant introductions within budget.

10.20 Aggregate verdict against the four properties

Across the four critical properties first articulated in Chapter 2, Section 2.9, the trajectory’s evidence permits a property-level verdict. Proposition 1 — interoperability, operationalised as semantic continuity and identity persistence across regime change, is strongest at S5b, where the dwelling crosses from Fully Accessible to Robust and the packet emits both the inherited typology trace and the activated Standard trace as a single composite overlay while the seven packets form a continuous identity chain; the verdict is boundary-limited acceptance, the boundary being the absence of an external interpretive-divergence study. Proposition 2 — transportability, operationalised as the conversion of broad-reconstruction verification into bounded local checking, is supported by the per-event scope-bounding of Section 10.10, strongest at the fork pair where two successor states emerge from one substrate under two verification scopes with neither pathway requiring the other’s modules to be re-verified; the verdict is boundary-limited acceptance. Proposition 3 — manipulability, operationalised through round-trip fidelity, is not exercised by the trajectory itself — the prototype runs one-way, from specification to packet — so the chapter’s contribution to Proposition 3 is the procedural-instantiation substrate against which Chapter 7, Section 7.20 exhibits the round-trip evidence proper, where the replay-determinism record reports twenty-two of twenty-two deterministic checks in agreement (fifteen replay tests, three boundary tests, and four verification re-runs) and the invariant-preservation record twenty-four of twenty-four across the twelve active boundary pairs; the verdict is boundary-limited acceptance via Chapter 7 inheritance. Proposition 4 — transformability, operationalised as legitimate variation under shared rules without rule violation, hidden coupling, or unstable interfaces, is the chapter’s strongest property-level claim: the fork at S4 produces two divergent successor states from a common substrate, neither requiring the other’s rules to be relaxed, and the variation-governance machinery absorbs the trajectory’s typed exceptions within budget; the adjudication is a pass within the exception-rate budget, the boundary being the trajectory’s single-fork corpus.

These verdicts are best read not as a set of four separate findings but as one evidence-bearing instantiation of the single architectural claim the thesis makes — that a governed-kernel modular architecture absorbs life-cycle and regulatory change at the governed instance library. That claim is discharged through several facets, each carrying a distinct part of the evidential burden: the interpretability, modular-fit, and burden evidence of the three result domains, and, decisively, the fork’s demonstration of governed divergence. The chapter does not claim contributions the artefact chapters have not independently established; it claims that those independent contributions, integrated through the trajectory, produce the integrated evidence the design theory requires.

10.21 Disconfirming and boundary outcomes

The negative-results policy pre-registered in Section 4.5 requires that adverse outcomes be recorded as first-class evidence. The chapter’s evidence base contains five. The first is the S0 baseline’s deliberate position at the low-volume edge of its cohort, within half a standard deviation of the cohort median and in-distribution but anchored low by design so that the additive events have room to grow; it is a scope-limit, and no architectural claim is falsified. The second and third are the BR2 and BR4 bedroom-area relaxations of Section 10.16 — typed Rule-4 exceptions on a non-required-element axis, recorded as evidence for Proposition 4 rather than against it, with BR4 at ten square metres sitting closest to the inferred geometric breakpoint below which a single-bed access pattern could not be realised; that breakpoint is reasoned rather than declared, and the chapter marks it as its most fragile inferential commitment on the trunk. The fourth is the S5b Robust governed-kernel composition: of the fifteen Robust-category predicates in the S5b packet, only two are directly authored from baseline-library entries (Kitchen and Exterior), the remaining thirteen inferred via Rule-4 substitution from their Fully Accessible parents — a direct-to-inferred ratio of about two to thirteen — and the most exposed substitutions are the bathroom reinforced-wall lining, the floor-planarity tightening, and the egress-path-count requirement, the last because egress-path count sits closer to a required-element commitment than to an optional attribute under most regulatory readings. No independent Robust-category baseline-library exists against which the inferred predicates could be falsified, so the gap is a substrate-tier scope-limit absorbed with explicit declaration, and the future-work entry to author the Robust entries directly is recorded. The fifth is the parser’s interface-obligations field of Section 10.7, a prototype scope-limit acknowledged in Chapter 9, Section 9.19. None of the five falsifies the architecture’s core claims; each is surfaced here rather than absorbed into selective omission.

10.22 Cross-chapter integration, limitations, and what the trajectory does not exercise

The trajectory’s evidence inherits substrate from each artefact chapter: the ambiguity-management discipline of Chapter 5, exercised at the S5b Standard activation; the modular-fit claim of Chapter 6, exercised at the per-state level; the round-trip evidence of Chapter 7, supplemented by the chapter’s procedural-instantiation substrate; the census statistics of Chapter 8, against which the seven states are positioned descriptively; and the deterministic-emission claim of Chapter 9, the mechanism that produces the seven packets. Each property-level verdict promoted to Chapter 11 carries its measure, its evidence object, and its adjudication category, satisfying the contribution-bridge condition that only claims with linked evidence and declared validity limits are promoted, and the four Technological Rules routed forward are read as the anticipated forms of the single architectural claim’s instantiations rather than as four independent contributions.

The chapter’s evaluation is candidate-driven throughout — the candidate authors the states and event specifications, operates the prototype, and reasons the matched-task baselines — and external practitioner-cohort evaluation is incompatible with single-doctoral-study scope. The boundary is recorded at the Technical-Risk-and-Efficacy classification of Section 4.4, and its consequence is uniform: each property-level adjudication is boundary-limited acceptance, or a pass within budget for Proposition 4, rather than a generalised pass. Five territories the trajectory does not exercise are recorded for Chapter 11’s future-work agenda: cross-corpus generalisation to cohorts beyond CP-D5; independent third-party round-trip verification against an external implementer’s engine; the Universal Design, Modular, and Improved Liveability design categories beyond the Fully Accessible and Robust pair exercised here; production-grade tooling beyond the minimum-viable prototype; and large-language-model-assisted authoring layered above the deterministic core. The five exclusions are consistent with the chapter’s instantiation-depth posture: the trajectory exercises the minimum-viable subset sufficient to test the architecture’s core claims, and the broader exercises are properly objects of subsequent inquiry. The conclusion that follows assigns a proposition-status verdict to each of Proposition 1 through Proposition 5 and hands them to Chapter 11.

The trajectory’s proposition-status verdicts close the chapter, and they are the input Chapter 11 synthesises into the thesis’s contribution. This conclusion recapitulates what the demonstration has established, assigns a verdict to each of the four critical properties and to the secondary burden measure, and states the contract under which those verdicts become Chapter 11’s empirical anchor.

The demonstration has shown the artefact suite operating as one integrated system across a full life-cycle dwelling trajectory. The six transformation events and the fork at S4 jointly exercise Proposition 4 transformability under regulatory and occupant change without breaking the architectural commitments declared in earlier chapters; the seven emitted documentation packets supply structured evidence for Propositions 1, 2, and 3 at every state; and the cumulative burden trace supplies the per-event Proposition 5 evidence of Section 10.17. This is the thesis’s first integrated, life-cycle-scope test of the architectural claim, and its evidence is the substrate against which Chapter 11 interprets the contribution — not as five separate results, but as the single architectural claim discharged through several evidence-bearing instantiations.

The proposition-status table adjudicates against the per-property aggregation scheme of Chapter 4, Section 4.5 — SUPPORTED at HIGH or MODERATE confidence, DECLARED-LIMITED, or REJECTED — into which the per-domain boundary-limited-acceptance verdicts of Section 10.17 promote. A domain verdict promotes to SUPPORTED where a property is carried by converging result domains within the trajectory’s scope, and to DECLARED-LIMITED where its evidence is bounded by inheritance (Proposition 3 from Chapter 7) or by candidate-driven scope (Proposition 5).

The verdicts carry three characteristic features. Propositions 1 and 2 are SUPPORTED at HIGH confidence within the tested-scenario class, and Proposition 4 at MODERATE confidence — the triad that Chapter 11 is asked to interpret. Propositions 3 and 5 are DECLARED-LIMITED with explicit boundaries: Proposition 3’s round-trip evidence is inherited from Chapter 7 and the trajectory does not extend it to a non-specialist usability study, and Proposition 5 is a sub-linear-growth claim under candidate-driven instrumentation rather than a practitioner-detected field claim. No proposition is REJECTED, because the four-result scheme reserves rejection for unambiguous threshold non-attainment without a recoverable boundary explanation, and the chapter’s disconfirming outcomes do not match that profile: the parser limitation, the variant-taxonomy ambiguity, and the Robust governed-kernel scope-limit are recorded as boundary conditions and routed into Chapter 11’s limitations rather than as proposition-level rejections.

The contribution-bridge condition of Section 4.5 governs what passes from this chapter into Chapter 11: only claims with linked EM-09-* evidence and declared validity limits are promoted, SUPPORTED-at-HIGH verdicts are eligible for unrestricted contribution-claim status, DECLARED-LIMITED verdicts only with the boundary attached, and no verdict authorises an unbounded claim. Six future-work directions are routed forward, each bounded by an explicit scope-limit: external multi-rater validation of the candidate-driven adjudication; cross-corpus generalisation against cohorts beyond CP-D5; the Universal Design, Modular, and Improved Liveability category expansion; production-grade tooling; large-language-model-assisted authoring above the deterministic core; and a multi-trajectory comparison testing whether the HIGH-confidence verdicts on Proposition 4 are robust beyond the single-dwelling design. The verdicts hold under the conditions declared in Section 10.1 and Section 10.17, and they are the empirical anchor against which Chapter 11 articulates the thesis’s contribution to design theory, to housing-adaptation practice, and to the regulatory-governance literature.

Navigation: ← 10.17 Results: Workflow Efficiency and Discussion of Results | ↑ Chapter 10 | Chapter 11: Discussion →