Introduction

Ability-diverse households — households whose abilities vary by disability, by age, or by life-stage — make demands of housing that evolve across the long life of their buildings. In Australia, the share of households requiring accessible or adaptable dwellings has risen steadily for two decades, driven by demographic ageing and rising disability prevalence, and reinforced by a policy shift toward sustained occupancy in place rather than institutional relocation.¹ Most of those households will be housed by stock that already exists, which makes the binding constraint the absorption capacity of the existing fabric — the rate at which dwellings already standing can take up the change that ability-diverse occupancy asks of them, a capacity held back, on our reading, as much by representational infrastructure as by physical fabric. Drawings, schedules, certification reports, occupancy protocols, regulatory texts: these bind a dwelling to the commitments under which it was originally consented, and they do not, by themselves, accommodate the regulatory and compositional churn that ability-diverse occupancy demands across a building’s lifetime.

Our response to that constraint is an architecture: a modular structure under which a tool, a method, and a model are held in disciplined relation, so that a local change to a dwelling’s representation can be made and checked at the point of change rather than re-rippled across the whole. It belongs to design science — a response is admissible as an artefact, or a coordinated suite of artefacts, whose theoretical content is exposed through declared mechanisms, indicators stated in advance, and falsifiers tabulated in advance.²³ The discipline this architecture institutes — holding a dwelling’s representational infrastructure in mutual consistency with its regulatory, stakeholder, and documentary commitments under change — is what we name representational governance (Chapter 3). What we develop, accordingly, is that architecture; what we test is whether it, instantiated through a coordinated artefact suite, can do the absorption that adaptation-heavy housing requires of its representational infrastructure.

The architecture inherits from two intellectual lineages. From the modularity-and-complexity literature (Chapter 2) — Simon’s account of near-decomposability, Parnas’s information-hiding criterion, and Baldwin and Clark’s formalism of design rules — we take the architectural finding that complex systems remain tractable to incremental change when their interdependence is bounded by stable interface contracts that publish at module boundaries.⁴⁵⁶ Architectural commitment, in this reading, lives at the partition between what publishes for cross-module use and what every module retains as its private decision space. From the Open Building literature (Chapter 2) we take a complementary housing-flexibility commitment: the layered reading of the residential dwelling that Habraken inaugurated in 1972 and that Kendall and Teicher consolidated in 2000, in which the dwelling functions as a substrate whose long-life support supplies a capacity that the short-life infill occupies and re-occupies across the building’s life.⁷⁸ The architectural move we attempt is, on our reading, Baldwin and Clark’s publish/hide partition transposed to a setting that differs from its software-and-manufacturing home along three axes: a module replacement in a dwelling is borne by the existing stock; the regulatory text governing the design is in natural language whose interpretive uncertainty resists test-automation; and the regulators who write the standards co-author the interface alongside the designer, a transposition developed at Section 1.3.

The architecture’s substantive move, made concrete, is a governed-kernel housing-module taxonomy that holds nine module-types and supports them with a stratified spatial vocabulary of seven primitives and seven composites, a small set of interaction rules, and a Rule-4 variant-inheritance mechanism that permits the architecture’s governed instance library to absorb life-cycle functional variation without altering the governed kernel itself (Chapter 6). The architecture publishes a nine-category reading of dwelling space — the connective spaces by which a home is entered and moved through (entry, circulation, and outdoor or external areas), the functional rooms in which daily life is conducted (sanitary or wet rooms, bedrooms, living areas, kitchens, and service or utility spaces), and the dwelling considered as a single governed whole — together with the boundary contracts that govern how those parts may legally combine; it hides the within-module choices through which each module’s quality is realised. This is modularity in Baldwin and Clark’s specific publish/hide sense, transposed for the dwelling, and it is the engine that every artefact in the suite exists to encode, instantiate, validate, or operate.

The empirical setting under which the engine is developed and tested is Specialist Disability Accommodation under the Australian National Disability Insurance Scheme (Chapter 5).⁹ The scheme funds purpose-designed dwellings for NDIS participants whose disability supports require dedicated housing, and a single mandatory design standard, whose six hundred and eleven normative clauses sort each dwelling into one of four certification categories — Improved-Liveability, Fully-Accessible, Robust, and High-Physical-Support — determines eligibility for the payments through which the scheme operates.¹⁰ The standard binds across the National Construction Code and the AS 1428 accessibility series, with registered SDA assessors required to trace each design decision back to the clause it implements.¹¹ We selected SDA because it concentrates into a single auditable instrument the diagnostic conditions under which the architecture must operate: a heterogeneous stakeholder coalition, a compliance text under repeated amendment, and cross-document obligations easily hidden in the gaps between artefacts. The architecture we develop is general; SDA supplies the evidence on which we rely.

We conduct the inquiry as design-science research (Chapter 4) in the Hevner-March-Smith tradition, treating construction, instrumentation, evaluation, and reflective audit as the four activities through which artefactual knowledge is produced and held open to test.¹² The architecture, on our argument, should sustain four properties developed across Section 2.9: interoperability, by which the representation travels intact across handover between actors whose disciplinary frames differ; transportability, by which a local change does not force global reinterpretation of the design; manipulability, by which an authoring or amending act on the representation can be replayed without information loss; and transformability, by which variant pathways become admissible under shared rules whose violations remain auditable. Each property is paired in Section 3.5 with a propositional mechanism intended to operationalise it for test (Proposition 1, semantic continuity; Proposition 2, bounded verification; Proposition 3, executable replay; Proposition 4, governed variation), and a fifth proposition — Proposition 5 — addresses what we tentatively call the integrated utility of the architecture across the dwelling’s diachronic trajectory.

The empirical test turns on a single fork. Chapter 10 takes the suite end-to-end through a single-dwelling tree-branching life-cycle trajectory whose state S4 — a one-hundred-and-sixty-five-square-metre four-bedroom envelope — branches two ways: a typical aging-in-place adaptation that keeps the original occupants ageing in place on one side, an SDA-Robust refurbishment with attached secondary dwelling housing an adult occupant whose support needs require SDA-Robust provisioning on the other. The architecture absorbs both branches at its governed instance library while the governed-kernel taxonomy stays fixed, and that absorption — two divergent regulatory-compliance pathways issuing from a common substrate without disturbing the governed kernel — is the architecture’s case for governing adaptation-heavy housing representation under regulatory change. The demonstration is bounded to one dwelling-and-lot under SDA, with the bounds recorded at Section 1.6.

Five artefacts compose the suite. We advance one architectural claim, and we render it inspectable through five evidence-bearing instantiations whose joint discharge constitutes the contribution, which the synthesis at Section 11.2 reads back as that one claim. The standardisation schema of Chapter 5 lifts the SDA Design Standard’s normative prose into a serialised, queryable representation, the clause-level surface to which the architecture’s primitives attach. The Governed Kernel Architecture of Chapter 6 is the engine’s home, carrying the nine-type taxonomy, the seven primitives and seven composites, the interaction rules, the Rule-4 variant-inheritance mechanism, the boundary contracts, and a baseline-library populated to nine entries at full schema depth. The two-layer planimetric notation of Chapter 7 encodes the engine’s compositions in a form whose verification properties can be examined under round-trip parsing. The substrate of Chapter 8 supplies empirical grounding, anchored in dimensional, occurrence, topological, and configurational evidence from a 745-plan corpus screened from the Australian residential floor-plan stock, and the generation-and-documentation pipeline of Chapter 9 supplies procedural instantiation. Each artefact renders one face of the single claim inspectable, and the contribution holds only when the five hold together, as Section 1.5 sets out.

The suite, in the order in which the chapters develop it:

The remainder of the chapter sequences the framing that supports the claim: the problem class at full depth in Section 1.2; the contribution-relevant gap in Section 1.3; aim, objectives, and research questions in Section 1.4; the five contributions, declared as the discharge instantiations of the single architectural claim, in Section 1.5; and scope, delimitations, and the thesis-structure map in Section 1.6.

Adaptation-heavy housing places a particular kind of demand on its representation. Across the long life of a dwelling, occupant trajectories shift — with age, with the onset or progression of disability, with the entry and exit of household members whose support requirements differ from the original brief. Regulatory texts move on cycles the original consent did not anticipate, as standards are amended, superseded, or extended by technical guidance between formal revisions and disability-housing policy undergoes recurrent reviews. Stakeholder coalitions turn over, so that the actor who authored the original specification is, in the general case, not the actor who reads, certifies, or modifies it later. Each shift is, on its own, modest, and each is meant to stay local — a kitchen renovation, a category re-certification, a clause update absorbed into a single drawing revision. What we observe in adaptation-heavy practice is that the changes do not stay local: a local change repeatedly forces global re-checking and re-issuing across the regulatory, stakeholder, compositional, and documentary commitments through which the dwelling is held together.

I shall argue that this propagation behaviour defines the problem class we develop. A problem belongs to the class when the representational network — through which a dwelling’s commitments are made traceable — cannot keep a local change local, so that ordinary maintenance, variation, or category-shift events propagate as global re-evaluation across the network. The class cuts across building type, software environment, regulatory regime, and institutional setting; it is the structural pattern Chapter 2 develops at depth, in the vocabulary of containing-versus-transmitting interfaces (Section 2.8) and the four representational properties — interoperability, transportability, manipulability, transformability — that a representation must possess if its interfaces are to contain rather than transmit change. Section 2.10 names the pattern adaptation intractability under ungoverned representational form, and names the disciplined maintenance of those properties under change representational governance.

The architectural response this class warrants is the response Section 1.1 introduced: a governed kernel whose variation lives at a bounded governed instance library, so that occupant and regulatory change can be absorbed at the library without the kernel itself moving. We develop the engine at depth in Chapter 6; here our task is the empirical reality of the class in the Australian SDA setting and the membership conditions on which the rest of the chapter relies. The case for the architecture turns on a single restraint that Chapter 10 puts to test: at state S4 — a one-hundred-and-sixty-five-square-metre four-bedroom envelope — a single substrate yields two divergent regulatory pathways, one a typical aging-in-place adaptation and the other an SDA-Robust refurbishment with attached secondary dwelling, precisely because the substrate is not asked to perform the divergent work, so that both branches are absorbed at the governed instance library while the governed kernel stays put. That restraint is, on our reading, the architecture’s claim for governing adaptation-heavy housing under change.

Three structural conditions, taken together, explain why interfaces in adaptation-heavy housing transmit rather than contain. The first is distributed stakeholders. A coalition of actors governs the dwelling whose decision rights, professional accountabilities, information demands, and disciplinary frames are heterogeneous: designers, builders, certifiers, occupants, funders, regulators, and — in disability-housing settings — clinical assessors and supported-decision-making coordinators. No single actor authors the representation in full. No single actor consumes it in full. Boundaries between authoring and consumption are negotiated under shifting institutional arrangements, and a change that crosses one of those boundaries cannot, on our reading, be kept inside a single actor’s internal records; the change must be re-authored at the boundary, and the boundary is a moving target.¹³¹⁴ The second condition is evolving compliance constraints. Regulatory texts that govern adaptation-heavy housing are themselves under continuous authoring activity by parties external to the original design. Australia’s National Construction Code is amended on a three-year cycle with intermediate technical updates;¹⁵ the SDA Design Standard is subject to NDIA technical guidance amendments outside its formal revision cycle;¹⁶ the AS 1428 accessibility series is revised through Standards Australia’s technical-committee process; and disability-housing policy is subject to recurrent reviews whose recommendations propagate into design guidance, certification protocols, and funding arrangements.¹⁷ A dwelling certified under one version of a standard does not automatically retain certification under the next. An extension authorised under one revision of the NCC may require re-certification under the revision in force at the time of the extension. The third condition is cross-document handovers that hide boundary obligations. A dwelling’s commitments are distributed across regulatory text, design documentation, certification reports, and provider documentation, and the boundaries between these documents are matters of professional convention, not architectural commitments embedded in the representation itself. Which clause is realised by which drawing? Mapping which assessor verdict is contingent on which specification element, and which operational protocol presupposes which design feature, is practice-knowledge held by individual practitioners. When a local change crosses one of these boundaries, the obligations the boundary was implicitly carrying are not, by themselves, visible to the actor making the change. They surface only when the change is re-presented to the next document in the handover sequence — at which point the change has already propagated.

The three conditions compound. Distributed stakeholders make evolving compliance harder to absorb, because no single actor can integrate the changes across the affected coalition. Evolving compliance makes cross-document handovers more brittle, because the conventions that previously held the documents in mutual consistency are themselves subject to revision. Cross-document handovers make distributed-stakeholder coordination harder, because each handover crossing requires re-establishing whose boundary obligations bind whom. The compound effect, on our reading, is that the representational network’s containment capacity scales inversely with the rate of change in the surrounding regulatory and institutional environment, so that intrinsic complexity of the dwelling fades as a determinant of the absorption load relative to that environmental volatility.

Three SDA examples make the pattern concrete at increasing cross-document scope. Consider an SDA-Robust upgrade first: a dwelling certified under the Improved-Liveability category must absorb an occupant whose support needs have escalated, and the upgrade extends well beyond a single re-certification event. The change propagates against the SDA Design Standard’s Robust-specific clauses — impact-resistant linings, secured fittings, line-of-sight requirements, secondary-egress provisions — against the NCC’s Class 1a provisions for the dwelling’s existing fabric, and against AS 1428.1 for the accessibility commitments that survive the upgrade unchanged.¹⁸ A clause in one document may bind to a drawing element whose specification is governed by a provision in a second document whose interpretation is contingent on a dimensional rule in a third — three documents authored under different drafting conventions, by different authorities, on different revision cycles. A kitchen-renovation cascade is the second example: an apparently localised intervention in an SDA-Improved-Liveability dwelling that propagates across circulation-space provisions, across bathroom grab-rail backing requirements where the renovation reroutes services through a shared wall, across laundry adjacency where the renovation reconfigures it, across NCC energy-efficiency re-certification if the renovation alters the dwelling’s energy profile, and across the SDA assessor’s prior certification if the renovation is interpreted as material under the standard’s variation provisions. A secondary-dwelling addition exercises cross-document handovers at maximal scope — for an aging parent, an adult child, an SDA-funded occupant on the same lot, or a multi-generational arrangement: acoustic envelopes under NCC Volume 2, thermal envelopes under NCC Section J, service envelopes for water and sewer and electrical and communications, accessibility envelopes under AS 1428.1 where the secondary dwelling must be accessible, and SDA-specific envelopes where it is SDA-funded. Each envelope is a representational interface whose containment behaviour was not declared in either dwelling’s original representation; negotiating the addition’s compliance is negotiating those interfaces in the absence of an architectural contract that would have made their containment behaviour visible from the start.

The absorption failure produces a characteristic cost signature in the workflow. Documentation re-work — the production of complete, consistent, queryable, and auditable packets through which a dwelling’s commitments are re-issued after each change — appears to dominate practitioner effort in adaptation-heavy housing under regulatory constraint. We supply the empirical anchor in Section 9.2, where the documentation bottleneck is diagnosed as the dominant practitioner cost, and where four families of partial response — BIM-based documentation, automated rule-checking, schema-driven document generators, large-language-model-assisted authoring — are characterised as addressing slices of the friction while leaving the underlying representational deficit untouched. A second anchor comes from Chapter 8. The 745-plan corpus extraction required roughly fourteen days of governed effort to produce schema-conformant plan records, and what consumed the budget was interpretive composition: the resolution of regulatory referents, the disambiguation of polysemous terms, the placement of plan-level evidence into categorial slots. Pure data entry was the smaller fraction. Current tooling handles the geometric and clerical layers — drawing emission, attribute capture, conformance checking against fixed rules — while the knowledge-bearing translation that resolves regulatory meaning remained human-borne throughout. The cost signature is thus a symptom of the absorption failure that defines the class, rather than the defining feature, which lies in the propagation behaviour upstream of it.

Three distinctions sharpen what kind of contribution the class warrants. Generic computer-aided design modelling is the first. It addresses the production of geometric and semi-semantic representations that support drawing emission, quantity take-off, and — under BIM extensions — clash detection across building services. The class at issue here subsumes generic CAD modelling and exceeds it in two respects: regulatory text enters the architectural commitment as a co-author of the interface, rather than being checked against the model after the fact; and the boundary between the model and its downstream documentation operates as a containment-or-transmission interface, rather than as a rendering convention solved by habit. Software-architecture modularity in the Baldwin-and-Clark tradition is the second, and the modularity reading carries the operative theoretical weight of this inquiry.¹⁹ We argue that the architectural response signalled above is Baldwin and Clark’s publish/hide partition — their design rules formalism — transposed to a setting that differs from its software-and-manufacturing home along three axes: module re-release is borne by the existing built stock; the governing regulatory text is natural language whose interpretive uncertainty resists test-automation; and the regulators who write the standards co-author the interface alongside the designer. The third distinction is from housing-design philosophy in the Habraken Open Building tradition.²⁰ Open Building operates at the level of design-philosophy commitment: it prescribes what designers ought to commit to in the architectural choices that produce a layered dwelling, so that physical change at the infill layer does not propagate into the support layer — the support-infill split in Habraken’s vocabulary. Our contribution operates at the level of governed-representation contract, specifying how those design-philosophy commitments are encoded, exchanged, queried, and revised in a form whose containment behaviour can be inspected, replayed, and held to account when the dwelling is asked to absorb change. The contract thus operationalises the design-philosophy commitment: the two levels stand in complementary specification, the contract supplying the inspectable encoding that the philosophy presupposes.

Our contribution is, we suggest, client-independent and organisation-independent. A reader can apply the architecture without the specific client data or institutional systems we used, provided the reader can define the local containment behaviour, acceptance criteria, and traceable interface obligations for the target setting. The architecture’s components — the standardisation schema, the Governed Kernel Architecture, the notation, the procedural pipeline, the empirical-substrate discipline — are specified at a level of abstraction that should admit transposition to other regulatory regimes, dwelling stocks, and authoring authorities. Transposition itself is the task of subsequent research and lies outside the present evidence chain — a scope limit declared at Section 1.6 and discharged in the bounds discussion at Chapter 10. The class so defined supplies the warrant for the contribution-relevant gap analysis that follows at Section 1.3.

Four traditions converge on the conditions for governing adaptation-heavy housing under change, and each remains productive in its home domain: modularity, Open Building, standards-formalisation, and specialist-housing-policy address, between them, what such governance requires. Each establishes a depth of precedent we draw on; what the contribution adds is the joint reading under which the architectural response of Section 1.2 extends from its home domain into the present setting. That extension carries a cost the modularity tradition takes for granted, since the home domain differs from ours: in software and product architecture, module re-issue is a release activity and the design team authors the interface alone, whereas in housing the dwellings bear module re-issue on their own stock and the regulatory text co-authors the interface alongside the designer. We take each tradition in turn — for the precedent it supplies and the extension the joint reading adds — then state the synthesis and name the five instantiations through which it is discharged.

Modularity supplies the most developed account of interface discipline as a strategy for managing complexity and absorbing change. Simon’s near-decomposability gives the systems-theoretic foundation;²¹ Parnas’s information-hiding criterion fixes the interface contract;²² Baldwin and Clark formalise design rules as the partition between visible design rules and hidden module parameters that generates option value under disciplined upfront specification;²³ Sanchez and Mahoney extend the analysis to organisational structure;²⁴ and Sullivan and colleagues quantify the resulting option value in software-design settings.²⁵ Baldwin and Clark’s publish/hide partition — visible design rules frozen, hidden module parameters free to vary — is the operative theoretical move. The architectural response signalled at Section 1.2 is that partition transposed to a setting that differs from its software-and-manufacturing home along three axes: module re-release is borne by the existing built stock; the governing regulatory text is in natural language whose interpretive uncertainty resists test-automation; and the regulators who write the standards co-author the interface alongside the designer. The first axis arguably inverts the option-value calculus the tradition takes for granted; the second leaves interface tests partly human-borne, the natural-language uncertainty being irreducible; the third widens the design surface the tradition attributes to the design team alone. The transposition the joint reading develops across these axes is faithful to the modularity literature’s normative content and extends the settings in which that content has been applied.

Open Building supplies the substantive housing-flexibility commitment most directly aligned with the problem class. Habraken’s Supports established the support–infill distinction — the architectural reading of the dwelling as a layered system whose layers are governed by distinct authorities, revision cycles, decision-rights allocations, and evidential requirements.²⁶ Cuperus codified the tradition’s commitments for the lean-construction audience;²⁷ Kendall and Teicher synthesised two decades of Open Building practice into an infrastructural reading;²⁸ Kendall’s later treatment extended the synthesis to institutional and tenure conditioning;²⁹ Lutolli re-reads the strategies through the lens of adaptive resilience under contemporary pressures.³⁰ The tradition operates at the level of design-philosophy commitment, prescribing what designers ought to commit to in the architectural choices that produce a layered dwelling; we add the governed-representation contract at which those commitments are encoded, exchanged, queried, and revised under regulatory change. A downstream certifier would need to verify that the support–infill stratification has been preserved across adaptation; stakeholders would need to query the representation to determine which decisions fall within their decision-rights allocation; and under a regulatory amendment re-allocating decision rights between authorities, the representation would need to be re-issued. The contribution is faithful to the Open Building philosophy and operationalises its commitments in a form amenable to representational governance under change.

Automated rule-checking is what the standards-formalisation literature supplies in most developed form. Eastman and colleagues’ foundational programme established that regulatory requirements can be formally encoded and computationally evaluated against building information models;³¹ Solihin and Eastman’s rule classification formalised the Class 1 / higher-class distinction;³² Hjelseth’s RASE methodology introduced clause-role mark-up;³³ Dimyadi and Amor traced the field’s trajectory across a decade;³⁴ Zhang and El-Gohary extended the tradition to integrated NLP-and-logic reasoning;³⁵ the BIM Handbook synthesised the broader tradition.³⁶ This tradition takes the standard as input and emits the verification verdict as output. We add the layer beneath it: a governed representation whose interpretive uncertainty is tracked as a first-class element of the schema, available for downstream stakeholders to inspect and negotiate rather than presupposed-resolved before checking begins; cross-document obligations declared as part of the architecture rather than emerging at handover as the verdict alone; and evolution under stakeholder change supported by the architecture rather than requiring re-encoding and re-validation of the rule-checking apparatus at each clause amendment. The relationship is complementary: BIM-based rule-checking and IFC-based interoperability supply essential infrastructure for computational compliance, and the contribution supplies the representational layer that infrastructure presupposes.

Specialist-housing-policy supplies the domain stakes and the human-rights framing within which the problem matters. The NDIS Act 2013 establishes the legislative framework for SDA;³⁷ the SDA Design Standard 2019 specifies the four certification categories;³⁸ Bringolf traces the Australian policy trajectory through which accessibility has been progressively incorporated into housing-design discourse;³⁹ Imrie documents the lived consequences of representational failure in accessibility-policy implementation — exclusion from housing tenure, retrofit costs borne by occupants, dignity loss, downstream economic exclusion;⁴⁰⁴¹ and Imrie and Hall extend the analysis to the broader inclusive-design movement.⁴² The literature operates at the level of policy outcomes, drawing its analytical apparatus from housing-studies, disability-studies, policy-analysis, and human-geography. We add the architectural framing for cross-document obligation tracing — spare commitments (bounded interfaces, hidden information, stratified substrates, falsifiable propositions) that carry the human-rights stakes the policy literature documents into the design-representation discipline those outcomes presuppose. The architecture sits at the layer beneath the policy outcomes and complements the policy-analysis apparatus.

The joint reading we develop holds the four traditions together: an architectural commitment whose components are auditable as the modularity tradition requires, whose substantive content is faithful to the Open Building philosophy, whose verification machinery is compatible with the standards-formalisation tradition, and whose human-rights stakes are answerable to the specialist-housing-policy tradition’s evidential standards. Its operative move is the publish/hide partition transposed to a domain where the regulatory text co-authors the interface and module re-issue is borne by the built stock. Counterfactual reasoning sharpens what each tradition establishes and what it leaves wanting here: modularity alone delivers bounded interdependence yet supplies none of the layering rationale adaptation-heavy housing presupposes; Open Building alone delivers a layered design philosophy yet none of the compliance-checking discipline regulatory adaptation requires; standards-formalisation alone yields verifiable compliance yet none of the lived-consequence traceability the policy regime exists to secure; and specialist-housing-policy alone delivers substantive accessibility yet none of the architectural tractability the diachronic burden demands. Each is sufficient within its home domain and insufficient for the joint conditions, and the additive contribution is the reading that holds across all four at once.

Five evidence-bearing instantiations discharge the joint reading — five faces of one architectural commitment, per the self-account at Section 11.2. The standardisation schema (Chapter 5) operationalises the standards-formalisation tradition’s verification ambition through a five-dimensional decomposition of the SDA Design Standard’s normative prose; it instantiates the queryable serialisation of regulatory text. The Governed Kernel Architecture (Chapter 6) is the engine’s home: the governed-kernel taxonomy, the stratified vocabulary, the Rule-4 variant-inheritance mechanism, and the boundary contracts under which composition operates. The notation (Chapter 7) encodes the engine’s compositions in a form whose verification properties can be examined under round-trip parsing. The generator (Chapter 9) operationalises cross-document obligation tracing through the procedural pipeline, drawing on the baseline-library from Chapter 6. The empirical substrate (Chapter 8) anchors the suite empirically in the 745-plan corpus against which the architecture’s commitments are calibrated. The theoretical-level articulation in Chapter 3 gives the joint reading its design-theory licence — a stratified-functional-structuralist commitment whose five propositional mechanisms (Proposition 1, semantic interface and identity persistence; Proposition 2, interface-bounded modularity; Proposition 3, executable transformation grammar; Proposition 4, platform-governed complement evolution; Proposition 5, integrated utility under diachronic burden) operationalise the four representational properties Section 2.9 identifies, together with the integrated-burden property Proposition 5 names; their one-for-one pairing is set out at Section 3.5. Aim, objectives, and research questions follow at Section 1.4; the contributions are stated in detail at Section 1.5.

Our aim is to develop a governed representational architecture for adaptation-heavy housing systems. We commit to an architecture that is stratified, modular, and auditable; we express it through a five-part suite that operationalises its commitments; and we validate it against an empirical substrate of Australian residential floor plans. Underlying the architecture is what we name Stratified Functional Structuralism (SFS), developed in Chapter 3 (Sections 3.2–3.5). SFS treats representational entities in housing as functional kinds realised through structural configurations over a discrete geometric substrate, governed by interfaces that may make change locally containable and globally auditable. We take representational architecture as the unit of analysis, in preference to methodology, process, or product, because methodologies for housing adaptation already exist in mature form while the failures at issue appear to arise one level lower, at the representational substrate on which any methodology operates and from which any process inherits. Simon’s architecture-of-complexity⁴³ and Baldwin and Clark’s design-rules formalism⁴⁴ together establish visible design rules as the locus at which complex systems remain governable, and we adopt that position. Specialist Disability Accommodation under the Australian National Disability Insurance Scheme⁴⁵⁴⁶ is the empirical anchor; the architecture is general, intended to transfer to any setting where adaptation-heavy housing is held in tension with an evolving regulatory text and where ability-diverse households must remain housed in place while their dwellings catch up to needs that shift faster than the fabric can be rebuilt.

Discipline for the aim comes from the pre-registered propositional contract at Section 3.5, where the architectural mechanism logic becomes a falsifiable design science research contract under the Hevner framework and the Gregor-Jones design-theory anatomy.⁴⁷⁴⁸ We name five propositional mechanisms: Proposition 1 (interoperability), pairing semantic interface and identity persistence; Proposition 2 (transportability), interface-bounded modularity; Proposition 3 (manipulability), executable transformation grammar; Proposition 4 (transformability), platform-governed complement evolution; and Proposition 5, integrated utility under diachronic burden. Each pairs one-for-one with the four critical-property names — interoperability, transportability, manipulability, transformability — developed in Section 2.9, with Proposition 5 carrying the integrated-burden property. Each instantiation’s contribution is testable against the falsifier specified at Section 3.5, and that evidential discipline subjects the architecture’s claims to operational test before they are credited.

Objective 1 — Theoretical foundation. Synthesise modularity theory (Chapter 2), platform architecture (Section 3.4), and wicked-problem-and-complex-adaptive-system thinking (Section 3.2) into a single design-theory scaffold under the Gregor-Jones anatomy. Discharge happens in Chapter 3 through four architectural commitments (semantic continuity, bounded composability, formal expressibility, procedural traceability) and the five propositions (Propositions 1 to 5) they license. The scaffold disciplines the vocabulary the architecture borrows from its parent literatures, so that the commitments stand internally consistent, the boundary conditions explicit, and the propositions ready for operationalisation; absent it, the architecture would carry an unaccountable mixture inherited from those literatures.

Objective 2 — The standardisation schema (Chapter 5). Specify the standardisation schema as a governed-representation schema for the NDIS Specialist Disability Accommodation Design Standard (Chapter 5), operationalised as a serialisation of the standard’s normative clauses across five evaluation dimensions — lexical ambiguity, scope ambiguity, deontic ambiguity, role ambiguity, deontic strength — together with a stratified mapping vocabulary that binds clauses to spatial primitives and composites. Discharge happens in Chapter 5 through the schema’s design rationale, the serialisation pipeline’s implementation, and the evaluation comparing schema-mediated coverage against the baseline ambiguity profile of the standard’s prose. Objective 2 is the suite’s regulatory anchor and supplies the SDA-specific evidence base for Proposition 1.

Objective 3 — The Governed Kernel Architecture (Chapter 6). Specify the Governed Kernel Architecture as a stratified vocabulary of housing module-types — seven primitives and seven composites — with formal interaction rules and constituent-element specification, instantiated in a nine-of-twenty-five Fully Accessible (FA) baseline-library bundle. Discharge happens in Chapter 6 through the move from triplets to primitives (Section 6.1), the module taxonomy and space categories (Section 6.2), the interaction rules including Rule 4 variant-inheritance for platform-governed complement evolution (Section 6.3), and the module-library specification with nine FA entries at full schema depth (Section 6.4). Modules are, on our reading, the bounded spatial units to which clauses attach, and the interaction rules govern how composition preserves the regulatory commitments expressed at the module level. Chapter 6 is the engine’s home, where the governed-kernel reading the contribution transposes from Baldwin and Clark’s publish/hide partition is set out in full.

Objective 4 — The notation (Chapter 7). Encode the architectural commitments in the notation as a two-layer scheme: RecPol is the formal core for discrete-grid geometry, PlaniSyn the semantically rich applied surface. Discharge happens in Chapter 7 through the RecPol specification (Section 7.3), the PlaniSyn grammar specification (Section 7.13), and the round-trip verification record. The notation is the form in which the standardisation schema’s predicates and the Governed Kernel Architecture’s interaction rules acquire formal expressibility, and it is the part of the suite under which manipulability admits operational test.

Objective 5 — The empirical substrate (Chapter 8). Validate the architecture empirically through the sealed v5.0 corpus of Australian residential floor plans: a complete census of 745 plans assembled in two construction strata — 572 plans in the October stratum and 173 in the August stratum — extracted under an agent-manual canonical protocol at a one-hundred-per-cent schema-validity rate at the categorisation gate. Discharge happens in Chapter 8 via the corpus-construction method, the dimensional and occurrence evidence streams, the topological and configurational streams, the integration of those streams into a single substrate claim, and the validation against held-out plans. The Governed Kernel Architecture of Objective 3 acquires here the dimensional and occurrence statistics that ground its primitives in the lived stock, and the census serves as the descriptive ecological-grounding reference distribution against which Objective 6’s demonstration trajectory is positioned along its key dimensions.

Objective 6 — The generator and the demonstration trajectory. Demonstrate operational viability through two evidence streams. The generator is a deterministic command-line Python prototype that ingests baseline-library entries and emits documentation packets, discharged in Chapter 9 (Section 9.6 procedural-generation pipeline; Section 9.12 pre-vetted-library; Section 9.19 prototype implementation against the nine FA baseline-library entries). A single-dwelling tree-branching trajectory exercises the suite end-to-end across a building life-cycle, discharged in Chapter 10 through seven dwelling states (S0–S5a, S5b) and six transformation events. The objective supplies the integrated evidence for Proposition 5 under the diachronic-trajectory measurement protocol.

The six objectives compose a single integrated claim, each carrying an independent evidential burden whose discharge is, we argue, necessary for the aim to hold. Objective 1 supplies the theoretical contract the instantiations must satisfy; Objectives 2 through 4 carry that contract down to a regulatory text, a spatial granularity, and a notation under which the governance is inspectable and verifiable; Objective 5 supplies the empirical substrate against which the commitments are tested; and Objective 6 supplies the procedural and demonstration evidence that the suite is operable end-to-end under the diachronic conditions the problem-class makes salient. Auditability appears preserved across the chain from Chapter 3’s theoretical declaration through the artefact chapters to Chapter 10’s evaluation and Chapter 11’s synthesis.

Five research questions operationalise the aim. Each binds one-for-one to a proposition of Section 3.5, asks whether a specific property of the architecture holds under operational test, and is answered by a specific evidence object emitted by an artefact-chapter and consolidated in Chapter 10. The four critical-property names from Section 2.9 and the propositional mechanisms from Section 3.5 surface together in each question, paired with the chapter at which each is settled — the formal contract that converts framing claims into examinable evidence against indicators the methodology chapter declares in advance.

RQ1 — Interoperability under semantic continuity (Proposition 1). Can a governed representational architecture make the regulatory obligations of adaptation-heavy housing more interpretable across multi-actor handover than the conventional clause-and-drawing workflow? The propositional mechanism pairs semantic interface and identity persistence; the indicator is divergence rate across handovers, measured against the current synchronous-or-bespoke interpretation workflow as comparator. Operationalisation runs through the standardisation schema’s five-dimensional decomposition of clause meaning (Chapter 5). RQ1 settles in two parts: the ambiguity-reduction part in Chapter 5 itself, and the cross-actor interpretability part in Section 10.4, where the per-state and per-event packets of the trajectory are read against the standards-interpretability measure declared in Section 4.5.

RQ2 — Transportability under bounded verification (Proposition 2). How are the boundaries of the architecture’s modules drawn so that local changes do not force global reinterpretation of the design, and is the bounded-check property maintained at materially divergent transformation scopes? The propositional mechanism is interface-bounded modularity; the indicator is the local-to-global verification scope ratio. Operationalisation runs through the Governed Kernel Architecture (Chapter 6). RQ2 settles in Section 10.10 via the modular-fit index (Chapter 10) across the seven dwelling states, where the divergent transformation scopes of the fork at S4 supply the empirical content: S5a’s typical aging-in-place preserves the S4 envelope with one module reclass, whereas S5b’s SDA-overlay-with-secondary introduces eight new modules, nine reconfigured modules, and nine new bilateral interface obligations.

RQ3 — Manipulability under replay (Proposition 3). Does the encoding (the notation) preserve the architectural commitments under round-trip parsing without information loss, such that a representation can be edited, re-emitted, and verified to be semantically and verifiably equivalent to its source? The propositional mechanism is the executable transformation grammar plus invariant checks; the indicator is replay consistency and invariant retention. The manipulability evaluation question is pre-registered at Section 4.5, and operationalisation runs through the notation (Chapter 7). Static round-trip fidelity is published in the replay evidence package (38 worked-example files, 18 trace records, zero divergence) and the invariant-test evidence package (24 invariant tests across 12 boundary pairs). The generator’s deterministic Python prototype (Section 9.19) supplies the procedural-instantiation substrate under which an independent party should re-execute the pipeline and obtain bit-identical packet content modulo emission timestamp.

RQ4 — Transformability under governed variation (Proposition 4). How does the architecture support divergent regulatory-compliance pathways from a common substrate under simultaneous occupant and regulatory change, without requiring revision of its governed kernel? The propositional mechanism is platform-governed complement evolution specified at Section 3.4; the indicator is legitimate variant diversity under shared rules. The question turns on the fork at S4, where the same one-hundred-and-sixty-five-square-metre four-bedroom envelope, module composition, and interface obligations are inherited by both branches. S5a absorbs operational-accessibility retrofits at the governed instance library under Section 6.4 Rule 4 (HLP variant of BED), serving ageing-in-place for the original occupants; S5b absorbs a full SDA-Robust overlay plus an attached secondary dwelling (LNK variant of EXT) for an adult occupant whose support needs demand it. The seven generator documentation packets read at this fork total approximately fifty-two kilobytes, and neither branch appears to require modification of the governed-kernel nine-type space taxonomy.

RQ5 — Burden under integrated diachronic measurement (Proposition 5). What burden does the architecture impose on the practitioner across a diachronic trajectory, and is that burden warranted by the gain in interpretability, transportability, manipulability, and transformability that the architecture delivers? The propositional mechanism is the composite-criterion measurement pairing a burden delta with a reliability delta against a matched-baseline workflow, under the claim that the integrated suite should hold verification burden at or below the baseline for equivalent assurance quality. Operationalisation runs in two phases: Section 4.5 declares the workflow-burden measure and the matched-baseline construction rule, and Section 10.17 reports the cumulative-burden trajectory and the burden ratio between S5a and S5b at the fork as the sub-linear-burden test.

The five questions interlock. RQ1 and RQ2 establish that the architecture’s commitments hold at substrate and module level; RQ3 establishes that those commitments transit through encoding without information loss; RQ4 establishes that the architecture absorbs simultaneous occupant and regulatory change at the governed instance library while leaving the governed kernel undisturbed; and RQ5 establishes whether the integrated burden is warranted by the cumulative gain across the prior four. We answer the questions against pre-registered indicators and falsifiers per Section 4.5, consolidated in Chapter 10, with proposition-status verdicts routed into Chapter 11 for synthesis. The contributions through which the joint reading is discharged are stated in detail at Section 1.5.

We advance one architectural claim, and we render it inspectable through five evidence-bearing instantiations whose joint discharge constitutes the contribution.⁴⁹ The claim is a governed-kernel engine — the nine-type housing-space taxonomy introduced at Section 1.1, with seven primitives and seven composites, Rule 4 variant-inheritance at Section 6.4, and boundary contracts at Section 6.3 — and we put it to operational test at the fork of state S4 in Chapter 10, where two divergent regulatory-compliance pathways are absorbed at the governed instance library from a common 165-square-metre substrate while the governed kernel remains undisturbed. Five instantiations follow, in the order in which their evidential warrant is discharged: the first supplies the design-theory licence; the second the form in which regulatory text becomes inspectable; the third the engine’s home; the fourth the verification notation; the fifth the empirical demonstration. Each pairs with a critical-property name from Section 2.9 (interoperability, transportability, manipulability, transformability) and a propositional mechanism from Section 3.5 (Propositions 1 to 5), the pairing converting each from a designed fact into a falsifiable architectural claim.

Contribution 1 — Stratified Functional Structuralism (Chapter 3)

Contribution 1 — Stratified Functional Structuralism — the design-theory licence under which the architectural claim is articulable. The first instantiation is theoretical: a synthesis fusing representational governance, modularity, and adaptation into one architectural commitment for housing systems, under which representations are treated as functional kinds realised through structural configurations over a discrete geometric substrate, with interface obligations derived from the stratification itself. Three parent literatures are integrated. The Open Building tradition supplies the capacity-providing-infrastructure commitment and the support–infill stratification that prefigures the suite’s stratified substrate.⁵⁰⁵¹ Platform-architecture scholarship supplies the visible design rules construct — the explicit, governed-kernel specification of inter-module interfaces under which legitimate complement variation may be audited at the boundary.⁵²⁵³ Standards-formalisation supplies the schema-driven discipline by which regulatory text may become queryable representation.⁵⁴ Each tradition establishes, on our reading, a distinct architectural commitment, and the joint reading we develop integrates the three into a governed-representation contract for adaptation-heavy housing under regulatory constraint. Operationalisation runs through a Type III prescriptive design theory under the Gregor-Jones anatomy, classified exaptive: modularity theory, relational graph topology, and text-as-notational-medium are each established in their home domains, and their integrated application to housing-governance representation constitutes a new application class.⁵⁵⁵⁶ The synthesis names seven constructs, five principles of form and function, three boundary conditions, and the five testable propositions (Propositions 1 to 5). Its scope is mid-range: a prescriptive claim advanced for housing adaptation and related change-intensive domains. Discharged in Chapter 3, Sections 3.2–3.5.

Contribution 2 — The standardisation schema (Chapter 5)

Contribution 2 — The standardisation schema (primary instantiation of Proposition 1) — the form in which regulatory text becomes inspectable as the architecture’s input.⁵⁷ The second instantiation is a five-dimensional decomposition of normative-clause meaning — lexical ambiguity, scope ambiguity, deontic ambiguity, role ambiguity, deontic strength — operationalised as a queryable serialisation of regulatory text together with a stratified mapping vocabulary that binds clauses to spatial primitives and composites. Compliance standards admit principled standardisation under a bounded-interdependence commitment, so that ambiguous and fragmentary regulatory prose may be lifted into a substrate whose interface obligations are explicit and whose change-impact appears locally containable. Proposition 1 is interoperability under semantic continuity, and the five dimensions supply the mechanism’s operational form: divergence between actors interpreting the same clause is bounded and inspected along each axis, then tested against calibrated coverage and ambiguity scorecards. The spatial counterpart, the stratified vocabulary, fixes the bounded units to which the standardisation schema’s predicates attach and supplies the architectural-granularity warrant that the Governed Kernel Architecture then discharges in spatial-rule form. Discharged in Chapter 5 (Sections 5.4–5.12), with supporting evidence at the ambiguity scorecard and the predicate-coverage register. Bounded: external multi-rater inter-rater-reliability evaluation is future work, and cross-corpus generalisation of the vocabulary beyond the present design corpus is future work.

Contribution 3 — The Governed Kernel Architecture (Chapter 6)

Contribution 3 — The Governed Kernel Architecture (primary instantiation of Proposition 2) — the engine’s home. The third instantiation is where the architectural claim of the framing paragraph is built: a stratified vocabulary of housing module-types — seven schematic primitives and seven elaborated composites — with formal interaction rules and constituent-element specification, instantiated in a nine-of-twenty-five Fully Accessible (FA) baseline-library bundle that supplies the suite’s authoring substrate. The commitment specified here is the engine itself — the governed-kernel nine-type space taxonomy (Section 6.2), together with the primitive–composite vocabulary, Rule 4 of Section 6.4 (life-cycle functional variation absorbed at the governed instance library without disturbing the governed kernel), and boundary contracts at Section 6.3 (cross-module compositional consistency). Four levers operate without disturbing the governed kernel, and the absorption capacity this affords is what Chapter 10’s fork at S4 (the fifth instantiation) is built to demonstrate. Contribution 3 instantiates Proposition 2 (transportability under bounded verification): declared interfaces convert broad reconstruction into local checking plus rule-governed escalation. The Governed Kernel Architecture also supplies the empirical substrate’s grounding axis, since the dimensional, occurrence, topological, and configurational evidence of Chapter 8 is computed against the scheme’s primitives, exposing the granularity claim to falsification against the 745-plan substrate. Discharged in Chapter 6 (Sections 6.2–6.5, with Section 6.2 nine-type taxonomy, Section 6.3 boundary contracts, and Section 6.4 module-library specification including Section 6.4 Rule 4 named). Bounded: only the nine FA-category baseline-library entries are authored at full schema depth, while the UD, MOD, RB, and LV entries are authored on-demand for Chapter 10’s trajectory.

Contribution 4 — The notation (Chapter 7)

Contribution 4 — The notation (primary instantiation of Proposition 3) — the notation under which the architectural claim is verifiable. The fourth instantiation is the planimetric notation: a context-free grammar, PlaniSyn, layered over the RecPol formal core and designed to encode the standardisation schema with verifiable round-trip fidelity. The notation is the suite’s encoding contract — the form in which the standardisation schema’s predicates and the Governed Kernel Architecture’s interaction rules acquire formal expressibility, and the form under which manipulability admits operational test. Proposition 3 is manipulability under replay, and formal expressibility here takes the operational form of executable transformation logic — a grammar whose named transformations, preserved invariants, and admissibility conditions admit deterministic inspection and replay. This executable-transformation-grammar reading is what Contribution 4 develops, extending the static-notation reading long served by the field into a form whose transformations can be replayed and checked. Two evidence packages discharge Proposition 3 directly. The first records deterministic round-trip replay across 38 worked-example files with 18 trace records and zero divergence. The second records 24 invariant tests across 12 boundary pairs, demonstrating the parser’s conformance gates against ill-formed inputs. Discharged in Chapter 7 (Sections 7.1–7.20), with the full grammar at the PlaniSyn appendix and the formal core at the RecPol specification. Bounded: independent third-party Goodman-criteria verification — syntactic disjointness, semantic finite differentiation, replete unambiguity — is future work, and internal verification through replay and invariant testing is the present warrant.

Contribution 5 — The empirical substrate, the generator, and the demonstration trajectory (Chapters 8 to 10)

Contribution 5 — The empirical substrate, the generator, and the demonstration trajectory (Chapters 8 to 10; primary instantiation of Proposition 4 and integrated instantiation of Proposition 5) — the empirical locus where the architectural claim is put to test at the fork. The fifth instantiation spans three discharges. The empirical substrate is the sealed v5.0 corpus: a complete census of 745 Australian residential floor plans assembled in two construction strata (572 plans in the October stratum and 173 in the August stratum), extracted under an agent-manual canonical protocol at a one-hundred-per-cent schema-validity rate. The generator is a deterministic command-line Python prototype that ingests baseline-library entries and emits documentation packets in conformance with the handoff-contracts specification. The demonstration trajectory is the single-dwelling tree-branching sequence in Chapter 10: seven dwelling states (S0 minimum-viable baseline through S5a typical-aging and S5b SDA-overlay-with-secondary), six transformation events, seven documentation packets totalling approximately 52 KB, and the fork at S4 where divergent regulatory-compliance pathways are absorbed at the governed instance library from a common substrate. Human stake enters at the fork: one branch accommodates typical ageing-in-place for the original occupants, while the other accommodates an adult occupant whose support needs require SDA-Robust provisioning across a secondary attached dwelling, and the same architectural core absorbs both. Proposition 4 (transformability under governed variation) is discharged through this fork, where the same 165-square-metre four-bedroom envelope, the same module composition, and the same interface obligations are inherited by both terminal states: S5a absorbs operational-accessibility retrofits at the governed instance library under Section 6.4 Rule 4 (HLP variant of BED), and S5b absorbs a full SDA Design Standard 2019 Robust overlay plus an attached secondary dwelling — a design-category transition, eight new modules, nine reconfigured modules, nine new bilateral interface obligations, and a multi-dwelling site composition under Section 6.3 boundary contracts — also at the governed instance library (LNK variant of EXT), with the secondary-dwelling tenancy declaration governed under Rule 4. Neither branch requires modification of the governed-kernel nine-type taxonomy. Proposition 5 (burden under integrated diachronic measurement) is discharged through the cumulative-burden trajectory at Section 10.17, where the fork’s burden grows with the scope of the regime change on two registers that Section 10.17 keeps apart: the measured documentation-emission cost rises about 1.4 times, from the S5a packet (7,736 bytes) to the S5b packet (10,797 bytes), while the matched-task practitioner-day baseline, reasoned event-by-event, rises roughly three-and-a-half-fold — an analytic estimate resting on no practitioner cohort (n = 0), not measured field data. Both are descriptive readings of a single dwelling’s trajectory rather than statistics over the census, and they support the burden-proportional-to-scope claim committed to in Chapter 4 Section 4.5 descriptively rather than as direct empirical confirmation. Discharged jointly in Chapter 8, Chapter 9, and Chapter 10 (Sections 10.2, 10.3, 10.4–10.17). Bounded: a single dwelling-and-lot, with population-scale generalisability awaiting cross-corpus and multi-trajectory validation, and third-party usability evaluation against a non-specialist user population future work in Chapter 9 Section 9.27.

Interlock — the architecture is the claim; the five instantiations are its evidence

The architecture is the claim; the five instantiations are its evidence-bearing discharges, and the aim of Section 1.4 holds only when the five hold together. Each carries an independent evidential burden whose discharge appears necessary for the architectural claim — that a governed representational architecture for adaptation-heavy housing systems may be constructible, instantiable, operable, and auditable under the diachronic conditions the problem-class makes salient — to hold, and the dependence runs in both directions: Contribution 1 standing alone would be theory awaiting instantiation; Contributions 2 through 4 would be designs awaiting empirical warrant; Contribution 5 alone would supply a corpus and a pipeline lacking a governing architecture. The audit chain makes the joint discharge inspectable end to end, running from the SDA Design Standard’s normative clauses, through the standardisation schema’s serialisation, through the Governed Kernel Architecture’s spatial primitives, through the PlaniSyn-encoded representations of the trajectory’s seven states, through the generator’s deterministic emission of seven packets, to the trajectory’s verdict against the four critical properties plus burden in Chapter 10 and the synthesis in Chapter 11. The integration of representational governance with modularity, and of adaptation with verification, under one architectural commitment expressed through five evidence-bearing instantiations and validated against a sealed substrate and a tree-branching trajectory, is, on our reading, the substantive contribution. Scope, delimitations, and the chapter-by-chapter map of the thesis follow.

The scope contract under which Section 1.5’s five contributions are advanced runs in two registers, both of them commitments: the in-scope register declares the demonstrations against which the suite remains auditable, and the out-of-scope register declares where later work begins, so that the limits it records leave the contribution’s claims intact when those claims are read with the scope clauses in view. This is the contract under which the Chapter 10 demonstration is reported and against which the examiner is invited to judge the contribution, and twelve chapters then compose the structural form within which it is discharged.

In-scope — what the inquiry undertakes to demonstrate. Four claims compose the in-scope register, each bounded by the evidential remit declared below and each discharged in the chapters that follow.

In-scope: the domain. The inquiry is bound to adaptation-heavy housing in the Australian regulatory context. Three instruments anchor the substrate. The NDIS Specialist Disability Accommodation Design Standard is the primary exemplar, selected because it concentrates the diagnostic conditions of representational governance into a single auditable text;⁵⁸ the National Construction Code Volume 2 supplies the baseline legal substrate against which the trajectory dwellings are deemed code-compliant;⁵⁹ and AS 1428.1 supplies the dimensional benchmarks against which the schema’s predicate coverage is calibrated.⁶⁰ Liveable Housing Australia design guidelines and state-level disability-housing regulation are referenced where they materially shape the SDA reading.

In-scope: suite coverage. Five artefacts compose the suite — design, specification, encoding, empirical validation, and procedural instantiation — and all five fall within the present evidential remit. The standardisation schema is validated against the full SDA clause-set at Chapter 5; the Governed Kernel Architecture’s Fully Accessible baseline-library is populated to nine entries at full schema depth at Chapter 6 Section 6.4; the two-layer notation is exercised across 38 worked-example files at Chapter 7; the empirical substrate is validated against the 745-plan corpus at Chapter 8; the generator — the deterministic command-line Python prototype — runs at Chapter 9; and end-to-end exercise occurs through the Chapter 10 trajectory. The five compose: each is necessary, none sufficient alone, and the integration itself is evaluated.

In-scope: the empirical substrate. The sealed v5.0 corpus is the inquiry’s ecological-grounding instrument — a complete census of 745 Australian residential floor plans assembled in two construction strata (572 October, 173 August), extracted under an agent-manual canonical multimodal-vision protocol at a one-hundred-per-cent schema-validity rate.⁶¹ Plans outside the sealed census fall outside the present remit. The corpus serves a dual function: it is the descriptive dimensional reference for the Chapter 10 trajectory, and the substrate for the Chapter 8 evidence streams.

In-scope: the demonstration. A single-dwelling tree-branching trajectory of seven dwelling states and six transformation events exercises the suite. The trajectory begins at S0 — a minimum-viable baseline at approximately 58m² on a 15×30m suburban lot — and proceeds through four trunk events (occupancy commitment; family-formation; family-growth with bathroom upgrade; empty-nest) to S4, where it forks: S5a applies a typical aging-in-place adaptation, and S5b applies an SDA-Robust overlay plus an attached secondary dwelling for an adult occupant. The trajectory exercises the interoperability, transportability, and manipulability propositions (Propositions 1, 2, and 3) at every state through round-trip serialisation; the transformability proposition (Proposition 4) across the six events and the fork; and the integrated-burden proposition (Proposition 5) cumulatively.⁶²

Out-of-scope — declared limitations. Eight delimitations bound the work explicitly, each paired with the chapter at which the corresponding scope-limit is articulated.

Out of scope: cross-corpus generalisation of the stratified vocabulary beyond the sealed census lies outside the warrant; Chapter 6 Section 6.1 tests cross-partition only, and cross-corpus comparison against AS 1428.1, ISO 21542, BS 8300, or LHDG is registered as future work.

Out of scope: independent third-party Goodman-criteria verification of the notation (syntactic disjointness, semantic finite differentiation, replete unambiguity) lies outside the present scope;⁶³ internal verification through round-trip replay across 38 worked-example files and parser-rejection invariants supplies the present warrant. External verification is registered as future work.

Out of scope: baseline-library population depth beyond the Fully Accessible category: nine Fully Accessible entries are populated to full schema depth, while Universal Design and Modification entries are authored on-demand at the depth that Chapter 10’s cases require. The bound reflects the tiered population verdict of the submission pathway, and full Universal Design and Modification population is future work.

Out of scope: external multi-rater inter-rater-reliability evaluation of the standardisation schema’s five dimensions and stratified vocabulary is registered as future work. Present calibration is candidate-as-domain-expert self-consistency rating, reported at Chapter 5 Section 5.12.

Out of scope: the SDA-Robust governed-kernel scope-limit at branch S5b: the SDA-Robust overlay is authored at full library-entry depth only for the kitchen and external-circulation modules, with the remaining Robust-category qualities (sanitary, bedroom, living, service, entry, circulation, and dwelling) inferred from Fully Accessible counterparts through the Rule 4 variant-governance protocol at Section 6.4. Rule-4-inferred qualities suffice to discharge S5b while leaving the governed-kernel contracts intact, and full Robust baseline-library population across all nine module types is future work.

Out of scope: production-grade tooling lies outside present scope. The generator is a command-line research instrument; practitioner-toolchain integration, multi-user concurrency, persistent storage, version-controlled library curation, GUI authoring, BIM-authoring-tool integration via the documentation-packet schema, rule-checking-engine integration along the Hjelseth-Nisbet pattern,⁶⁴ and LLM-assisted authoring layers above the deterministic pipeline are declared future-work at Chapter 9 Section 9.27.

Out of scope: post-submission revisions, viva preparation, and publication-stream papers lie outside present scope. The unit of contribution is the work as submitted, and downstream knowledge-transfer activities are flagged for the post-submission programme.

Out of scope: the synthetic-but-empirically-grounded character of the Chapter 10 trajectory dwellings: the seven states are authored synthetic dwellings whose key dimensions sit within the corpus’s descriptive distribution, and they are not field-extracted instances. S0’s deliberate low-tail anchoring — approximately 58m² against a corpus median of approximately 98m² — is defended as a methodological choice that supports transformation-event demonstrability. Population-scale generalisability through field-extracted longitudinal dwelling histories is future work.

Scope closure. The implication is a deliberately narrowed evidential surface. The contribution is the architecture and its auditable instantiation within the regulatory exemplar and the empirical substrate against which it was built; universal generalisation beyond corpus and exemplar is future work. The design-theoretic commitments — stratification, bounded interdependence, governed interfaces, and round-trip verifiability — are advanced as transferable principles, and their transfer to other domains, other regulatory regimes, and other empirical substrates is the task of subsequent inquiry.

Thesis structure. Twelve chapters compose the inquiry, organised into four argumentative arcs — framing (Chapters 1–4), artefactual contribution (Chapters 5–9), demonstration-and-synthesis (Chapters 10–11), and closure (Chapter 12) — and the sequence tracks the design-science research logic by which problem, theory, method, artefact, evaluation, and reflection are sequenced into a single auditable evidence chain.⁶⁵⁶⁶ Each chapter discharges a specific evidential burden, declared at the chapter’s opening and closed against the contribution claims of Section 1.5. The structure is deliberately additive: each artefactual chapter consumes the deliverables of its predecessors and supplies the inputs of its successors, so the evidence chain reads forward without back-channel dependencies.

Chapter 1 — Introduction. Frames the modular architecture for adaptation-heavy housing — Baldwin and Clark’s publish/hide partition transposed to the dwelling — and the problem class it answers, representational governance under adaptation pressure; locates the contribution-relevant gap at the intersection of four reading-frames in the literature; states the aim with six objectives and five research questions; declares five contributions and their interlocking evidential burden; bounds the scope of the inquiry; and maps the structure of the work.

Chapter 2 — Literature Review and Theoretical Framework. Traces five reading-frames — housing as a person-environment fit problem, the supply-side architecture of change, bounded optionality and dimensionality reduction, modularity as bounded interdependence, and representational governance as a design problem — each supporting an architectural commitment that the suite presupposes; develops the four critical-property names at Section 2.9 (interoperability, transportability, manipulability, transformability).

Chapter 3 — Theory: Modularity and Complexity. Builds the theoretical scaffold and produces Contribution 1 (Stratified Functional Structuralism); develops wickedness as emergent from complex-adaptive dynamics (Section 3.2), modularity as the explanatory frame for interface-governed change (Section 3.3), platform architecture as the system-scale operationalisation that supplies Baldwin-Clark’s visible-design-rules construct (Section 3.4), and the Gregor-Jones design-theory contract under which the five propositional mechanisms, Propositions 1 to 5, are stated (Section 3.5).

Chapter 4 — Methodology. Commits the inquiry to design-science research as paradigm; specifies the DSR cycles that govern construction and evaluation; declares the evaluation strategy by which Propositions 1 to 5 are tested; documents the data-collection and pipeline architecture under which the sealed corpus is built; addresses validity, reliability, ethics, and reproducibility.

Chapter 5 — A Queryable Schema for Accessibility Standards. Produces Contribution 2, the standardisation schema; presents the five-dimensional decomposition of normative-clause meaning; specifies the serialisation pipeline; evaluates ambiguity reduction and predicate coverage against the calibrated scorecards.

Chapter 6 — A Governed Kernel Architecture for Housing. Produces Contribution 3, the Governed Kernel Architecture, and is the engine’s home: the governed-kernel nine-type space taxonomy at Section 6.2, the stratified vocabulary (seven primitives and seven composites), Rule 4 variant inheritance at Section 6.4, and the boundary contracts at Section 6.3 are specified here; the Fully Accessible baseline-library is populated to nine entries at full schema depth.

Chapter 7 — A Formal Notation for Floor Plans. Produces Contribution 4, the notation; specifies RecPol as the formal core for discrete-grid geometry, PlaniSyn as the applied grammar layered over RecPol, and the round-trip evidence across 38 worked-example files (deterministic round-trip replay: 18 trace records, zero divergence; conformance-gate invariants: 24 invariant tests across 12 boundary pairs).

Chapter 8 — Evidence from a Census of Australian Floor Plans. Produces the empirical half of Contribution 5, the empirical substrate; constructs the sealed corpus (745 plans, 100% schema validity); presents dimensional, occurrence, topological, and configurational evidence streams; integrates them into a single substrate claim and validates it against held-out plans.

Chapter 9 — Generating Documented Dwelling Variants. Produces the procedural half of Contribution 5, the generator; diagnoses the documentation bottleneck the suite addresses; presents the four-stage ingest-validate-transform-emit pipeline; specifies the pre-vetted-library concept that joins the standardisation schema, the Governed Kernel Architecture, and the notation operationally; reports the prototype implementation against the nine Fully Accessible baseline-library entries.

Chapter 10 — Demonstration and Evaluation. Tests the suite end-to-end through the seven-state trajectory; specifies the canonical trunk at Section 10.2 and the fork at S4 at Section 10.3 — the locus at which the architectural claim is put to test, where two divergent regulatory-compliance pathways are absorbed at the governed instance library from a common substrate without disturbing the governed kernel; reports three results streams — standards interpretability (Section 10.4), modular fit via the modular-fit index (Section 10.10), and workflow efficiency (Section 10.17) — and assigns proposition-status verdicts to each of Propositions 1 to 5.

Chapter 11 — Discussion. Synthesises the contributions, with Section 11.2 developing the synthesis under which the single architectural claim is borne out through five evidence-bearing instantiations; draws theoretical and practical implications; declares limitations in honest dialogue with the scope contract above; identifies a future-work programme aligned with the delimitations registered here.

Chapter 12 — Conclusion. Summarises the argument as a whole; restates the validated contribution claims — the single architectural claim borne out through five evidence-bearing instantiations — with explicit pointers to the supporting evidence; advances final recommendations to regulators, design practitioners, the research community, and policy analysts; offers concluding remarks on the trajectory of the research programme.

The twelve chapters together compose a single audit trail running from the regulatory source to the validated artefact suite. The four arcs operate as nested contracts: the framing arc fixes the contract under which the artefactual arc is warranted; the artefactual arc supplies the deliverables that the demonstration arc exercises; the demonstration arc produces the proposition-status verdicts that the synthesis arc interprets; and the closure arc restates the validated position against the framing arc’s original claims. At each chapter boundary the inputs and outputs are declared; at each contribution claim the evidence pointer is explicit. The literature review and theoretical framework follow.