Literature Review and Theoretical Framework

Four bodies of literature compose this chapter’s cumulative argument: person–environment fit, supply-side architecture, bounded optionality, and modularity theory. Each demonstrates a distinct failure mode in housing adaptation that the others cannot explain, and each draws on an established research tradition with its own evidential base. Their convergence on a common gap — the absence of governed representational infrastructure — is, on our reading, a finding of this review rather than a pre-theoretical commitment.

Dynamic misfit requires a demand-side explanation. Housing need is a dynamic person–environment fit problem: misfit accumulates because adjustment is discontinuous and high-friction (Sections 2.1–2.3). Supply-side architecture depends on separation of concerns and interface discipline that may preserve local optionality (Sections 2.4–2.5). Bounded optionality remains narrowed by interaction costs, administrative burden, and coordination demands — a reduction we formalise as dimensionality reduction through enabling constraints (Sections 2.6–2.7). Modularity provides the structural mechanism: bounded interdependence governed by explicit interfaces keeps change manageable, under six derivable conditions, and yields four critical properties that an adaptation-heavy housing representation must possess (Sections 2.8–2.9).

Representational governance is the integrative concept that emerges from this convergence. The final section produces five meta-requirements (MR1–MR5) and five research questions (RQ1–RQ5) from the asymmetry between strong descriptive knowledge and fragmentary prescriptive knowledge (Sections 2.10–2.11). Each answer requires a designed artefact as its solution form.

Adaptation-heavy housing is a relational, time-varying person–environment fit problem: a condition that holds when a dwelling and its surrounding socio-institutional context remain adequate for a household whose abilities, resources, and circumstances change across the long life of the building.¹² Fit holds when environmental demands sit within the household’s capacity to meet them without excessive cost, risk, or strain; misfit arises when those demands exceed household capability for a sustained period, or when the dwelling–tenure–location bundle ceases to support the household’s functioning given its current requirements.³

The ‘person’ side, treated here as a household-level capability set, is multi-dimensional and time-varying.⁴⁵ Household capability includes income and liquidity, time availability, health and functional capacity, knowledge and information access, social supports, and household composition — the latter encompassing dependants, caring responsibilities, and the interdependencies of ‘linked lives’ that shape who must be accommodated and supported. Because capability moves across so many dimensions over time, housing need is poorly captured, we argue, by any single demographic marker or administrative category. The ‘environment’ side is correspondingly broad. Environmental demands include the physical dwelling, the locational context, and the institutional and market context that determines what adjustments are feasible.⁶ Tenure security, permission to modify, rent-setting practices, financing constraints, eligibility rules, waiting lists, and administrative complexity all function as environmental "press" — recurring demands on household time, cognition, and resources — so that housing operates as a bundle of material, legal, administrative, and spatial conditions that jointly shape what households can actually do.⁷

Competence–press model of person–environment fit The Lawton–Nahemow competence–press model: fit holds within a tolerable zone where environmental demands are calibrated to household capability; misfit accumulates as press exceeds competence or competence declines below press.

Two implications follow. First, housing need is inherently dynamic because both household capability and environmental demands change. Households undergo transitions, employment shifts, relationship changes, additions or losses of household members, caring episodes, migration, health shocks, and gradual capability changes; housing environments meanwhile change through rent resets, neighbourhood transformation, policy changes, service availability, and market cycles. Fit must therefore be re-established repeatedly through ongoing adjustment, because capability and demand both keep shifting.⁸ Second, misfit is structurally likely to persist because housing adjustment is typically slower and more discontinuous than household change. Housing stock is durable and place-fixed; modifications require capital, permissions, coordination, and skilled labour; relocation entails search, negotiation, and non-trivial transaction, social, and emotional costs.⁹ A better fit may exist in principle yet remain infeasible within relevant time and resource constraints, and households frequently tolerate growing mismatch instead of correcting it continuously.

For analytical traction, misfit is treated as an observable gap between household requirements and the effective housing bundle. It manifests along six dimensions — space and configuration, accessibility and safety, affordability and financial exposure, security and control, location and connectivity, and administrative and informational burden.¹⁰¹¹¹² Misfit is distinct from crisis: a crisis is a threshold-crossing event such as eviction, homelessness entry, or forced relocation, and misfit may exist well before such visible rupture. The key mechanism is accumulation — small mismatches compound when households and other local actors lack feasible, low-friction routes to realign housing with changing requirements — and over time misfit may cross financial, functional, psychological, or institutional thresholds after which adjustment becomes abrupt, high-cost, and welfare-reducing.¹³

If housing need is a dynamic person–environment fit problem, the immediate question is where misfit originates. The life-course perspective supplies analytic tools for specifying how and when housing requirements change, and why housing careers become heterogeneous.¹⁴ Life-course analysis treats the household as a unit in patterned motion, foregrounding change over time and the way historical and institutional contexts shape it. Its core concepts work as devices for tracing the production of misfit through time. Cohorts are groups exposed to similar macro-conditions at key stages — policy regimes, labour markets, and housing-market conditions can durably shape later tenure access and wealth accumulation. Trajectories are long-run patterns of stability and change in housing careers that may include repeated tenure shifts and varying degrees of security. Transitions are discrete role or status changes — partnership formation or dissolution, childbirth, employment changes, caring episodes, illness — that frequently reconfigure space, location, affordability, and support requirements. Turning points are critical events that shift the subsequent pathway, often altering future feasibility sets and not merely generating short-term disruption. Linked lives name the interdependence of individual lives within households and networks: a transition for one person may alter the housing requirements and constraints of others, and therefore the whole household’s fit condition.

These concepts matter because they specify the mechanisms by which ordinary life-course change reconfigures housing requirements across multiple dimensions. Transitions frequently shift space and configuration needs (household size, privacy requirements, work-from-home arrangements, caregiving space), affordability constraints (income shocks, childcare costs, health expenditures), support needs (proximity to kin, services, schooling, healthcare), and location requirements (commuting, access to jobs, safety, transport). A dwelling that fits well at one stage may become misfit at the next, because the household’s capability set and requirements have shifted while the housing bundle remains relatively fixed. Cohort constraints sharpen the point. Households face different opportunity sets at comparable life-course stages: macro-conditions at key transitions — labour-market security, interest rates, rent levels, credit availability, the supply of suitable dwelling types — may permanently alter later tenure access and housing stability. Cohort effects therefore operate as structurally different feasible sets, shaping whether a household can meet transitions with incremental adjustments or is forced into discontinuous, high-stakes changes that risk entrenching misfit. Biography explains when requirements change; cohort conditions strongly influence whether re-fitting is feasible when those changes occur.

Once housing need is framed this way, heterogeneity in housing outcomes becomes expected. Longitudinal research on housing insecurity identifies distinct subpopulations with highly divergent pathways, from largely stable trajectories to episodic instability and a smaller subset experiencing chronic, compounding insecurity.¹⁵ This heterogeneity foreshadows the later argument about divergence and stickiness: pathways differ because households differ and because they face different feasible sets and different capacities to re-fit when transitions occur. Because housing is durable and adjustment is costly and delayed, many misfits persist instead of being corrected smoothly — and the next section turns to why.

Misfit persists because the housing system is structurally biased towards inertia. Dwellings are durable, place-fixed assets; they cannot be reconfigured, replicated, or relocated on demand, and the surrounding institutional environment — planning controls, tenure rules, market structures — constrains how quickly supply and household–dwelling matches can change.¹⁶ Households that recognise a mismatch still adjust within a slow-moving stock, with limited vacancies in relevant locations and typologies, and institutions that often delay or prohibit timely modification and mobility.

The central friction is that adjustment costs are typically non-convex: a fixed burden is incurred whatever the size of the change, so small corrections are dominated by their threshold cost and become practically infeasible. Moving home demands search, negotiation, inspection, applications, reference checks, legal processes, and often simultaneous coordination of exit and entry dates, and these burdens are compounded by direct financial costs (fees, taxes, relocation expenses) and by the disruption of routines, services, and local networks.¹⁷¹⁸¹⁹ Where the adjustment route is modification, non-convexity returns through approvals, landlord permissions, contractor availability, and financing sequencing, so that even an in-place change becomes a thresholded project. The minimum viable correction is therefore large: the next feasible step out of a visibly imperfect fit is a discontinuous jump, which is why households so often stay put.

These frictions extend beyond finance. Time and coordination costs accumulate as a distinct form of press. Adjustments may require synchronising multiple parties (landlords, agents, lenders, builders, local authorities, household members, service providers) and navigating queues and delays (inspection cycles, application timelines, approval windows, settlement periods). In private rental contexts, critical life events can trigger acute risk precisely because tenancy systems require rapid administrative responses — bond transfers, reapplication, lease renegotiation — under time pressure and uncertainty.²⁰ These temporal and coordination burdens matter because they suppress frequent micro-adjustments and push households toward delayed, discontinuous changes — exactly the condition under which misfit accumulates.

The Stock Adjustment Model formalises this concisely: households occupy an actual level of housing — the dwelling they have — while holding a desired level — the bundle that would best fit their income, household size, preferences, and constraints. Because adjustment is costly, households close the gap only partially in any given period; the adjustment rate sits substantially below one, so much of the mismatch carries from one period into the next.²¹ The behavioural implication anchors the argument here: when the cost structure is non-convex, inaction stays rational across a wide range of mismatches, and adjustment waits until a threshold is crossed.

Empirical estimates corroborate the scale of this inertia. Detailed urban-housing data show that only around 14–27 per cent of the gap between actual and desired housing closes in a typical period, so most mismatch carries forward.²² The same data reveal an asymmetry: households adjust more rapidly when moving ‘up’ (increasing housing consumption) than when moving ‘down’. In Phoenix the estimated adjustment rate is markedly higher for increases (0.370) than for reductions (0.128); Pittsburgh shows a similar though smaller asymmetry (0.151 upward against 0.127 downward).²³ The asymmetry is consequential for misfit: mismatches that demand downsizing, reducing costs, or relinquishing valued attributes tend to persist longest, even when objectively inefficient.

The practical consequence is that misfit often accumulates. Households adapt informally — tolerating overcrowding, under-occupation, long commutes, unsafe layouts, unaffordable costs, or insecure tenure — until the accumulated strain exceeds a functional, financial, or psychological threshold. Adjustment then occurs abruptly and often under duress, through forced mobility, crisis-driven tenure shifts, or disruptive ‘lumpy’ interventions that are welfare-reducing relative to earlier, incremental re-fit.²⁴ Once persistence becomes cumulative, it becomes path dependent — producing the lock-in, hysteresis, and ‘sticky’ housing trajectories that the next section develops.

Persistence becomes stickiness once repeated or prolonged misfit stops merely inconveniencing the household and starts to reshape the future feasible set itself, stabilising long-run secure and insecure pathways through three closely related mechanisms. Path dependence, lock-in, and hysteresis together generate trajectory formation, which complex-systems language describes as movement into stabilised states and difficulty in exiting them.²⁵²⁶

Path dependence — cumulative constraint. Early conditions, decisions, and shocks alter the structure of subsequent opportunities, so that later outcomes register the accumulated history of a pathway and not merely its current circumstances.²⁷ In housing, small differences in initial conditions appear to compound over long horizons through reinforcing mechanisms that run in opposite directions for different starting positions: secure tenure supports stable employment and savings, which in turn underwrite further security; insecure housing erodes creditworthiness, weakens social networks, and amplifies exposure to repeated administrative and market pressures.²⁸ Longitudinal evidence illustrates the asymmetry concretely. Cohort analyses show that privileged pathways — transitioning from owner-occupied parental homes into earlier mortgage homeownership — are associated with substantially higher housing and non-housing wealth accumulation by midlife, while steady renters experience compounding disadvantage across the same horizons.²⁹ The same logic operates at the insecurity end of the distribution, where early-life perturbations appear to carry delayed but durable impacts as they reorganise later risk environments and the resources available for re-fit. Victoria’s International Youth Development Study, for instance, finds that high levels of family conflict in early adolescence predict homelessness in young adulthood, even after accounting for other contextual risks, with peer processes mediating part of the association.³⁰³¹ The mechanism claim is therefore narrow and load-bearing: early trajectory positions shape the later feasible set.

Lock-in — layered exit costs. Lock-in operates as a specific manifestation of path dependence in which a system becomes trapped in a sub-optimal configuration because the costs of exit are prohibitive relative to perceived gains.³²³³ In housing, lock-in is rarely “just preference”; it is produced by layered exit costs operating at three registers simultaneously: financial (taxes, fees, deposit gaps, credit constraints); institutional (eligibility rules, tenancy conditions, planning constraints); and social (loss of local support networks, disruption to caregiving, education, and employment routines).³⁴ What matters for the misfit argument is that the exit pathway itself is high-friction. Households perceive the mismatch clearly and still remain in states widely recognised as inefficient or harmful, because the layered cost of exit is too high to absorb in a single small step. Lock-in therefore converts persistent misfit into a durable trajectory position, with spillover effects on wider system dynamics through constrained vacancy chains and reduced mobility options for the households that would otherwise have followed.

Hysteresis — irreversibility. Hysteresis sharpens the argument by focusing on irreversibility: a temporary shock produces long-lasting structural effects such that, even after the shock is reversed, the system does not return to its prior state.³⁵ In housing pathways the shock may be an illness, an unemployment spell, an interest-rate spike, or a relationship dissolution that forces a household into lower-security tenure or a less suitable dwelling. Income recovery does not restore the prior pathway, because the return route is itself foreclosed by lost equity, credit scarring, changed prices, and disrupted networks. The point holds at the socio-cultural level as well, where hysteresis has been used to explain persistent expectation and aspiration shifts following adverse market conditions — situations in which objective economic improvement leaves prior subjective housing trajectories unrestored.³⁶³⁷ The analytic implication is direct: hysteresis is the mechanism by which short-run disruption becomes long-run trajectory divergence, and it is the irreversibility of the initial shock, more than its depth, that does the work.

These three mechanisms admit a complex-systems reading in terms of attractor-like trajectories and multistability, provided the terminology is handled carefully. In dynamical-systems vocabulary, an attractor is a state or behavioural regime that a system tends to settle into; the basin of attraction is the set of conditions from which the system tends to return to that regime following perturbation.³⁸³⁹ Applied to housing, “attractor-like” carries neither determinism nor mysticism; it is descriptive shorthand for the observation that certain housing states are sustained by feedbacks and exit barriers more than by the household’s current preferences. Once a household occupies a secure and resource-generative position, reinforcing processes — wealth accumulation, residential stability, institutional recognition, credit access — make that position resilient to moderate shocks. Once a household enters severe insecurity, interacting disadvantages — health deterioration, network severance, labour-market exclusion, administrative burden — make exit difficult and relapse likely.⁴⁰⁴¹ Housing systems therefore exhibit multistability in the practical sense relevant here: multiple durable pathway regimes — secure, precarious, chronically insecure — coexist and remain relatively stable once established, with transitions between them concentrated at moments when the basin barriers are temporarily weakened.⁴² One consequence follows directly: isolated, short-duration interventions may target exactly the right households and still appear weak, because their weakness traces to the interacting basin barriers left jointly intact — credit constraints, service fragmentation, network disruption, and ongoing administrative press operating together.⁴³

Trajectory Divergence under Stickiness Divergent trajectory regimes — secure, precarious, chronically insecure — stabilised by feedbacks and exit barriers across path-dependent, locked-in, and hysteretic housing careers.

These mechanisms generate empirically detectable consequences: housing systems produce divergent housing careers — pathways that differ in stability, tenure security, and exposure to disruption, with divergent downstream effects. Two complementary descriptive approaches map this divergence. Transition probabilities and state-based modelling represent housing careers as movements between defined housing states — private renter, social renter, mortgaged owner, outright owner, transitional accommodation — using panel datasets such as the Household, Income and Labour Dynamics in Australia (HILDA) survey to quantify how life-course events and market conditions shift transition risks across multi-year tenure sequences.⁴⁴⁴⁵ Sequence analysis complements this by treating housing careers as ordered strings of states across time, identifying trajectory typologies through pattern similarity without imposing linearity.⁴⁶⁴⁷ The latter approach aligns closely with the stickiness argument because it captures volatility (frequent transitions), persistence (long spells in one tenure), and scarring patterns (movement into insecurity followed by constrained re-entry). Across this literature, the traditional housing career — parental home → renting → mortgage → outright ownership — appears increasingly fragmented and obsolete as a general model.⁴⁸⁴⁹⁵⁰ Modern sequences show high volatility alongside constrained stalling patterns in which households remain in imperfect fits because the next feasible move is discontinuous and costly.

The significance of these mapping results extends beyond description. If housing pathways differ systematically in stability and security, they should also differ in cumulative outcomes reflecting prolonged fit or prolonged misfit — and on the evidence available, they do. Longitudinal evidence from Australia, examining individuals aged sixty-five and older and adjusting for earlier-life socio-demographic circumstances, shows that owner-occupiers exhibit higher Total Life Expectancy (TLE) and Disability-Free Life Expectancy (DFLE) than private renters. Among men, owner-occupiers at sixty-five record 19.0 years of TLE against 16.7 years for renters — a 2.3-year difference, accompanied by a 1.8-year DFLE gap. Among women, the corresponding figures are 22.9 against 20.6 years — again a 2.3-year TLE difference, accompanied by a 3.1-year DFLE gap.⁵¹ These differences are consistent with the broader claim that housing careers operate as cumulative exposures — to stability, control, and material security on one side, or to sustained press, insecurity, and constrained re-fit capacity on the other.⁵² Inference must, however, be managed carefully. Tenure is not randomly assigned, and selection processes — healthier or higher-income households being more likely to enter and remain in ownership — plausibly contribute to the observed gradient. Even with adjustment for prior socio-demographic circumstances, residual confounding appears likely, and the observed differences should therefore be read as associations consistent with cumulative advantage and disadvantage rather than as definitive causal estimates.

Stickiness, finally, operates at the system level as well as the household level. Durable housing stock and institutional fragmentation produce bottlenecks that limit the feasible set even for households not yet in chronic insecurity, foreshadowing why the argument must now move from individual sticky trajectories to the supply-side architecture that constrains them. Section 2.4 takes up that shift, where the question becomes whether local supply systems are organised to expand or contract the option set through which households execute timely, low-friction re-fit.

Explanatory emphasis now shifts from the household’s changing requirements to the supply-side architecture of change: how the housing system can be organised so that re-fit proceeds through small, reversible interventions, with lumpy crisis-driven moves held in reserve as a fallback. Ordinary change becomes expensive, slow, and disruptive precisely when the built environment and its governance are organised in ways that block low-friction adjustment.⁵³⁵⁴

The starting point we adopt is a reconceptualisation: the built environment is understood less as a set of finished objects and more as an evolving socio-technical process, so that design and construction govern the management of change over time. On our reading, separating long-lived capacity from short-lived configuration lets future adjustments stay local instead of propagating destructively across the whole. The practical meaning of “local optionality” follows directly: it is the availability of feasible next steps that can be executed under real constraints of money, time, permissions, coordination capacity, and uncertainty. Local optionality is achieved when change is engineered and governed so that it remains local — confined to parts of the system whose alteration does not trigger cascading redesign, heavy disruption, or irreversible commitments.

Housing adjustment becomes discontinuous because the built environment is commonly organised as a tightly coupled bundle: structure, spatial layout, and building services are physically interwoven, and control over change is simultaneously coupled to legal and administrative permissions. In this condition, “change” ceases to be a local edit to one part of the system and becomes a cascading intervention propagating across subsystems. The supplied literature describes this lock-in dynamic most explicitly through the concept of path dependence — initial decisions constrain future options, and the resulting trajectory becomes self-reinforcing and increasingly difficult to deviate from.⁵⁵ A path-dependent system is prone to “lock-in,” where increasing returns and positive feedback may make reversal prohibitively expensive or practically impossible.⁵⁶⁵⁷ In housing the mechanism is concrete and material: early technical choices harden into irreversible commitments because key systems are embedded within one another — plumbing integrated into load-bearing concrete walls, room sizes fixed through structural partitions, services cast into structure. Once the structure is poured, the dwelling is committed to a particular configuration.

Once systems are entangled in this way, changing one element forces changes elsewhere — altering spatial layout requires demolition of structural walls, and updating services requires coring concrete and reworking connected systems. These physical cascades amplify organisational ones: specialist expertise, coordination among trades, and approval processes that classify the intervention as major work demanding the procedures of incremental maintenance.⁵⁸ Under tight coupling a routine alteration therefore escalates into structural demolition, and the corresponding switching costs are multi-dimensional — money, time, disruption (loss of use, relocation), specialised expertise, coordination, and permissions.⁵⁹ When switching costs are high, households rationally delay adjustment and tolerate misfit, and the “next step” becomes a disruptive cliff. Misfit accumulates until a threshold is crossed; change then turns forced and discontinuous — major renovation, relocation, or abandonment — beyond the reach of routine adaptive moves. This is the supply-side substrate of lumpy adjustment.

If tight coupling converts ordinary change into demolition-level intervention, the corresponding supply-side solution is a separation-of-concerns problem: the housing system must be organised so that parts with different change-rates can evolve without forcing one another to change. Reconfiguration capacity is preserved when long-life capacity is separated from short-life configuration, with the boundary between them explicit enough to prevent cross-layer cascade. A first step is recognising layered lifetimes. The literature frames this explicitly through Simon’s near-decomposability — interactions within subsystems are stronger than interactions between subsystems, allowing higher-frequency dynamics to change without destabilising lower-frequency structures.⁶⁰ In built form this becomes fast-changing layers (use patterns, internal partitions, furniture arrangements) and slow-changing layers (structural frames, foundations).⁶¹ When fast and slow layers are fused, everyday change becomes structurally expensive; when separated, fast layers can adapt while slow layers remain stable.

Layered Lifetimes Layered lifetimes after Brand’s shearing-layers — fast-changing layers (use patterns, stuff, space plan) adapt without destabilising slow-changing layers (structure, site).

Separation-of-Concerns Design Logic Separation-of-concerns design logic — long-life capacity is held stable while short-life configuration is permitted to evolve, with the boundary between them disciplined by interfaces.

The layered-lifetime view carries an organisational corollary: task partitioning. If different layers change at different rates, their responsibilities, risks, and rights should be partitioned too, so actors are not forced into decisions outside their horizon or control. Task partitioning operates as the judicious separation of design, production, and maintenance tasks based on component lifespan and territorial control.⁶² The same logic frames a hierarchical structure of intervention — higher levels (urban planning, base building) set the stage for lower levels (individual floor plans) without strictly determining them, which requires interfaces between levels to be specified as technical, legal, and spatial boundaries.⁶³ When long-life capacity is built as a stable platform and short-life configuration is treated as changeable, fewer decisions become irreversible commitments, and change can stay “within layer,” avoiding the cascading redesign that tight coupling produces.

The Open Building / Skeleton–Infill family supplies a worked precedent we use here instrumentally, as an entrepôt or staging platform. It shows, in a fully articulated housing precedent, what it means to separate long-life capacity from short-life configuration. The tradition emerged as a technical and organisational response to obsolescence in mid-twentieth-century mass construction, and its intellectual lineage may be traced from the elemental decomposition of the primitive hut (hearth, roof, enclosure, mound) to Habraken’s formalisation of independent building levels.⁶⁴ The canonical mechanism is the Support–Infill split — a deliberate separation of a building volume into two layers with different lifetimes, owners, and change rights.⁶⁵⁶⁶ The Support (also called the Base Building or Skeleton) is the long-life shared capacity: primary structural frame, load-bearing walls, façade, entrances, staircases and elevators, and the main infrastructural lines for utilities, engineered for an operational lifespan on the order of fifty to one hundred years.⁶⁷ The Infill (fit-out, recheio) is the short-life configuration within that capacity — non-load-bearing internal partitions, modular interior finishes, cabinetry, and secondary, user-specific mechanical and electrical routing — under the domain of the individual user or occupant. Disentangling installations and reducing cross-dependencies shifts complexity toward the less complex, more readily changeable layer, and the precedent’s own evaluative concept for this — transformation capacity, a measurable phenomenon indicating how efficiently a building may adapt to evolving spatial, functional, and environmental demands without requiring structural demolition⁶⁸ — appears as the supply-side analogue of local optionality.

“Local change” is never only physical. The Open Building tradition explicitly involves physical, legal, and financial separation between layers, so the ability to change a part depends on who controls it and which permissions are required. At the governance level this is framed as a polycentric structure: decision-making is segmented by building level to decouple parts controlled by different parties and built by different trades.⁶⁹ The same framing aligns with subsidiarity ideas in institutional economics — higher governance levels recognise and protect the rights of lower levels, so that initiative remains distributed across actors.⁷⁰

Taken as an illustrative precedent, separation provides an existence proof that reconfiguration can be made feasible; its value here is the demonstrated mechanism, which the discipline of Open Building puts to work as an entrepreneurial staging platform. It shows, in a housing-ready pattern, how long-life capacity can remain stable while short-life configuration remains revisable, lowering switching costs and increasing the practical availability of next steps. Separation is, however, necessary but not sufficient: optionality only materialises when the interfaces between layers are explicitly disciplined, which is the concern of Section 2.5.

Separation-of-concerns in housing becomes operational once the boundary between layers is treated as an interface that carries enforceable rules, and in the built environment that interface is plural. Spatial interfaces fix which volumes and zones belong to which layer; dimensional interfaces fix where elements must land in plan and section; servicing interfaces fix how utilities cross layers without entanglement; connection interfaces fix how parts join, tolerate movement, and can be removed without collateral damage to neighbouring assemblies. Support and Infill remain independent yet cohesive only under rigorous rules for positional and dimensional coordination across this plural interface set.⁷¹⁷²

The core discipline — and this appears to be the load-bearing move of the entire Open Building lineage — is to treat dimensional coordination as a binding contract. Open Building operationalises that contract through modular coordination tied to standards: a basic module of 10 cm under NEN 5700 and the relevant ISO codes, with Support–Infill coordination governed through tartan grids typically in the 10–20 cm range. Historical implementations show how the contract acquires governance force in practice. At La Maison Médicale, a 30 × 30 cm tartan grid was integrated into a static concrete support — pillars plus load-bearing walls — that governed flexible interior distributions while leaving the structural system untouched at each rework.⁷³⁷⁴ The discipline decomposes further into modular subsets — structural, installation, geometry, and element modules — so that spans, shafts, service routes, and surface-defining elements each “agree” on where they can and cannot land, reducing the probability that a change in one category forces redesign in another.⁷⁵ Standardisation functions as interface specification: it stabilises boundaries so that reconfiguration stays local instead of cascading across layers.

Service interfaces are at least as decisive as geometric ones. Where services are embedded inside static walls or cast into structure, change becomes destructive by default — the everyday alteration converts into demolition before the household has framed it as renovation. Open Building interface logic therefore extends into service routing and accessible service zones, and the engineering target is to avoid complex, irreversible entanglement between element types — especially the binding of moulded-on-site materials (‘M’) to highly finished prefabricated parts (‘O’).⁷⁶ The Matura Infill System illustrates the principle compactly: conduits and cabling are redistributed through a floor-based “Matrix tile” and an L-shaped baseboard channel, while vertical walls remain free of wiring — so rapid post-occupancy alteration proceeds without tearing open walls.⁷⁷ The Passe-Partout family extends the same logic at infrastructure scale, where an independent frame separates the roof structure from gables and façades, making the roof a replaceable subsystem whose alteration triggers no cascade.⁷⁸ Empirical evaluation reinforces the design intuition. Post-occupancy assessment of Open Building dwellings at IBA Hamburg, conducted seven years after habitation, reports measurably greater spatial reconfiguration by residents in Open-Building units than in equivalent conventional units — empirical traction for the near-decomposable layering set out in Section 2.4.⁷⁹

Joint and tolerance discipline is, in practical terms, what keeps loose coupling functional over time and prevents its decay into the ad hoc workarounds it was designed to displace. A boundary prevents cascade only when its connections are designed so that parts can be installed, removed, and replaced within known tolerances, and the detailing must anticipate variation in advance instead of relying on site correction at each rework. Here the common failure mode appears: standardisation without detailing competence. Pedagogical evidence makes the mechanism legible. Students adopt Open Building and modular logics, and successfully engage user agency at the infill level, yet insufficient detailing experience inverts the intended outcome — structural rigidity, reduced façade flexibility, and inadequate technical resolution at precisely the joints meant to absorb change.⁸⁰ The lesson is sharp: optionality emerges only when interfaces are engineered so that local variation is absorbed inside a layer; declaring layers, by itself, achieves nothing.

Interface discipline therefore operates as an enabling constraint. It imposes local limits — grids, service zones, connection rules, tolerances — and those limits are precisely what give the system access to a wider range of feasible reconfigurations without redesigning the whole. Interfaces are the price of optionality: boundary specifications that keep separation from degenerating into hidden coupling and demolition-by-default. Physical interfaces appear necessary but insufficient, since change is also governed through institutional interfaces that fix who is authorised to alter which layer, and under what conditions.

From a lifecycle perspective, separability is an investment trade-off that carries a real upfront premium. The discipline can raise initial capital cost, and the aim is to reduce the frequency of destructive renovation by making replacement and reconfiguration technically feasible within the short-life layers. Quantitative claims in this area are indicative, and we treat the figures below as illustrative magnitudes rather than settled constants. BIM-enabled lifecycle modelling has been used to compare 30-year (traditional), 50-year (semi-open), and 100-year (fully open) designs across renovation cycles.⁸¹⁸² One reported simulation identifies an upfront capital premium of approximately 20% — associated with enhanced structural capacity, raised access floors, exposed conduits, and interstitial service space — paired with whole-life cost approximately 24% lower over a 100-year horizon, and waste output roughly three times higher in the traditional comparator.⁸³ What converts higher capex into lifecycle savings is the avoided destructive renovation. The same mechanism underpins circularity and embodied-impact claims. Skeleton–Infill analyses report embodied-carbon reductions of 45–55% relative to monolithic cast-in-situ baselines, an effect attributed to extending the service life of the carbon-intensive structural frame while enabling frequent reconfiguration through replaceable infill.⁸⁴⁸⁵⁸⁶

Separability also carries an adoption logic that depends on the surrounding ecosystem, and that logic appears decisive for whether the lifecycle case becomes investable at scale. When interfaces are stable, specialised infill supply chains can emerge, and manufacturers can reliably design to known boundaries. Industrialised infill describes how standardised interfaces allow commodity services and fixtures to be bundled with project-specific prefabrication into delivered “infill packages”, with reconfiguration timelines as rapid as 10 working days for new fit-outs.⁸⁷⁸⁸ The “infill market” industrialises only on dependable interface contracts; absent them, costs revert to bespoke, high-friction site work.⁸⁹ The reasons stakeholders often withhold investment in separability are correspondingly structural, embedded in capability, governance, finance, and the existing stock. Capability is the first barrier: modular intent without detailing competence produces structural rigidity, reduced façade flexibility, and poor technical resolution, so the system loses the optionality it sought.⁹⁰ Standard-setting and governance form a second barrier, since polycentric layer separation depends on clear rights and responsibilities across levels, and on higher-level authorities recognising the governance space of lower levels under a subsidiarity-compatible structure that distributes control across loci.⁹¹⁹² Financing horizons and market structure form a third: those who pay the upfront premium may not be those who capture the long-run benefits. Retrofit constraints form a fourth, because much of the existing stock is built around embedded services and tightly coupled assemblies, so the transition cost can be high even where the lifecycle case is strong.⁹³

The bounded conclusion follows directly. Separability is a conditional investment: its payoffs are contingent on technical-interface discipline, on alignment of governance rights, and on connecting short-run cost-bearers to long-run benefit-holders, and they materialise only where those three conditions hold together. That conditional structure carries a further implication, and this implication is the load-bearing transition to Section 2.6. If optionality requires layered parts, disciplined interfaces, and explicit constraints, then those elements must be representable, and the representation must allow local re-fit options to be generated and checked under real decision conditions. The design problem becomes one of making the next feasible move legible: clarifying what may be changed now, by whom, within time, money, and coordination limits. Separation-of-concerns must therefore be specified as structured information — an engineered arrangement of layers, interfaces, and governance rights expressed as entities, relations, and constraints that support local exploration.⁹⁴⁹⁵

Representable Separation-of-Concerns Layers, interfaces, and governance rights rendered as structured information — the representational substrate on which bounded optionality (Section 2.6) is operationalised.

Supply-side architecture and interface discipline together may render local re-fit feasible in principle. What households, providers, and regulators must act on, however, is whether that nominal feasibility survives once bounded rationality, administrative burden, and coordination cost are admitted as binding constraints. Two senses of optionality have tended to be conflated, obscuring the diagnostic work that follows: nominal optionality, which counts the pathways formally available, and effective optionality, which counts the pathways an actor may plausibly exercise under real constraint. Their divergence widens in adaptation-heavy housing, where it operates as a structured, three-mechanism phenomenon governed by its own internal logic.

Nominal vs effective optionality. This distinction is load-bearing for the argument that follows and warrants careful specification. Nominal optionality enumerates the configurations, interventions, or trajectories possible in principle — the formal pathways a regulatory text, a procurement frame, or a service architecture is on record as offering. Effective optionality is the subset of those pathways an actor may actually exercise under binding cognitive, financial, administrative, and coordination constraints. The gap between the two is, in our framing, the empirically observed reduction of the feasible-next-step space. Six criteria operationalise the distinction at the level of a single move: a feasible next step is timely (implementable within the actor’s binding timeframe), permissible (with a knowable approvals pathway), affordable (within budget once transaction and compliance costs are admitted), co-ordinatable (with manageable cross-party synchronisation), legible (with stable rules and credible estimates), and reversible (with exit routes available at tolerable cost). These six criteria are provisional analytical constructs, offered as design-relevant specifications for what it would mean to expand effective optionality in practice; their reconciliation against the four critical properties developed at Section 2.9 is taken up at the appendix table referenced below.⁹⁶

Bounded rationality and the cognitive substrate. The first of three coupled mechanisms by which effective optionality falls below nominal is the boundedness of the deciding actor. Simon’s foundational claim — that real actors search and satisfice rather than optimise across the full feasible space, settling for workable options under constraint — supplies the cognitive substrate on which the remainder of the argument is built.⁹⁷ In housing, the difficulty lies in how public-service interactions impose learning, compliance, and psychological costs whose magnitude tracks the actor’s cognitive resources and falls unevenly across the population. Administrative-burden research makes the point a structural one: converting nominal entitlements into used services exacts costs that are unequally distributed and politically shaped, with the heaviest weight falling on those who hold the least bandwidth, the least institutional literacy, and the least slack to absorb a procedural setback.⁹⁸⁹⁹ Bounded rationality, on our reading, matters here less as a free-standing theory of individual choice than as the cognitive substrate through which institutional complexity is metabolised into unequally distributed access to nominally universal pathways.

Paradox of choice and overload. The second mechanism inverts the intuitive equation of more-options-with-better-outcomes, and its force is the easier to overlook for that reason: unconstrained optionality may itself function as a friction. Iyengar and Lepper’s paradox-of-choice findings established the canonical case: large option sets, beyond a threshold, appear to reduce decision quality, satisfaction, and the likelihood of follow-through, even when the additional options are nominally welfare-improving.¹⁰⁰ Tversky and Kahneman’s earlier characterisation of choice under uncertainty — heuristic, biased, and shaped by the framing of the option set itself — supplies the complementary mechanism by which large menus interact with limited evaluative capacity to produce systematically distorted outcomes.¹⁰¹ Translated into housing, the regulatory, market, and administrative environments routinely present option sets — programmes, exceptions, pathways, bespoke conditions — that may overflow household cognitive capacity, so that decision-makers defer, satisfice on impoverished criteria, or wait for the crisis that forces a choice. The reframing the overload literature affords this inquiry is, we suggest, its central one: unconstrained optionality functions as a friction in its own right, generating cost where the more-options-is-better intuition expects relief.

Administrative burden in the housing-specific register. The third register through which the bounded-rationality and overload mechanisms become visible is the empirical record of housing-specific administrative-burden research. Aiken, Ellen, and Reina document, in the context of emergency rental assistance, that documentation and verification requirements suppress take-up among precisely the households whose need is most acute and whose capacity to absorb procedural cost is lowest.¹⁰² Robinson and colleagues, working in homelessness services, characterise the cumulative effect of repeated procedural demands as exhaustion — a “burnout” produced by the institutional architecture itself, separate from the underlying housing scarcity.¹⁰³ Herd and Moynihan’s broader synthesis frames the distribution as policymaking by other means: burdens are designed allocative mechanisms whose unequal incidence is, in the end, the policy itself.¹⁰⁴ The housing-specific evidence base, taken together, may support the structural inference the section requires: the gap between nominal and effective optionality is neither incidental nor evenly distributed.

Nominal-to-effective optionality funnel The funnel from nominal pathways to effective next steps, with bounded-rationality, overload, and administrative-burden mechanisms acting as the successive narrowing constraints.

Permutation explosion as a coupling problem. The third coupled mechanism connects the present section to the modularity argument developed at Section 2.8, and it is structural where the first two are cognitive. When the options a system offers interact, “more” ceases to be additive and becomes, in the technical sense, combinatorial. A system offering N independent options yields a choice space linear in N; once those options couple combinatorially, the choice space scales as a product of option counts, producing factorial-or-worse growth that may overwhelm cognitive and computational capacity well before the option list reaches the levels familiar in housing adaptation. The housing translation is concrete: layout, services, accessibility, tenure, financing, and compliance options couple — and they do couple, because a change to one routinely conditions the feasibility of the others — to produce a combinatorial space no actor can survey or compare comprehensively. The mechanism is the structural counterpart of the behavioural overload effect: bounded rationality and paradox-of-choice describe the cognitive insufficiency of the deciding actor, while permutation explosion describes the formal intractability of the space the actor is asked to decide over. Coupling converts more options from a benefit into a burden, and the conversion is monotonic in the density of interaction. The connection to the modularity argument is direct. The patterns-of-interdependence framing developed at the next major section — interfaces, near-decomposability, and the disciplined containment of cross-module dependency — supplies, in the present section’s terms, the architectural response to the permutation-explosion mechanism specified here.¹⁰⁵

Combinatorial choice space under coupling Linear growth in the configuration space under independent options versus combinatorial growth under coupling, with the feasible subset narrowing as interaction density rises.

From three mechanisms to enabling constraint. Three coupled mechanisms — bounded rationality, paradox-of-choice overload, and permutation explosion under coupling — may together reduce effective optionality below nominal in adaptation-heavy housing. The diagnostic question the next section takes up is how this reduction is best characterised mathematically, and how the resulting account of bounded optionality as an enabling constraint may be operationalised in the architectural language the remainder of the chapter develops.

Bounded rationality, the paradox of choice, and permutation explosion together reduce effective optionality below its nominal headline count, and the next analytical step is to ask what kind of representation could characterise that reduction precisely enough to permit a design response. Two complementary moves are required. The first move is mathematical: a vocabulary in which the gap between nominal and effective optionality can be described as a structural property of high-dimensional, interaction-heavy spaces. The second move is architectural: a design stance under which constraint is reinterpreted as the mechanism that converts an unmanageable space into a navigable one. The curse-of-dimensionality literature from statistics and machine learning supplies the first; bounded optionality reread as enabling constraint supplies the second. The two moves operate together — diagnosis without remedy would leave the deficit intact, and remedy without diagnosis would lack the disciplined criterion under which a constraint could be judged constructive rather than merely restrictive.

The curse-of-dimensionality, named in Bellman’s foundational work on dynamic programming and elaborated extensively in the high-dimensional statistics and machine-learning literature, refers to a family of pathologies that arise as the number of degrees of freedom in a decision or inference problem grows.¹⁰⁶¹⁰⁷ Three symptoms recur across the literature. Sparsity appears first: as dimensionality rises, any finite sampling of the space becomes vanishingly thin relative to the volume it must cover, and points become far apart in a way that frustrates generalisation from precedent. Distance concentration appears second: under broad conditions the ratio between nearest and farthest neighbour collapses, so options that should be discriminable become difficult to rank by similarity-based heuristics. Sample-complexity escalation appears third: stable estimation and reliable rules of thumb require exponentially more observations as dimension rises, which renders flexible inference infeasible even when computational power is generous. Applied to housing-adaptation option spaces, the construct functions as an analogue rather than a literal claim — household choices are not random samples and the relevant metric is not Euclidean — but the mechanism transfers cleanly to the case at hand. As the dimension of the option space grows, with layout choices multiplying against service choices multiplying against accessibility choices multiplying against tenure choices multiplying against compliance choices, any given household’s lived experience covers a vanishingly small fraction of the space, and the signal-to-noise of option comparison degrades to the point where actor-level discrimination becomes unreliable. High-dimensional unconstrained option spaces become, on this reading, structurally uninformative for actor-level choice.

If unconstrained option spaces become uninformative under dimensional escalation, then the design response that follows cannot consist of further enumeration. The relevant move is the imposition of constraint that enables rather than constraint that forecloses — what the complex-systems literature calls an enabling constraint. The mechanism operates through three sub-moves that we name explicitly. Pruning eliminates dimensions of variation that do not reward actor-level discrimination; interfaces that fix dimensional coordination to a small modular grid, for instance, suppress an infinity of millimetre-resolution choices without foreclosing the substantive choices that matter for occupant function. Low-dimensional control reduces the actor-facing choice surface to a small set of substantively distinct options whose downstream consequences can be reasoned about within bounded cognitive load. Reliable solution arrival completes the move: an actor working within an enabling constraint can reach a feasible solution in bounded time, where the constraint substitutes for an exhaustive search that would otherwise be unattainable. Six tentative criteria specify what an enabling constraint for bounded optionality would look like in practice — approvals stability, interface explicitness, reversibility, limited co-ordination span, substitutability for atypical needs, and low administrative burden. These criteria are derived analytically from the cross-domain literature and held under explicit evidential test against the artefact-mediated demonstration of Chapter 9 and the integrated discussion of Chapter 10, where each criterion is evaluated as a property of the proposed standardisation schema, module taxonomy, and procedural-generation pipeline, with adjudication recorded against the four critical-property names of Section 2.9 and Table 2.A.1.¹⁰⁸

Enabling constraint operates at the level of the system rather than at the level of the individual actor, and this distinction is load-bearing for the argument. The constraint bounds the option space within which actor-level decisions occur; it does not eliminate actor autonomy within those bounds. Standardisation and interface discipline, treated extensively in the next section as the structural form of modularity, are enabling constraints in this precise technical sense. They fix the coordination protocol so that within-protocol decisions remain feasible and reasonable for the actors who must execute them. Where the constraint mis-specifies — by fixing the wrong dimensions, by setting the grid too coarse, by foreclosing legitimate substitutions — the option space contracts in welfare-reducing directions. Where the constraint specifies well, the option space remains large within the protocol while becoming navigable to actors operating under bounded rationality.

A criterion that the enabling-constraint framing must satisfy if it is to perform the design work asked of it is substitutability for atypical needs. The constraint must preserve the ability of actors with atypical requirements — disabled occupants, multi-generational households, households with culturally specific spatial practices — to find feasible solutions within the protocol. Constraints that work for typical actors but exclude atypical ones are, in practice, foreclosing constraints even when they appear enabling for the median case. The criterion is decisive for accessibility-housing design specifically and for any setting in which the regulatory text codifies typicality assumptions that may not survive the diversity of the population the architecture must serve. Substitutability accordingly bridges directly to the representational-governance discussion of Section 2.10, where the Governed Kernel Architecture is required to admit governed variation at the governed instance library without disturbance to the governed kernel.

Enabling constraint does not, however, secure itself without governance, and three governance risks may convert a well-specified constraint into a foreclosing one over time. Standard-setter capture is the first risk: constraints may be specified by actors whose interests do not align with those of the typical adapter, and the resulting bounds may then privilege production efficiency, regulatory clarity, or political legibility over actor welfare. Stale constraints are the second risk: a constraint set against the conditions of one era may become foreclosing when the context shifts. Dimensional standards designed for the typical-sized household of one period, for instance, become foreclosing when household compositions change. Interface decay is the third risk: standardisation requires active maintenance, and absent maintenance the drift between practice and standard may convert an enabling constraint into a foreclosing one without any explicit policy change.¹⁰⁹ These risks point, on our reading, to the need for representational governance — the disciplined maintenance of the constraint’s enabling character over time, developed at the joint reading of Section 2.8 and the meta-requirements of Section 2.11.

From dimensionality reduction to modularity — containing interactions to expand feasible local next steps. The dimensionality argument identifies the diagnosis: high-dimensional coupled option spaces are unmanageable under bounded rationality. The bounded-optionality argument identifies the design move: enabling constraint as the mechanism through which the unmanageable is rendered navigable. The next section specifies the structural form the enabling constraint takes — bounded interdependence governed by explicit interfaces, which is what the modularity literature names and analyses.¹¹⁰¹¹¹ Section 2.8 develops modularity as an abstract analytical description of bounded interdependence — interactions concentrated within bounded sub-systems, with cross-boundary effects carried by explicit, limited interface channels — and specifies the conditions under which change remains manageable across the boundary. Sections 2.9–2.10 develop the four critical properties — interoperability, transportability, manipulability, transformability — and the representational-governance problem class that emerges from the joint reading of dimensionality, enabling constraint, and modular interdependence. The transition is load-bearing for the chapter’s architecture: it converts the chapter’s diagnostic work into the analytical-and-architectural vocabulary against which the artefact suite of Chapters 5 through 9 is built and evaluated.

Modularity, as the term is used here, names an abstract analytical description of bounded interdependence in housing systems; it does not name a construction method, prefabrication technology, or deregulation programme. The lens organises a working vocabulary — bounded versus propagating interdependence, route availability versus route usability, locality of change, tractability of adjustment — supported indirectly across modularity research, product-architecture theory, network science, housing-pathways scholarship, and path-dependence work. What converges, on our reading, is one shared explanans: tight cross-domain linkage is the condition under which housing change is most often costly, and bounded interdependence through governed interfaces the condition under which it may remain practicable.

A first distinction does much of the analytical work. Route availability is the formal existence of a pathway — necessary but not sufficient for a feasible next step. Route usability is the practical possibility of exercising it under actual burdens. The distinction explains why more routes can still yield less usable adaptability: proliferation alone does not deliver feasibility when each route crosses interfaces with real costs attached. Four cost categories are kept distinct. Switching costs cover the broad costs of moving — financial charges, search effort, information costs, attachment to place — and Australian downsizing research shows how these frictions suppress adjustment even where more suitable dwellings exist.¹¹²¹¹³ Transaction costs are narrower — legal, tax, settlement, and brokerage charges. Administrative burden concerns the learning, compliance, and psychological costs of claiming assistance,¹¹⁴¹¹⁵ and housing-specific studies show that documentation requirements suppress nominal access.¹¹⁶ Coordination burden arises when change depends on synchronised action across actors; the CAPABLE trial costed that requirement at roughly US$3,300 per participant over four months.¹¹⁷

Behind that divergence lies a more fundamental variable: the pattern of interdependence among the domains a route crosses. Four overlapping but non-identical traditions converge on the diagnosis. Simon’s near-decomposability concerns the relative strength of interactions across subsystems and supplies the foundational architectural intuition.¹¹⁸ Product-architecture treats modular and integral architectures as ideal types differing in how functions, components, and interfaces are arranged.¹¹⁹ The housing-pathways framework conditions trajectory on social and institutional context across the life course.¹²⁰ Path-dependence work explains why sunk costs, routines, and institutional rules make reversal difficult over time.¹²¹ Bounded interdependence (consequences largely contained within related domains) and propagating interdependence (change in one domain forces wider adjustment) are adopted as working analytical terms; their value is diagnostic rather than taxonomic.

Modularity itself has been defined across disciplines converging on one intuition — bounded change through governed interfaces. Network science treats it as a partition in which within-group ties are denser than between-group ties.¹²² Parnas’s information-hiding principle holds that volatile design decisions should be encapsulated behind stable interfaces.¹²³ Baldwin and Clark’s visible design rules describe how stable shared interfaces permit semi-independent variation.¹²⁴ Translated into housing, modularity names a comparative pattern of organised interdependence — an arrangement is more modular where change in one domain can proceed with limited knock-on adjustment elsewhere, and less modular where modest change requires multi-domain renegotiation. The term as used here is analytical rather than technological; no claim is made that housing is literally equivalent to a product, software system, or network partition.

Near-decomposability: geometric intuition Geometric intuition of a nearly decomposable system: dense within-subsystem interactions along the diagonal blocks, weak but non-zero cross-subsystem interactions, permitting quasi-independent analysis and local change.

Two distinctions organise the interface analysis. Locality of change refers to how far consequences spread across domains. Tractability of adjustment refers to whether the work of enacting change remains within manageable bounds. Interfaces themselves take three broad forms. Physical interfaces connect components, layers, and service systems within the dwelling; the Open Building tradition shows that when long-life support is separated from short-life infill, internal layout changes can proceed without triggering structural demolition.¹²⁵¹²⁶ Administrative interfaces connect programmes, approvals, and eligibility regimes; documentation and verification rules act as decisive interfaces determining whether nominal help becomes material access. Organisational interfaces connect the actors who must sequence or align their work. Interface governance, on the polycentric reading, requires structured nesting rather than rigid centralisation or ungoverned fragmentation.¹²⁷ The critical analytical claim is that interfaces may either contain or transmit disruption. Containing interfaces limit what must cross the boundary and what must be rechecked. Transmitting interfaces require that a change in one domain be repeatedly validated, documented, or renegotiated in others — a distinction supported by Open Building’s disciplined physical boundaries and the service-interface typology in modular health-care provision.¹²⁸

Three interface types distinguished Physical, administrative, and organisational interfaces distinguished by unit of analysis, coupling failure mode, and characteristic cost structure.

Containing versus transmitting interface behaviour A containing interface absorbs a change event within its originating module; a transmitting interface propagates the event across module boundaries, generating cascading rework.

Three burden forms compound as adaptation traverses an interface sequence. Sequencing burdens arise when steps must occur in order and delays compound.¹²⁹¹³⁰ Coordination burdens arise when multiple actors must align their action.¹³¹ Cumulative compliance burdens arise when each interface adds documentation, waiting, or re-assessment. Tight coupling converts ordinary change into invasive intervention at dwelling level; fragmented assistance produces the same effect administratively. The same structure is heavier for some actors than others because cognitive resources, time, and prior procedural exposure differ across the population.¹³² The complex-systems literature supplies the formal analogue: tightly coupled architectures turn local optimisation into cascading disruption,¹³³ and systems-dynamics modelling of homelessness shows intervention effects depend on where propagation pathways are interrupted.¹³⁴

Burden cascade across the interface sequence Sequencing, coordination, and cumulative compliance burdens compound as a single adaptation initiative traverses the physical, administrative, and organisational interface sequence.

Modularity is always relative to the frame of analysis. A dwelling may exhibit bounded interdependence at the level of internal layout while exhibiting propagating interdependence at the level of tenure, finance, or service coordination.¹³⁵ Path dependence adds a temporal version: what is manageable short-term may create harder-to-reverse constraints later. Modularity is therefore a diagnostic and comparative concept within an explicit frame of domain, scale, and time — not a binary label.

The reviewed literatures support a working synthesis of six derivable conditions under which housing change is more likely to remain practicable. (1) Containment holds that consequences should remain more local than cross-domain, consistent with Simon’s near-decomposability. (2) Manageable interface load holds that the interfaces through which change passes should be explicit and stable enough that crossing them does not trigger repeated renegotiation, a point supported by both modularity and service-interface literatures. (3) Limited coordination span holds that the number of actors whose synchronised action is required should remain small enough that the route does not collapse into a coordination project in its own right, a condition the administrative-burden literature and the CAPABLE trial together evidence. (4) Incrementality holds that change should, where possible, proceed in steps rather than through whole-system commitment, with strongest support from Open Building and flexible-housing research.¹³⁶ (5) Reversibility is multi-dimensional: physical reversibility appears in adaptable-architecture work, financial reversibility in switching-cost and downsizing literatures,¹³⁷ while legal and social reversibility are less directly evidenced. (6) Absorptive capacity holds that the burden imposed by change must remain within what households and relevant institutions can manage, with its strongest conceptual grounding in the competence-press model of person-environment fit and demand for lower-burden tenure arrangements.¹³⁸¹³⁹ Support is strongest for physical containment, interface discipline, and administrative tractability; thinner for incrementality, reversibility, and absorptive capacity. Taken jointly, the six conditions distinguish routinely manageable change from change deferred until crisis intervenes.

Six conditions for manageable change The six derivable conditions under which housing change is more likely to remain practicable: containment, manageable interface load, limited coordination span, incrementality, reversibility, absorptive capacity.

The six conditions specify when change is likely to remain manageable; they do not specify what a housing representation must possess if those conditions are to become checkable governance claims. Section 2.9 names the four critical properties — interoperability, transportability, manipulability, transformability — that convert these conditions into checkable claims, which in turn translate into the propositional mechanisms (Proposition 1 through Proposition 5) developed in Chapter 3 under the Gregor-Jones design-theory anatomy.

The argument turns here from diagnostic to prescriptive. The six conditions developed in the preceding module — containment, manageable interface load, limited co-ordination span, incrementality, reversibility, and absorptive capacity — diagnose when housing change is likely to remain manageable; they do not specify what a housing representation must possess if those conditions are to operate as checkable governance claims on representational artefacts rather than as aspirational descriptions of well-behaved adaptation. The four properties named here perform that turn. Each operationalises one or more of the six conditions into a checkable claim on the representational form through which adaptation is enacted, and together with their integrated-burden complement they constitute the necessary-and-sufficient property set claimed for adaptation-heavy housing representations.

Interoperability (under semantic continuity): housing-representation meaning survives crossing actor, tool, and version boundaries without uncontrolled drift, such that semantic content can be re-instantiated by a downstream actor without the underlying obligation degrading at each handoff. The anchors are the information-hiding principle as developed at Section 2.8 and a NIST cost study quantifying the cost incurred when interoperability fails across construction-industry handoffs.¹⁴⁰¹⁴¹ The housing-domain warrant comes from documented suppression of nominal access by documentation and verification rules, which appears to treat semantic continuity as a property of the representation rather than of any single actor’s interpretive labour.¹⁴²

Transportability (under bounded verification): compliance commitments and verification obligations move through handover networks at finite, declared scope rather than as whole-system re-evaluation. The anchor is Baldwin and Clark’s design-rules logic — applied at Section 2.8 for general modularity — which applies to transportability by showing how a stable shared interface licenses semi-independent variation in compliance commitments and verification obligations specifically, rather than in design parameters at large.¹⁴³ The housing warrant follows from the administrative-burden evidence reviewed earlier: transfer between agencies, programmes, or assessment cycles forces re-documentation of facts already established elsewhere, converting nominally bounded verification into cumulative re-litigation across institutional encounters.¹⁴⁴ Transportability names the property that prevents this collapse — each handover crossing requires only the verification declared at its interface, and cumulative load may therefore remain proportional to declared scope.

Manipulability (under replay): transformation operations on the representation are inspectable, replayable, and verifiable through declared grammar rather than through narrative description of what someone took the rule to mean. The anchor is Goodman’s set of notational-adequacy criteria — disjointness, finite differentiation, semantic unambiguity — which establish what a notation must satisfy if it is to support replicable manipulation rather than interpretive improvisation.¹⁴⁵ The housing warrant returns to the containing-interface analysis: a containing interface is not merely one that absorbs disruption in some run, but one whose absorption behaviour can be predicted through the declared structure of the representation itself. Manipulability — on our reading — is what permits segmentation, substitution, and recomposition to be inspected in advance and replayed after the fact, rather than reconstructed forensically after a change has propagated through actors and tools whose interpretations may have diverged.

Transformability (under governed variation): legitimate variant evolution proceeds through a stable shared core plus rule-governed complement extension, rather than through ungoverned local divergence. The anchor is the platform-evolution argument that platform architecture, governance, and environmental dynamics co-evolve, and that the legitimacy of variation depends on whether core and complement are jointly governed by declared rules rather than allowed to drift.¹⁴⁶ The housing warrant follows from the path-dependence and fragmentation evidence in the preceding module: variant evolution within a governed platform is what may make contained change durable across cohorts of adaptations rather than achievable only in single demonstration cases.

These four properties, together with their integrated-burden complement — handled separately because it concerns cumulative load over diachronic adaptation, and taken up at Chapter 3 Section 3.5 under Proposition 5 (integrated utility under diachronic burden) and the evaluation-readiness meta-requirement (MR5) — constitute the necessary-and-sufficient property set. Property-mechanism translation under the Gregor-Jones design-theory anatomy runs one-for-one onto the Chapter 3 propositions: interoperability ↔︎ Proposition 1 (semantic interface and identity persistence); transportability ↔︎ Proposition 2 (interface-bounded modularity); manipulability ↔︎ Proposition 3 (executable transformation grammar); transformability ↔︎ Proposition 4 (platform-governed complement evolution); integrated burden ↔︎ Proposition 5. Property-to-measure mapping is set out at Chapter 4 Section 4.5.

The literatures reviewed across Sections 2.1 through 2.8 converge on a single problem class — adaptation intractability under ungoverned representational form — addressing structurally distinct failure modes: dynamic misfit accumulation under person-environment tension; supply-side rigidity enforced by monolithic representational form; reduction of nominal options to a narrower effective set under bounded rationality and administrative burden; and propagation of change under ungoverned coupling. Convergence on a common representational gap is a finding of the review, not a pre-theoretical commitment, and licenses a design science research framing in which adaptation becomes intractable when requirements, spatial units, constraints, and consequences cannot be represented in ways that keep change local, intelligible, and governable.^147148149150

Three Cycle View of Design Science Research The Three Cycle View mapped onto the present study: literature review and theoretical framework sit within the Rigor Cycle, artefact chapters within the Design Cycle, and demonstration and evaluation within the Relevance Cycle.

Two forms of knowledge constitute the scientific base for design under the Gregor-Hevner classification.¹⁵¹ Descriptive knowledge (Ω) comprises theories, observations, and empirical regularities that describe and explain phenomena; prescriptive knowledge (Λ) comprises design theories, technological rules, and artefact descriptions that specify how to achieve a goal, with Type V theory — theory for design and action — its most mature form.¹⁵² Across the reviewed literatures the Ω base appears strong — person-environment fit theory, administrative-burden analysis, and modularity theory collectively explain how misfit accumulates, options narrow, and change propagates — while the Λ base appears fragmentary, with prescriptive contributions distributed across disciplines (layering, interface discipline, modular decomposition) whose integration is the contribution we develop.¹⁵³ This Ω-rich, Λ-fragmentary asymmetry places the present contribution in the improvement quadrant of the Gregor-Hevner Knowledge Contribution Framework, with elements of exaptation through transfer of modularity principles from software and product architecture to housing adaptation.

Gregor-Hevner Knowledge Contribution Framework The combination of strong descriptive knowledge and fragmentary prescriptive knowledge places the contribution in the improvement quadrant, with exaptation through transfer of modularity principles from software and product architecture to housing adaptation.

Partial prescriptive responses exist and must be acknowledged. BIM rule-checking and automated compliance — Class 1 and Class 2 rule classification, regulatory process modelling, and the normative BIM layer — demonstrate that regulatory requirements can be formally encoded and computationally evaluated when expressed as logical predicates over well-typed building elements.^154155156157 The shared commitment in these schemes is that the semantic content of regulatory requirements has been correctly interpreted before rule-checking begins. The IFC schema (ISO 16739-1:2024) provides a standardised data model for building elements with identity, spatial decomposition, property sets, and relationship structures, operating at the geometry-level and element-level.¹⁵⁸ Open Building modular catalogues operationalise the support/infill rationale through component libraries that separate long-life structural from short-life fit-out, at the physical-component level.¹⁵⁹¹⁶⁰ The positioning that follows is one of complementary layering: BIM rule-checking and IFC provide geometry-level and element-level governance — essential infrastructure on which any computational compliance regime depends — while the contribution developed here addresses the layer beneath, namely requirements-level semantic governance that produces the interpretive inputs on which rule-checking operates. External benchmarking against IFC4 and the Building Topology Ontology is reported at Chapter 5, with results and data catalogued in the Supplementary Chapter 5 Results and Data Package.

A second body of work bears not on whether requirements-level semantic governance is needed but on the form it must take. Cognitive linguistics treats built-environment concepts not as bare objects but as construals — meanings built by elaborating schematic conceptual baselines such as space, boundary, and objecthood into more specific concepts through profiling, image-schema extension, and contextual framing.^161162163 On this account a room is bounded space construed as enclosed and habitable, a barrier a boundary construed as obstruction, and a regulatory context a frame fixing which requirements apply — a layered structure that maps onto the governance planes a representational scheme must keep distinct. The implication is that a controlled vocabulary for adaptation cannot treat all its terms as equally primitive: it must distinguish schematic primitives from the composites elaborated upon them and from the operators that perform the elaboration. This cognitive grounding is developed as theory in Chapter 3, Section 3.4 and operationalised as the artefact vocabulary in Chapter 6, Section 6.1.

Four continuity demands surface across the reviewed literatures, one in each body of work: dynamic misfit requires the conditions defining adequate housing to be carried forward as circumstances change;¹⁶⁴ layered architecture achieves physical independence between subsystems but leaves open the question of how the rationale for that separation persists across later modifications; administrative-burden analysis describes how eligibility and documentation obligations accumulate across institutional encounters; and modularity theory establishes that governed interfaces are necessary for bounded interdependence at the software and product-architecture level, which the present contribution transfers and operationalises for component boundaries and interaction conditions maintained across housing adaptation. Representational governance is the integrative capacity inferred from this convergence — the disciplined maintenance of meaning, boundaries, and permissible transformations across adaptation processes — whose meta-requirements and research questions are taken up in the next module.¹⁶⁵

The problem formulation built across Sections 2.1 through 2.8 yields four domain-level objectives that any adequate response must satisfy, together with a fifth cross-cutting requirement imposed by the research paradigm. In the Gregor-Jones anatomy these five constitute the meta-requirements of the design theory: the class of goals the proposed artefact suite must address.¹⁶⁶ They are abstract enough to apply across multiple instantiations yet specific enough to constrain what counts as an adequate response.

Meta-Requirement 1 (semantic continuity). The first objective preserves the meaning of accessibility requirements as they pass between actors, contexts, and decision points. It derives from the dynamic-requirements diagnosis at Section 2.1 and the administrative-burden mechanisms catalogued at Section 2.6: housing need is itself diachronic, and interpretive handoffs across multi-actor pipelines may degrade requirement meaning unless governance is exercised at the representational layer where meaning is carried.¹⁶⁷

Meta-Requirement 2 (bounded composability). The second objective decomposes spatial and functional units so they may be recombined without uncontrolled propagation. It derives from the coupling and propagating-interdependence reading at Section 2.8, where local change converts into systemic burden whenever interface crossings transmit rather than contain disruption — a behaviour Simon’s near-decomposability argument anticipates and the Baldwin-Clark design-rules formalism encodes as the publish-and-hide partition.¹⁶⁸¹⁶⁹

Meta-Requirement 3 (formal expressibility). The third objective represents units, relations, and constraints in forms amenable to structured reasoning and transformation. It derives from the interface-governance requirement for explicit rules and verifiable constraints — what Parnas’s information-hiding decomposition articulates at the module-boundary level — and commits any adequate response to a notational substrate that may be parsed, queried, and transformed under deterministic rules.¹⁷⁰

Meta-Requirement 4 (procedural traceability). The fourth objective makes adaptation outputs auditable from requirement through decision to documented outcome. It derives from the administrative-burden and coordination-failure mechanisms at Section 2.6: where institutional encounters accumulate eligibility, documentation, and learning costs without integration, the chain from regulatory requirement to documented outcome may fragment, and ungoverned handoff is, on our reading, the proximate mechanism by which traceability decays.¹⁷¹

Meta-Requirement 5 (evaluation-readiness). The fifth objective is cross-cutting and derives from the research paradigm rather than from any one domain literature. Design science research requires a nominal sequence of problem identification, objective definition, design, demonstration, evaluation, and communication, and the proposed response must therefore exhibit amenability to demonstration and assessment against the declared problem class.¹⁷²

The derivation chain from the four literature domains through the meta-requirements to the research questions is made explicit in Figure 2.3, where the traceability of the synthetic contribution from diagnosis to artefact-design commitment is laid out.

Figure 2.3 — Meta-Requirements Derivation Network Derivation chain from the four literature domains of Sections 2.1 through 2.8 to the five meta-requirements (Meta-Requirement 1 through Meta-Requirement 5) and onward to the five research questions (RQ1 through RQ5).

Five research questions operationalise the aim developed in Chapter 1. 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 an evidence object emitted by an artefact-chapter and consolidated in Chapter 10. The four critical-property names — interoperability, transportability, manipulability, transformability — surface alongside the propositional mechanisms Proposition 1 through Proposition 5 in each question.

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; the comparator is the current synchronous-or-bespoke workflow. Operationalisation runs through the standardisation schema’s five-dimensional decomposition of clause meaning in Chapter 5, with RQ1 settled at Section 10.4 under the interpretability measure pre-registered at 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 in Chapter 6, with RQ2 settled at Section 10.10 via the modular-fit index across the seven dwelling states; the divergent transformation scopes at the S4 fork supply the empirical content.

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. Operationalisation runs through the notation in Chapter 7; static round-trip fidelity is published in the replay-fidelity evidence (thirty-eight replay files, eighteen trace records, zero divergence) and the invariant-test evidence (twenty-four invariant tests across twelve boundary pairs).

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 (theoretical framework) and Section 6.4 (operational Rule 4 variant-inheritance instantiation); the indicator is legitimate variant diversity under shared rules. Decisive evidence comes from the S4 fork: the same envelope, module composition, and interface obligations are inherited by both branches, with S5a absorbing operational-accessibility retrofits and S5b absorbing a full SDA-Robust overlay without modification of the governed-kernel 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 that pairs a burden delta with a reliability delta against a matched-baseline workflow; the architecture’s claim is that the integrated suite should not increase verification burden for equivalent assurance quality. Operationalisation runs through the composite burden-and-reliability measure declared at Section 4.5 and reported at Section 10.17.

The five questions interlock. RQ1 and RQ2 may establish that the architecture’s commitments hold at substrate and module level; RQ3 may establish that those commitments transit through encoding without information loss; RQ4 may establish that simultaneous occupant and regulatory change are absorbed at the governed instance library without disturbing the governed kernel; RQ5 may establish that the integrated burden is warranted by the cumulative gain across the prior four. The mapping of each question to its primary chapter, the artefact discharged, and the meta-requirements addressed is laid out at Table 2.1.

A single cumulative argument has been advanced across the preceding four sections. Housing need was framed as diachronic at Section 2.1, where misfit emerged as the structurally predictable accumulation across occupant life courses against a synchronous supply side. At Section 2.4, the mismatch was located in monolithic representational form, under which any change requires re-coordinating the whole. At Section 2.6, nominal options were narrowed by bounded rationality, administrative burden, and coordination costs to a much smaller set of effective feasible moves — a narrowing structurally produced rather than reducible to individual decision-making failure. At Section 2.8, the theoretical basis for managing these dynamics was located in modularity, near-decomposability, and interface governance, and the four critical properties — interoperability, transportability, manipulability, transformability — were named as what a governed representational architecture must exhibit. What we develop here is the reading that the convergence is structural rather than coincidental.¹⁷³

Where modularity has been taken motivationally across the preceding sections, its formal operationalisation is taken up in Chapter 3. Definitional axioms grounded in a nine-hundred-and-five-record cross-domain corpus are developed at Section 3.3, alongside semantic primitives and trigram-based interface atoms. The representational architecture is stratified into three planes at Section 3.4. The propositional contract Proposition 1 through Proposition 5 and the falsifier-matrix that anchors the evaluation programme are advanced at Section 3.5. The conditions under which artefact-level commitments become testable are, in this sense, the proper subject of the chapter that follows.