Skip to content
modumatics Modular Infrastructure for Inclusive Housing Tran Thien Toan Ngo · PhD Dissertation

Front Matter

Title Page

Modular Infrastructure for Inclusive Housing

Ground-up Application of a Modular Framework to Improve Housing for Ability-Diverse Living


Tran Thien Toan Ngo

Student ID 22148804


A thesis submitted in fulfilment of the requirements for the degree of

Doctor of Philosophy


Centre for Infrastructure Engineering

School of Engineering, Design and Built Environment

Western Sydney University


Principal Supervisor: Dr Pejman Sharafi, Associate Professor

Co-Supervisor: Dr Jun Wang, Senior Lecturer



Submitted in candidature for the degree of Doctor of Philosophy

Abstract

Households whose abilities differ — with age, with disability, with the changing membership of a life-stage — make demands on their housing that shift across a building’s long life. Most will be housed by stock that already stands, so the binding question is how much change an existing dwelling can absorb. That capacity is limited, on our reading, as much by a dwelling’s representational infrastructure — its drawings, schedules, certificates, and the regulatory texts they answer to — as by its physical fabric: a change meant to stay local repeatedly forces global re-checking across every commitment that holds the dwelling together. Keeping a local change local — a problem of representational governance — is what this thesis treats as a design-science problem and answers with a modular architecture for the dwelling.

The response transposes two lineages onto housing. From the study of complex systems — Simon, Parnas, Baldwin and Clark — it takes the principle that change stays tractable when interdependence is bounded by stable interfaces that publish what others may rely on and hide what lies within; from Open Building, the layered dwelling whose long-life support carries a renewable short-life infill. Joining them yields the thesis’s central artefact, a Governed Kernel Architecture: a governed kernel — a small set of design rules, nine functional space-types over a stratified vocabulary of seven primitives and seven composites, a variant-inheritance rule, and explicit interface contracts — around which life-cycle variation is absorbed in a bounded governed instance library. The kernel is governed rather than frozen: fixed for any given configuration, yet revisable through a rule-bound, versioned path. The architecture is instantiated as a coordinated artefact suite: a design theory; a schema that renders regulatory prose queryable; the module system that houses the kernel; a planimetric notation that can be written and read back without loss of meaning; a generation-and-documentation pipeline; and an empirical substrate of 745 Australian floor plans. Specialist Disability Accommodation under Australia’s National Disability Insurance Scheme serves as the proving ground, not the subject, because it concentrates the difficulty into one auditable instrument.

The decisive test sits at a single fork: one four-bedroom dwelling branches two ways — the original occupants ageing in place, or a higher-support refurbishment with an attached secondary dwelling — and both, on our reading, are absorbed without the governed kernel moving. A representation that travels intact between actors, and one that confines the effects of a local change, appear well supported; governed variation under shared rules holds at moderate confidence within the single branch examined; faithful replay of authoring acts, and net benefit across a dwelling’s life, remain only partly established. Wider rater studies, further trajectories, other building stocks and regulatory regimes, and a production implementation would extend the warrant.

Keywords: modular architecture and design rules; governed kernel; housing adaptation; representational governance; Open Building; planimetric notation; design science research.

Statement of Originality

Statement of Authentication

The work presented in this thesis is, to the best of my knowledge and belief, original except as acknowledged in the text. I hereby declare that I have not submitted this material, either in full or in part, for a degree at this or any other institution.


………………………………………………..

Tran Thien Toan Ngo

Date:

Acknowledgements

Acknowledgements

I am grateful to my principal supervisor, Associate Professor Pejman Sharafi, and my co-supervisor, Dr Jun Wang, whose guidance, patience, and critical reading shaped this work at every stage. Their willingness to let the argument develop on its own terms, while holding it to account, made the thesis possible.

I thank the Centre for Infrastructure Engineering and the School of Engineering, Design and Built Environment at Western Sydney University for the resources, candidature support, and research environment that sustained this project.

Any errors that remain are my own.

Table of Contents

Front Matter

Background

Main Body

Bibliography

The thesis uses inline footnote citations throughout the prose; full reference entries are aggregated at the end of the manuscript and cross-referenced from each footnote callout.

List of Figures

The list of figures inventories the figure callouts harvested from chapter prose. Figures carry no numbers: each entry pairs the figure title with a wikilink to the host section module that renders it (or, where a standalone figure module is still load-bearing, to that module).

Chapter 2 — Literature Review and Theoretical Framework

  • Person-environment fit: competence-press dynamics — view
  • Lawton-Nahemow zone model: competence-press space with blocked adjustment — view
  • Separation of concerns: monolithic vs layered systems — view
  • Shearing layers of the built environment (layer lifetimes) — view
  • Nominal to effective optionality funnel — view
  • Six conditions for manageable change — view
  • Chapter 2 argument map — view
  • Gregor-Hevner knowledge contribution framework — view
  • Hevner’s DSR three-cycle view — view
  • Meta-requirements traceability network — view
  • Near-decomposability: block-diagonal interaction structure — view
  • Containing vs transmitting interface behaviour — view
  • Life-course trajectory divergence — view
  • Permutation explosion: independent vs coupled decision variables — view
  • Three interface types in housing adaptation — view
  • Burden accumulation across interface sequence — view

Chapter 3 — Theory: Modularity and Complexity

  • Chapter 3 theory map — view
  • CAS feedback loops in the housing adaptation domain — view
  • Modularity constructs: interface-governed interaction density — view
  • Schematic Model of a Module — view
  • Platform architecture: from requirements to verification — view
  • Design-theory scaffold (after Gregor and Jones 2007) — view

Chapter 4 — Methodology

  • DSR cycle architecture — view
  • Four-layer data pipeline architecture — view
  • Evaluation traceability map — view
  • Thesis alignment with Hevner’s three-cycle DSR framework — view
  • Validity threat matrix — view

Chapter 5 — A Queryable Schema for Accessibility Standards

  • Five-layer standardisation schema architecture — view
  • Five-stage serialisation pipeline — view
  • Resolution rate profile by dimension — view
  • Resolution rate by clause class — view
  • Progressive schema layer contribution — view
  • External benchmarking coverage — view
  • Primitive ablation information loss — view
  • Threshold sensitivity: polysemy ratio across frequency cutpoints — view
  • Cross-channel fragmentation: text–figure normative coverage overlap — view
  • Design Science Research iteration history for the standardisation schema — view
  • Deontic force decomposition: three-path honest reframing — view
  • Strongly coupled primitive co-occurrence pairs — view
  • Literature positioning: four-stream intersection of the standardisation schema — view
  • SDA foundational primitive taxonomy — view
  • Artefact suite dependency chain with handoff contracts — view

Chapter 6 — A Governed Kernel Architecture for Housing

  • The Governed Kernel Architecture: three layers and the governed instance library — view
  • The stratified vocabulary: seven primitives, seven composites, ten operators — view
  • Twelve-pair boundary network with interaction differential heatmap — view
  • Verification sequence DAG — view

Chapter 7 — A Formal Notation for Floor Plans

  • Two-layer notation architecture — view
  • Cell taxonomy by free-edge count — view
  • Three-plane stratification with entity types and verb families — view
  • Local and global grids across Shape, Form, and Instance — view
  • RecPol EBNF production rules and LL(1) decision points — view
  • RecPol syntax by example: coordinates, joints, and free-edge counts — view
  • Raw and canonical form with the canonicalisation operator — view
  • Composite shape construction (worked example, S1) — view
  • Composite construction (abstract concept) — view
  • Lateral attachment of two distinct shapes via edge selection — view
  • Reflective self-attachment with the canonicalisation operator — view
  • Lanes and permeability flags as receptacle-prefix components — view
  • Permeability matrix: conditions in which two receptacles may overlap — view
  • Object exclusivity cases: disjoint, adjacent, and overlapping configurations — view
  • Instance placement walk-through (faithful worked example, RecPol syntax) — view
  • Instance placement (clean conceptual: Place, Mirror, Rotate) — view
  • PlaniSyn three-layer architecture with tag types and interaction mapping — view
  • Verification sequence constraint encoding (HC-6C) — view

Chapter 8 — Evidence from a Census of Australian Floor Plans

  • Geographic and market coverage of the census’s two strata (August and October) — view
  • Corpus construction data flow — view
  • Dimension frequency distribution with Gaussian-mixture-model components — view
  • 4D composite-coverage curve for module candidate selection — view
  • Soft and hard coverage profiles across module candidate range — view
  • Coincidence family network at the 25 mm module — view
  • Space-category connectivity frequency heatmap — view
  • Space-category adjacency coupling network — view
  • Hard- and soft-constraint adjacency segregation subgraph — view
  • Generator constraint funnel: from all polyominoes to feasible packings — view
  • D4 symmetry group and canonical form for polyomino deduplication — view
  • Representative arrangement permutations from the generator — view
  • Representative packing permutations showing port and attachment semantics — view
  • Micro-to-meso module LCM staircase hierarchy — view
  • the empirical substrate to the generator handoff contracts — view

Chapter 9 — Generating Documented Dwelling Variants

  • Procedural generation pipeline stages — view
  • the generator prototype: end-to-end run flow — view

Chapter 10 — Demonstration and Evaluation

  • Trajectory tree (S0 → S4 → {S5a, S5b}) — view
  • State progression — footprints (S0 to S4) — view
  • Fork path comparison (S5a vs S5b) — view
  • Standards-interpretability per-state profile — view
  • Modular fit (Module-Inheritance Ratio) per-state metrics — view
  • Per-event workflow burden across the trajectory — view
  • Evaluation surface (cross-state EM-09-01..04 composite, 2×2 panel grid) — view

Appendices

The appendix-data and supplementary-context appendices contain illustrative figures that are not numbered into the main-text figure sequence. See the appendix data index for the full appendix inventory. Appendix figures (when numbered) follow the “Appendix Figure A.N” convention to distinguish them from main-text figures.

  • Appendix Figure: Cross-chapter figure-traceability matrix — view

List of Tables

The list of tables inventories the tables harvested from chapter prose. Tables carry no numbers: each entry pairs the table title with a wikilink to the source module. Where a table is rendered through a dedicated atomic module under the tbl_ prefix scheme, the link is to that module; where the table is inline within a section file, the link is to the host section.

Chapter 2 — Literature Review and Theoretical Framework

  • Research questions, artefacts, and meta-requirements (problem-class to designed-response mapping) — view

Chapter 3 — Theory: Modularity and Complexity

  • Principles of form and function (design-theory anatomy) — view
  • Meta-requirement to proposition adjacency mapping (MR1–MR5 × P1–P5) — view
  • Proposition contract matrix (mechanism, indicator, baseline, evidence, falsifier) — view

Chapter 4 — Methodology

  • Requirements–Design–Evaluation Traceability Matrix — view

Chapter 5 — A Queryable Schema for Accessibility Standards

  • Chapter 4 abstract measures mapped to the standardisation schema operationalisations — view
  • Per-dimension resolution profile (identity, structural, deontic, lexical, mapping) — view
  • Per-clause-class resolution rate scorecard — view
  • Deontic force decomposition: coverage gap, force collapse, and force preservation pathways — view
  • Schema-variant ablation: aggregate comparison delta across null, minimal, and full variants — view
  • Layer-by-layer marginal-gain decomposition — view
  • Threshold sensitivity: polysemy ratio across frequency cutpoints — view
  • Cross-channel validation summary (text-figure normative coverage overlap) — view
  • Failure mode to design principle to schema layer mapping — view
  • Ambiguity-treatment categories in prior standards-formalisation work — view

Chapter 7 — A Formal Notation for Floor Plans

  • Governed-kernel stratified vocabulary (seven primitives, seven composites, ten operators) mapped to PlaniSyn syntactic forms — view
  • Pre/Post/Effect specification for the four PlaniSyn interaction types — view
  • PlaniSyn interaction-type coverage of the three governed-kernel interface types (HC-6D, NR-014) — view

Chapter 10 — Demonstration and Evaluation

  • Evidence-object register: primary (matrix-declared) and inline-synthesised identifiers exercised across the trajectory — view
  • Per-state corpus position against the general-residential cohort distribution — view
  • State-by-state interpretability profile across the four sub-criteria — view
  • Per-state modular-fit profile across the three components — view
  • Per-state module-inheritance ratio — view
  • Per-event workflow burden across the trajectory’s seven dwelling states — view
  • Synthesis matrix: result domain → measure → evidence object → proposition → adjudication — view

Chapter 12 — Conclusion

  • Final recommendations (audience → recommendation → evidence base) — view

List of Abbreviations

The list of abbreviations consolidates the acronyms and short forms used across the thesis prose. Each entry pairs the abbreviation with its expansion, a brief operational definition, and the chapter in which it is first introduced. For full conceptual definitions, see the Glossary; for reference targets such as evaluation measures, requirements, and propositions, see the Evaluation Workbench and the Requirements Register.

Regulatory and policy

  • ABCB — Australian Building Codes Board. Joint Commonwealth–State authority responsible for the National Construction Code. Ch1 §1.2.
  • AS — Australian Standard. Standards Australia document series; the AS 1428 series governs accessibility design. Ch1 §1.2; Ch5.
  • AS 1428 — Australian Standards series for design for access and mobility (Parts 1–4 cited where relevant). Ch1 §1.2; Ch5 §5.4.
  • DSS — Department of Social Services (Cth). Policy authority for disability programmes that interact with the SDA regime. Ch1 §1.2.
  • NCC — National Construction Code. The unified building-code instrument for Australia, revised on a three-year cycle. Ch1 §1.2.
  • NDIA — National Disability Insurance Agency. The Commonwealth agency administering the NDIS, including SDA technical guidance. Ch1 §1.2.
  • NDIS — National Disability Insurance Scheme. The Australian Commonwealth scheme established by the National Disability Insurance Scheme Act 2013 (Cth). Ch1 §1.1.
  • SDA — Specialist Disability Accommodation. The NDIS-funded built-environment regime governed by the SDA Design Standard. Ch1 §1.1.

SDA design categories

  • FA — Fully Accessible. SDA Design Standard category for occupants with significant physical impairment. Ch1 §1.4; Ch5–Ch6.
  • HPS — High Physical Support. SDA Design Standard category for occupants with very high support needs. Ch5; Ch10.
  • IL — Improved Liveability. SDA Design Standard category for occupants with sensory, intellectual, or cognitive impairment. Ch5; Ch10.
  • RB — Robust. SDA Design Standard category for occupants requiring resilient construction. Ch5; Ch10.

Modularity-scheme primitives (Chapter 6)

  • BED — Bedroom (sleeping) module type within the governed-kernel nine-type taxonomy. Ch6 §6.3.4.
  • BR1, BR2, … BRn — Instance-suffix notation for BED-class module instances within a single dwelling (BR1 is the first BED-class instance, BR2 the second, and so on). Used in Ch10 demonstration prose where each instance must be referred to by ordinal identity (e.g., the master bedroom BR1 versus the secondary bedroom BR2). BR is not a separate module type; every BRn instance is a BED-class instance under the canonical nine-type taxonomy. Ch10 §10.3–§10.4.
  • BA1, BA2, … BAn — Instance-suffix notation for SAN-class module instances within a single dwelling (BA1 is the first SAN-class instance, typically the master ensuite or principal bathroom; BA2 the second). Used analogously to BRn in Ch10 demonstration prose. BA is not a separate module type; every BAn instance is a SAN-class instance under the canonical nine-type taxonomy. Ch10 §10.3–§10.4.
  • CIR — Circulation module type. Ch6 §6.3.4.
  • DIN — Dining variant tag (LIV variant under §6.4 Rule 4, for dining-zone configurations). Ch6 §6.4; Ch10.
  • DWL — Dwelling-envelope module type. Ch6 §6.3.4.
  • ENT — Entry module type. Ch6 §6.3.4.
  • EXT — External (site/secondary-dwelling) module type. Ch6 §6.3.4.
  • GKA — Governed Kernel Architecture; the three-layer architecture of Chapter 6 (Layer 1 Generative Grammar and Layer 2 Modular Contract System together form the governed kernel; Layer 3 is the Governed Instance Library). Ch6.
  • HLP — Helper/carer variant tag (BED variant under §6.4 Rule 4, for episodic helper or carer occupancy). Ch6 §6.4.
  • KIT — Kitchen module type. Ch6 §6.3.4.
  • LIV — Living module type. Ch6 §6.3.4.
  • LNK — Link variant tag (EXT variant under §6.4 Rule 4). Ch6 §6.4.
  • LV — Improved Liveability design category (auxiliary scope). Ch6 §6.3.
  • MOD — Modular design category (auxiliary scope). Ch6 §6.3.
  • OFC — Office variant tag (LIV variant under §6.4 Rule 4, for home-office or workspace configurations). Ch6 §6.4; Ch10.
  • SAN — Sanitary (bathroom/WC/laundry) module type. Ch6 §6.3.4.
  • SVC — Service (utility/storage) module type. Ch6 §6.3.4.
  • UD — Universal Design category. Ch6 §6.3.

Notation, schema, and methodology

  • API — Application Programming Interface. Used in the generator’s handoff context. Ch9.
  • BIM — Building Information Modelling. Family of object-oriented building representations; treated as a parent literature. Ch2 §2.5; Ch5.
  • CAD — Computer-Aided Design. Generic geometric modelling tradition referenced for distinction. Ch1 §1.2; Ch2.
  • CAS — Complex Adaptive System. Theoretical framing for wicked-problem dynamics. Ch3 §3.2.
  • CBS — Composition-Boundary-Stability. The former name of the trajectory’s modular-fit measure, now the Module-Inheritance Ratio (see MIR). Ch10 §10.5.
  • CFG — Context-Free Grammar. Underlying formalism for PlaniSyn. Ch7.
  • CLI — Command-Line Interface. Form factor of the generator prototype. Ch9 §9.5.
  • CP-D5 — Corpus Population D5. The sealed 745-plan census of Australian residential floor plans (the empirical substrate; two construction strata — 572 October and 173 August). Ch4 §4.3; Ch8.
  • DSR — Design Science Research. The overarching methodological paradigm. Ch4 §4.2.
  • EBNF — Extended Backus–Naur Form. Grammar specification syntax used for RecPol. Ch7 §7.4.
  • GFA — Gross Floor Area. Dimensional measurement used in corpus and trajectory positioning. Ch8; Ch10.
  • IFC — Industry Foundation Classes. ISO-standardised BIM data schema. Ch2 §2.5; Ch5.
  • JSON — JavaScript Object Notation. Serialisation format used in the generator’s output. Ch9.
  • LCM — Least Common Multiple. Used in micro-to-meso module synthesis. Ch8 §8.8.
  • LL(1) — Top-down parser class with one-token lookahead. RecPol parser characteristic. Ch7 §7.4.
  • MIR — Module-Inheritance Ratio (formerly the Composition-Boundary-Stability Index, CBS). The proportion of modules in a trajectory state inherited unchanged from the immediate predecessor, relative to the total module count at that state; reported as the trajectory’s modular-fit measure. Ch10 §10.5.
  • PE-fit — Person–Environment fit. Theoretical framing inherited from Lawton and Nahemow. Ch2 §2.1.
  • PlaniSyn — Planimetric Syntax (the notation’s applied layer). The semantically rich notation grammar (v6.0). Ch7 §7.5.
  • RecPol — Rectangular Polyomino (the notation’s formal core). The discrete-grid geometric notation. Ch7 §7.4.
  • SFS — Stratified Functional Structuralism. The thesis’s design-theoretic position. Ch3 §3.5.
  • SVG — Scalable Vector Graphics. Figure asset format used in the publish suite. Ch1 §1.8 (figure conventions).
  • XML — Extensible Markup Language. Generic structured-text format referenced for distinction. Ch5.

Evaluation, propositions, and traceability codes

  • CW — Candidate-Work. Identifier prefix for items deferred to candidate post-submission work (e.g., Ch5-CW-04). Ch4; Ch5.
  • DDG — Decision Document. User-decision artefact referenced in governance prose. Ch4.
  • DF — Design Feature. Element of the Requirements–Design–Evaluation Traceability Matrix. Ch4 §4.4.
  • EM — Evaluation Measure. Element of the Evaluation Workbench (e.g., EM-4W-01). Ch4 §4.4.
  • EQ — Evaluation Question. Element of the Evaluation Workbench (e.g., EQ-01). Ch4 §4.4.
  • ER — Environment Requirement (environment-derived requirement; ER-01 to ER-06). The minimum problem scope the artefact suite must address, indexed by the Requirements–Design–Evaluation Traceability Matrix. Ch4 §4.4; see the Environment-Derived Requirements Register.
  • EVID — Evidence object identifier (e.g., EVID-P3-REPLAY, EVID-P3-INVARIANTS). Ch4; Ch7.
  • EXP — Experiment file. A sealed experiment record under experiments/recpol-v6/ (e.g., EXP-7.4-round-trip-fidelity.md), cited as the provenance for a notation or semantic result, including round-trip replay testing. Ch7 §7.6.
  • HC — Handoff Contract identifier (e.g., HC-6A–HC-6D for Ch6→Ch7; HC-8A–HC-8D for Ch8→Ch9). Naming the inter-chapter substrate contracts. Ch6 §6.8; Ch8 §8.8.
  • NR — Notation Rule identifier (e.g., NR-014). Ch7.
  • P1–P5 — The five testable propositions of the Stratified Functional Structuralism design theory: P1 semantic interface and identity persistence, P2 interface-bounded modularity, P3 executable transformation grammar, P4 platform-governed complement evolution, P5 integrated utility under diachronic burden. Ch3 §3.5.
  • REC — Recommendation identifier (used in Ch12 final recommendations). Ch12 §12.3.
  • TR — Technological Rule (TR-01 through TR-04). Mid-range design-knowledge contributions formalised in Ch11. Ch11 §11.3.

State and trajectory codes (Ch10 demonstration)

  • E0→1 … E4→5 — Transformation event identifiers between trajectory states. Ch10 §10.3–§10.4.
  • S0–S5 — Dwelling state identifiers across the demonstration trajectory; the trajectory forks at S4 into S5a (typical aging-in-place) and S5b (SDA-overlay-with-secondary). Ch10 §10.3–§10.4.

Symmetry and combinatorics

  • D4 — Dihedral group of order 8 (rotations and reflections of the square); used for polyomino canonicalisation. Ch8 §8.7.

Glossary

This glossary consolidates governed concept modules and separates foundational thesis terms from supporting clarification terms.

Core thesis concepts

  • actor — a schematic primitive of the Generative Grammar (Chapter 6, Section 6.1): the human participant a requirement addresses (participant, resident, wheelchair user, support worker). Made explicit under the cognitive primitive–composite–module re-stratification; inherits the participant lemmas formerly recorded under role, while role becomes a composite expressed through serves_role.

  • ageing in place Ageing in place is the sustained ability to live safely and meaningfully in one’s home and community as needs and capacities change.

  • bidirectional conversion Bidirectional conversion is the capability to map floor plans into a textual representation and reconstruct floor plans from that representation with controlled loss and explicit assumptions.

  • diachronic Diachronic describes change through time. In housing, diachronic change includes shifting capability, household structure, care arrangements, and regulatory expectations across the life-course.

  • housing optionality Housing optionality is the set of feasible, timely, and affordable housing adjustments available to a household as conditions change.

  • housing trajectories Housing trajectories are longitudinal housing pathways composed of states, transitions, and turning points that accumulate advantage or vulnerability over time.

  • interoperability Interoperability is the capacity to exchange information across tools, actors, and phases without losing semantic intent. It is not merely file transfer: interoperability requires that meaning and constraints survive translation.

  • life-course housing need Life-course housing need is the changing fit requirement between households and dwellings across developmental, social, and health transitions rather than a fixed age-band category.

  • modularity Modularity is an architectural property of bounded interaction: dense internal coherence is preserved while exchanges across boundaries are governed through explicit interfaces. In classic accounts, modularity manages change by hiding volatile details behind stable contracts.12

  • representation Representation is used here in an operational sense: the objects, relations, and permissible transformations a system can express; what must be explicit rather than inferred; and what survives exchange without semantic loss.

  • round-trip fidelity Round-trip fidelity is the degree to which a floor plan representation preserves relevant geometric, topological, and semantic constraints after an encode-decode cycle.

  • serialisation Serialisation is the encoding of representational state into a stable, exchangeable textual form so that structures can be stored, diffed, transported, and reprocessed across tools and actors.

  • structural coupling Structural coupling is a history of recurrent interaction through which a system and its environment become mutually adapted and constrained.

  • synchronous Synchronous describes a point-in-time framing: requirements are treated as satisfiable at a moment through a one-shot specification and a one-shot check. In housing, synchronous governance appears in certification and compliance regimes that validate an end-state snapshot.

  • text-based floor plan representation A text-based floor plan representation is a machine-processable textual encoding of plan semantics and geometry (for example, graph tuples, token sequences, or structured JSON) that can be interpreted by humans and/or computational systems without relying on raster drawing interpretation alone.

  • verification Verification is the process of checking whether a design, model, or dwelling satisfies stated requirements. Verification is low-friction when constraints are explicit and queryable; it becomes high-friction when meaning is implicit in drawings or depends on tacit interpretation.

Clarification concepts

  • adaptable housing Adaptable housing is housing designed so that key elements can be modified over time as needs and contexts change–ideally quickly, safely, and at lower cost than ad hoc retrofit.3

  • adaptation Adaptation is the governed modification of a dwelling, component, or rule-bound arrangement in response to changing functional conditions across time.

  • allostasis Allostasis is the achievement of functional stability through behavioural or physiological change in response to stress and variability.

  • autopoiesis Autopoiesis is the self-producing and self-maintaining organisation of a system that preserves identity while components and states change.

  • BIM Building Information Modelling (BIM) is a digital approach to representing built assets using structured model data rather than only drawings. In practice, BIM is often used within toolchains that still validate deliverables as point-in-time artefacts, which limits how well models support lifecycle adaptation and re-checking after change.

  • boundary conditions Boundary conditions are the explicit scope statements under which a design theory’s claims are warranted to apply: the contextual presuppositions, environmental constraints, and use settings that must hold for the artefact’s prescriptive propositions to retain their evaluative force, declared as integral components of the theory rather than appended as late caveats.4

  • boundary governance Boundary governance is the meta-question — and the corresponding rule apparatus — of how module boundaries are stipulated, made observable, monitored across change, and revised when revision is warranted, such that the interface contract at every boundary remains finite, declared, and accountable rather than tacit or interpretively reconstructed.56

  • bounded interaction Bounded interaction is the property that dense internal coherence is preserved while cross-boundary exchanges are restricted to explicit interfaces.

  • CAD Computer-Aided Design (CAD) refers to software-supported drafting and modelling used to author and document design. CAD workflows historically privilege geometric description and document outputs, which can make semantic intent and constraints difficult to preserve across handovers.

  • Cognitive cost Cognitive cost is the mental effort required to interpret, reconcile, and reason about a representation well enough to act safely. In housing delivery and adaptation, cognitive cost rises when intent is implicit, dispersed across documents, or lost across handovers.

  • cognitive map A cognitive map is an internal representation that supports orientation, navigation, and updating under change–commonly discussed in relation to how organisms move through space. The concept emphasises that robust action depends on stable substrate plus meaningful anchors and transition logic.

  • complex adaptive system A complex adaptive system is a system in which many interacting agents co-produce macro-level behaviour through feedback, adaptation, and path-dependent dynamics. Causal effects are often delayed, non-linear, and distributed across interacting components rather than localised in one decision point.

  • complement tier (superseded; also governed complements, complement library) — see governed instance library, the revised name for Layer 3 of the Governed Kernel Architecture. The earlier labels named the same referent — the slot, its contents (per Baldwin & Clark), and the artefact respectively — and are retained here only as cross-references.

  • configurative plane The configurative plane is the representational layer that binds identity and intent to assemblies of primitives (e.g., rooms, paths, interfaces), so that “what something is for” does not have to be re-inferred from geometry alone.

  • context Context is the declared boundary condition set within which claims are interpreted, design choices are justified, and evaluation outcomes are judged. It includes problem setting, actors, constraints, timescale, and transfer limits.

  • controlled coupling Controlled coupling is coupling that is explicit, bounded, and governed through defined interfaces, rather than hidden in tacit conventions or informal dependencies.

  • coordination requirements Coordination requirements are the minimum cross-boundary communication, dependency resolution, and verification actions needed for a system to remain valid during change.

  • delays and path dependence Delays and path dependence in housing refer to the combined mechanism by which non-convex adjustment costs impose delay in housing transitions and early trajectory conditions constrain later feasibility, such that misfit accumulates rather than dissipating and household pathways resist correction even when current circumstances would theoretically permit it.

  • design rules Design rules are explicit, stable coordination rules that separate what is fixed at system level from what can vary locally within modules. They govern interfaces, compatibility, and verification expectations so distributed design work remains coherent.

  • design science research Design science research is a research paradigm that produces and evaluates purposeful artefacts to solve class-relevant problems while extracting transferable design knowledge. Its quality depends on explicit problem grounding, explicit artefact logic, and explicit evaluation criteria.

  • design theory Design theory is a structured explanatory-prescriptive account of how and why an artefact class should produce desired outcomes under stated conditions. It specifies constructs, principles, justificatory knowledge, boundaries, and testable propositions.

  • distributed agency distributed agency denotes a named construct used in this thesis to state and test claims about housing representations and governance under change.

  • entropic loss Entropic loss is a representational failure mode where semantic intent decays across handovers: what begins as meaningful constraints and rationale gradually collapses into inert geometry, prose fragments, or tool-specific conventions.

  • environmental press Environmental press is the total demand load imposed by the dwelling and social setting on a person’s available competence.

  • epistemic uncertainty Epistemic uncertainty is uncertainty arising from incomplete knowledge of causal structure, delayed effects, and confounding in open systems.

  • explicit invariants and degrees of freedom Explicit invariants and degrees of freedom is the paired-contract commitment in which a module boundary states, in declared and checkable form, both what must remain true across permitted change (invariants) and what is permitted to vary, within what range, without escalating beyond the module (degrees of freedom).

  • form function principles Principles of form and function are the component of the Gregor-Jones design-theory anatomy that translate between what the artefact is (its structural, representational, and architectural commitments) and what the artefact does (the governance functions those commitments enable), specifying both halves of the translation in a form that downstream artefact decisions can be checked against.

  • frame of analysis A frame of analysis is the declared perspective that fixes what counts as a system element, what interactions are measured, and what boundary conditions are considered valid for evaluation.

  • frozen core The frozen core was the original name for the stable base of the module system; in the revised architecture the positive construct is the governed kernel, and “frozen core” is retained only in its pejorative sense — the failure mode in which a core over-expands into rigidity and forbids revision outright, trading evolvability for uniformity. The positive base is the deliberately minimal layer of a platform architecture that contains only the stable governance terms required for cross-version comparability and bounded verification (the grammar, module contracts, constituent specifications, and interaction rules that downstream artefacts may consume but not unilaterally alter); that base is now named the governed kernel precisely because it is governed — revisable through a versioned promotion path — rather than frozen.

  • functional boundary A functional boundary is a boundary defined by governable exchange conditions and verification obligations, rather than by geometric separation or naming convention alone.

  • governed instance library The governed instance library is Layer 3 of the Governed Kernel Architecture: the governed environment in which concrete baselines, variants, adaptation records, provenance, and supersession are managed. It instantiates the module contracts of the governed kernel (Layers 1 and 2) without altering them, and answers the question what can vary without breaking the system?

  • governed kernel (also: Governed Kernel Architecture) — the stable but revisable base of the three-layer architecture (Layer 1 Generative Grammar + Layer 2 Modular Contract System); see the concept note. The governed kernel is the stable but revisable base of the three-layer Governed Kernel Architecture: Layer 1 (the Generative Grammar — the stratified vocabulary of primitives, operators, and composites) together with Layer 2 (the Modular Contract System — the module contracts, constituent specifications, constraints, interface types, and verification sequence). Downstream artefacts may consume but not unilaterally modify the kernel; the kernel itself is revisable only through a rule-bound, evidence-based, versioned, and auditable promotion path. Its stability is disciplined revisability, not immobility.

  • functional usability verification Functional/usability verification is checking that a dwelling works as an integrated system for the intended occupants, not merely that individual features satisfy dimensional proxies.

  • geometry Geometry refers to metric and spatial description (shape, size, position). Geometry is necessary for design and checking, but geometry alone typically under-specifies functional and regulatory meaning (e.g., why a clearance exists, what path must remain continuous, what invariants must hold after change).

  • grid Grid denotes a stable metric substrate: a coordinate-like basis that makes spatial measurement and comparison reliable. In cognitive science, grid-like systems provide a reusable metric substrate on which environments can be mapped.7

  • grid cells Grid cells are neurons whose firing fields form a lattice-like pattern, widely interpreted as implementing a metric code that supports navigation and path integration.8

  • homeodynamics Homeodynamics is dynamic stability maintained through continual re-regulation and shifting baselines rather than strict constancy.

  • hypertrophic complexity Hypertrophic complexity is a representational failure mode where meaning is trapped in heavyweight encodings that are difficult to query, extend, validate, or safely change. Complexity “grows” around the representation because changes require global reinterpretation.

  • hysteresis Hysteresis is path-dependent asymmetry where the conditions required to recover a prior state differ from those that caused the transition away from it.

  • IFC Industry Foundation Classes (IFC) is an open, vendor-neutral schema intended to support exchange of building information across tools and phases.

  • information flows Information flows are the pathways by which meaning, constraints, and decisions propagate through a socio-technical system across actors and artefacts.

  • interaction differential Interaction differential is the relative difference between within-cluster interaction intensity and cross-cluster interaction intensity under a declared frame of analysis.

  • interactive plane The interactive plane is the representational layer that carries context and state for a specific dwelling at a particular time, including as-built deviations and the record of permitted changes.

  • interface An interface is an explicit boundary specification that governs interaction: what can be exchanged, what may vary, what must remain invariant, and what must be re-checked after a permitted change.

    Merged aliases: interfaces.

  • interface control Interface control is the ability to govern exchanges across a boundary through an explicit, finite set of conditions, rather than through tacit interpretation or pervasive coupling.

  • interface drift Interface drift is the failure mode in which the nominal rule surface of a module boundary remains in place, but its meaning changes across actors, revisions, or tools, so comparability decays beneath apparent formal consistency.

  • internal consistency verification Internal consistency verification is checking that changes do not silently violate dependencies across artefacts, phases, or representations.

  • justificatory knowledge Justificatory knowledge is the component of the Gregor-Jones design theory anatomy that carries the body of theoretical and empirical knowledge explaining why the proposed design should produce its intended outcomes.

  • manipulability Manipulability is the ability to perform controlled edits with predictable effects at the level of intent (not only at the level of geometry). A manipulable representation exposes guarded degrees of freedom and makes dependencies explicit, so competent actors can change what is allowed to change without silently violating hidden constraints.

  • module A module is a cluster of system elements whose internal interactions are, within a given frame of analysis, substantially stronger or more frequent than their external interactions. This differential interaction facilitates a functional boundary governing every intended exchange via an interface of finite and expressed set of conditions, thereby enabling the cluster’s interaction with the external environment with reduced coordination requirements while sustaining its role within the larger whole. Evidence anchor: Supplementary Author Research Corpus Assertions.

  • monolithic representation A monolithic representation binds many concerns into a single tightly coupled encoding, making local change difficult because dependencies are implicit and widely distributed. In such representations, a small change can force global reinterpretation and broad re-checking.

  • movement Movement denotes transition logic: the pathways and permissible transformations that connect states. In navigation, movement is not only distance; it is a sequence of permitted transitions under constraints.

  • near decomposability Near decomposability describes complex systems in which short-run behaviour is dominated by within-subsystem interactions, while cross-subsystem effects appear as weaker, slower perturbations. This structure makes systems intelligible and modifiable because local change does not require constant global recomputation.9

  • Physical cost Physical cost refers to material and bodily costs imposed by change: waste, invasive works, disruption, and physical burden associated with retrofit or rework.

  • place Place denotes identity and semantic attachment: the ability to treat a location as a coherent unit with meaning (what it is, what it affords, what constraints apply).

  • place cells Place cells are neurons that fire for specific locations, commonly interpreted as supporting place identity and contextual anchoring within a cognitive map.10

  • Planimetric System The Planimetric System is a representational architecture organised around the grid–place–movement triad, motivated by convergent evidence from spatial cognition. The aim is to separate metric substrate, semantic identity, and transition logic so that meaning can be preserved and updated under change without global recomputation.11

  • platform architecture Platform architecture is a system architecture pattern with a stable core contract and governed complements that can evolve independently under shared interface rules. The key design condition is not static uniformity, but controlled variation with compatibility and verification discipline.

  • point-in-time Point-in-time denotes a governance pattern in which compliance and fitness are validated as a snapshot at a single moment (e.g., design approval or certification), rather than as a sequence of permissible changes through the lifecycle.

  • primitive plane The primitive plane is the representational layer that carries stable metric substrate: discrete geometric primitives and coordinate commitments that support measurement and clearance checking.

  • Real cost Real cost refers to direct monetary cost (fees, construction, retrofit, delay penalties, relocation) imposed by design, delivery, and adaptation.

  • regulatory verification Regulatory verification is checking conformance to codified requirements (e.g., building codes, standards, approval rules).

  • representational constraint Representational constraint is the limit on what actors can compare, defend, and enforce, imposed by the artefacts and routines that carry meaning through a system.

  • residential inertia Residential inertia is reduced movement propensity caused by attachment, transaction frictions, and accumulated location-specific capital.

  • rigidification Rigidification is the substrate-induced suppression of adaptability: representational constraints make change costly enough that actors rationally freeze layouts, standardise, and treat adaptation as exceptional.

  • semantic intent Semantic intent is the “meaning that matters” for governance: why an element exists, what role it plays, what constraints apply, and what must remain true after change. Semantic intent differs from raw geometry because it includes functional and regulatory relations.

  • semantic interface A semantic interface is the interface type that governs meaning across a module boundary: it specifies the terms, object identities, and requirement references that must remain interpretable, type-stable, and referentially continuous when an entity is shared between, or transmitted across, adjacent modules.

  • semantic primitives Semantic primitives are minimal meaning units that can be composed, transported, indexed, and checked. The point is not that a primitive captures everything, but that it stabilises a small unit of meaning so interfaces can be explicit and verifiable.

  • semantics Semantics concerns meaning: the relations and constraints that make a representation interpretable and governable (not merely drawable). In rule-based checking, semantics must be explicit enough to query and validate.

  • serialised text Serialised text is the linear, character-discrete encoding of a governed housing representation into a form that is simultaneously human-readable and machine-parseable, such that each token’s type is positively and reliably identifiable and each compositional unit (primitive, module, constituent, interaction declaration) carries an explicit name, type, and scope rather than being implied by spatial position or tool-internal state.1213

  • serialised transformation logic Serialised transformation logic is the representation of permitted change operations as named, text-expressed grammar constructs that carry explicit preconditions, postconditions, invariant declarations, and responsibility metadata, such that the same notation that captures what a housing module is also captures what may legitimately change about it in a form that survives handover, inspection, replay, and verification without bespoke reconstruction.1415

  • Skill cost Skill cost is the dependence on scarce expertise to make small but consequential changes safely. When representations do not expose explicit constraints and permissible operations, routine adaptations are pushed into specialist workflows.

  • stable referential identity Stable referential identity is the property that named entities — modules, primitives, constituents, interaction declarations, baseline-library entries — retain identifiable identity across serialisation, deserialisation, transformation, handover, and tool replacement, such that a reference made by one actor at one time can be resolved by a different actor at a different time without re-interpretation of the underlying object’s name, type, or scope.1617

  • standardisation schema A standardisation schema is a representational contract that constrains permissible structure and term-role assignment across the clauses and categories of a regulatory text, converting narrative compliance prose into a queryable, machine-checkable substrate by codifying the mapping from regulatory language to typed primitives and explicit relations under declared validation and ambiguity-handling rules.

  • stratification Stratification is an architectural mechanism that separates representation into planes so coupling is controlled and change can be local. It applies the systems logic of bounded interaction to representation: interfaces define how layers depend on one another.

  • stratified discrete substrate A stratified discrete substrate is a representation designed to be serialisable, comparable, and transportable (discrete), while controlling coupling through layered planes and explicit interfaces (stratified).

  • testable propositions Testable propositions are the component of the Gregor-Jones design-theory anatomy that converts the theory’s claims into statements amenable to empirical examination and internal-consistency checking, distinguishing a prescriptive design theory from an aesthetic, hortatory, or merely descriptive framework by binding mechanism claims to declared indicators, comparator baselines, and falsifiers.18

  • Time cost Time cost is the delay imposed by coordination, reinterpretation, re-approval, and rework. Time cost increases when artefacts cannot be trusted and must be reconstructed at each handover.

  • transformability Transformability is the ability to perform structured transformations while verifying invariants. A transformable substrate allows change to be expressed as a sequence of named operations with explicit preconditions and postconditions, so validity can be re-checked after each step.

  • transformation Transformation is a change operation applied to a representation. In lifecycle governance, the key question is not “can it be changed?”, but “can it be changed in a bounded way while preserving invariants and remaining verifiable?”

  • transformation interface A transformation interface is the interface type that governs admissible change: it specifies operation types, allowable parameter movement, and invariant-preservation conditions under which local variation remains legitimate.

  • transportability Transportability is the ability to move a representation into a new context while retaining the conditions under which it remains valid. It is interoperability plus trust: meaning should travel with scope, assumptions, and validity conditions.

  • triggered interface checking Triggered interface checking is the verification regime in which a change at a declared boundary condition reactivates a finite, locatable set of re-checking obligations, rather than relying on either continuous whole-system revalidation or ad-hoc, actor-discretionary inspection.

  • trigram (Chapter 3) = triple (Chapter 5) = triplet (Chapter 6, Chapter 7) — the minimum-sufficient three-element governance unit (source–relation–target / entity–relation–qualifier). The label varies by chapter: Chapter 3 §3.3 introduces trigram from formal and cognitive arguments; Chapter 5 §5.1 uses triple to emphasise the applied schema vocabulary (the reconciliation is documented in-text at §5.1:42); Chapter 6 and Chapter 7 use triplet to emphasise the substrate-data vocabulary. The structural commitment is identical across the three labels: each governed representation is a three-element unit with stable identity, typed relation, and scoped qualifier.

  • ungoverned complements Ungoverned complements is the platform failure mode in which local elaborations of the platform — case instantiations, transformation sequences, notation expressions, and procedural outputs that should remain rule-bound complements of a shared core — proceed without rule-bounded compatibility, so that hidden coupling accumulates, exceptions multiply without typed justification, and verification scope re-globalises beneath an apparently stable interface vocabulary.19

  • universal design Universal design is an approach that aims to make environments usable by the broadest range of people, reducing exclusion by designing for diversity rather than by treating accessibility as an exception handled through special adaptation.

  • upstream prevention Upstream prevention is intervention at inflow and early-risk nodes to prevent entry into high-cost housing insecurity states.

  • value conflict Value conflict is disagreement among stakeholders about what ought to be optimised and what trade-offs are acceptable.

  • verification capacity Verification capacity is the system’s ability to check constraints and claims after change without reconstructing meaning from scratch.

  • verification interface A verification interface is the interface type that governs re-checking: it links changes to the evidence, tests, or review procedures that must be reactivated once a boundary condition has been touched.

  • wicked problems Wicked problems are planning problems that resist definitive formulation and definitive solution: there is no stopping rule, and outcomes are judged as better or worse under conflicting value frames rather than as true or false.20

    Merged aliases: wicked problem.

  • within_context — the tenth relation operator of the Generative Grammar (Chapter 6, Section 6.1): frames a term within a regulatory or situational scope (a design-category context), recorded as a governed link distinct from the context entity itself. Added under the re-stratification to complete the operator layer.

Methodological terms

  • census A census, in this thesis, is a complete enumeration of an assembled population rather than a probability sample drawn from it: every unit that passes the screening gate is included, and no sampling frame mediates between the population and the records analysed. The empirical substrate of Chapter 8 is a census of 745 Australian residential floor plans in this sense — the screened, agent-manually extracted stock in full, constructed in two strata (572 October and 173 August), not a sample of it.

  • coverage ratio A coverage ratio, in this thesis, expresses the scale of an assembled census relative to the national stock it is drawn from, offered as credibility context for the corpus’s non-triviality rather than as a basis for statistical generalisation. The 745-plan floor-plan census enumerates roughly one in ten thousand of Australia’s separate-house stock — the 2021 Census of Population and Housing records 10,852,208 private dwellings, of which about seventy per cent are separate houses, approximately 7.6 million.

  • Module-Inheritance Ratio (formerly the Composition-Boundary-Stability Index) The module-inheritance ratio (formerly the Composition-Boundary-Stability Index) is a chapter-level operationalisation of modular fit, defined as the proportion of modules at a given trajectory state that are inherited from the immediate predecessor without spatial change, relative to the total module count at that state: for a state S_n the ratio is |{m ∈ M_n : m ∈ M_{n-1} and spatial_dimensions(m) = spatial_dimensions_{n-1}(m)}| / |M_n|, where M_n is the module set at state S_n and spatial_dimensions is the module’s declared area-and-position vector.21

  • staged piloting Staged piloting, in this thesis, is the design-science practice of working out a data-processing codification on small, deliberate batches before applying it to a full corpus — building, evaluating, and refining the method, then scaling it under the refined rules. It is method development, not a separate corpus or a sample.

Writing Philosophy and Principles Commitment

This thesis is written and revised as a modular architecture of argument, where each chapter, module, claim, and evidence object has a defined function and a traceable contribution to the whole.

The thesis is written as a governance instrument for change, with modularity as the organising discipline: explicit meaning, explicit boundaries, explicit verification, and auditable adaptation over time.

  1. Problem-first fidelity: writing remains anchored to the diachronic housing-governance problem rather than rhetorical novelty, so module-level work does not drift from system-level purpose.
  2. Definition discipline: adopted definitions remain fixed and consistent, especially for module, modularity, and interface terms, so boundary conditions remain stable across chapters.
  3. Representation as governed obligation: obligations and invariants are encoded explicitly so meaning survives handover and transformation across representational modules.
  4. Boundary and interface explicitness: finite, declared interface conditions are preferred over implicit interpretation, so exchanges between sections, artefacts, and claims remain controlled.
  5. Evidence traceability: every substantive claim is expected to be supportable, inspectable, and reproducible from local evidence, so each module can be audited without inference gaps.
  6. Verifiability under change: claims are accepted only when post-change validity can be checked without global reconstruction, preserving local revision with whole-thesis coherence.
  7. Writing clarity with technical precision: concise plain English is maintained while high-precision terminology is used only where it carries analytical necessity, keeping modular interfaces legible.

This commitment is used to maintain modular argument continuity, methodological discipline, and evidence-grounded accountability across the thesis.

Notes

  1. On the Criteria To Be Used in Decomposing Systems into Modules DL Parnas Carnegie-Mellon University, accessed on July 23, 2025, http://sunnyday.mit.edu/16.355/parnas-criteria.html ↩︎
  2. Baldwin, C. Y., & Clark, K. B. (2000). Design Rules, Volume 1: The Power of Modularity. MIT Press. https://direct.mit.edu/books/monograph/1856/Design-Rules-Volume-1The-Power-of-Modularity (accessed 2026-02-03). ↩︎
  3. Adaptable Housing for People with Disability in Australia, accessed on July 18, 2025, https://humanrights.gov.au/sites/default/files/document/publication/monash-adaptablehousing2021-digital.pdf ↩︎
  4. S. Gregor and D. Jones, “The Anatomy of a Design Theory,” Journal of the Association for Information Systems, vol. 8, no. 5, pp. 312-335, 2007, doi: 10.17705/1jais.00129. ↩︎
  5. D. L. Parnas, “On the Criteria to Be Used in Decomposing Systems into Modules,” Communications of the ACM, vol. 15, no. 12, pp. 1053-1058, 1972, doi: 10.1145/361598.361623. ↩︎
  6. C. Y. Baldwin and K. B. Clark, Design Rules, Volume 1: The Power of Modularity. Cambridge, MA, USA: MIT Press, 2000. ↩︎
  7. A principle of economy predicts the functional architecture of grid cells | eLife, accessed on August 30, 2025, https://elifesciences.org/articles/08362 ↩︎
  8. A principle of economy predicts the functional architecture of grid cells | eLife, accessed on August 30, 2025, https://elifesciences.org/articles/08362 ↩︎
  9. Simon, Herbert A., “The architecture of complexity”, Proceedings of the American Philosophical Society, (1962), accessed on October 24, 2025 ↩︎
  10. What Are Grid-Like Responses Doing in the Orbitofrontal Cortex? - PMC - PubMed Central, accessed on August 30, 2025, https://pmc.ncbi.nlm.nih.gov/articles/PMC8299309/ ↩︎
  11. A principle of economy predicts the functional architecture of grid cells | eLife, accessed on August 30, 2025, https://elifesciences.org/articles/08362 ↩︎
  12. J. Haugeland, “Analog and Analog,” Philosophical Topics, vol. 12, pp. 213-225, 1981. ↩︎
  13. N. Goodman, Languages of Art: An Approach to a Theory of Symbols. Indianapolis: Bobbs-Merrill, 1968. ↩︎
  14. D. L. Parnas, “On the Criteria to Be Used in Decomposing Systems into Modules,” Communications of the ACM, vol. 15, no. 12, pp. 1053-1058, 1972, doi: 10.1145/361598.361623. ↩︎
  15. C. Y. Baldwin and K. B. Clark, Design Rules, Volume 1: The Power of Modularity. Cambridge, MA, USA: MIT Press, 2000. ↩︎
  16. D. L. Parnas, “On the Criteria to Be Used in Decomposing Systems into Modules,” Communications of the ACM, vol. 15, no. 12, pp. 1053-1058, 1972, doi: 10.1145/361598.361623. ↩︎
  17. C. Y. Baldwin and K. B. Clark, Design Rules, Volume 1: The Power of Modularity. Cambridge, MA, USA: MIT Press, 2000. ↩︎
  18. S. Gregor and D. Jones, “The Anatomy of a Design Theory,” Journal of the Association for Information Systems, vol. 8, no. 5, pp. 312-335, 2007, doi: 10.17705/1jais.00129. ↩︎
  19. A. Tiwana, B. Konsynski, and A. A. Bush, “Platform Evolution: Coevolution of Platform Architecture, Governance, and Environmental Dynamics,” Information Systems Research, vol. 21, no. 4, pp. 675-687, 2010, doi: 10.1287/isre.1100.0323. ↩︎
  20. What’s a Wicked Problem? | Wicked Problem - Stony Brook University, accessed on July 18, 2025, https://www.stonybrook.edu/commcms/wicked-problem/about/What-is-a-wicked-problem ↩︎
  21. C. Y. Baldwin and K. B. Clark, Design Rules, Volume 1: The Power of Modularity. Cambridge, MA, USA: MIT Press, 2000. ↩︎