Skip to content

Repairability

The scored design property of a physical artefact measuring how cheaply and reliably a failed instance can be restored by component replacement rather than whole-unit replacement — a conjunction of accessible disassembly, modularity, documentation, parts availability, and software support.

Core Idea

Repairability is the design property of a physical artefact that determines how cheaply, quickly, and reliably a failed instance can be returned to functioning condition by replacement of faulty components rather than replacement of the whole unit. The property is not a single design choice but a conjunction of concrete physical, informational, and economic commitments: fasteners that are accessible and removable without destructive force (as opposed to adhesive bonding, welding, or ultrasonic welding that prevents non-destructive disassembly); a modular architecture with standardized interfaces that allow individual components to be identified, removed, and replaced without disturbing unrelated parts; documented diagnostics and service manuals that allow a technician to localize the fault; spare parts availability at a price ratio low enough that repair is economically rational compared to replacement; and tooling requirements within the reach of the target repair population (independent shop, consumer, or field technician). The canonical composite measure is the repairability index introduced by France in 2021 and adopted in EU ecodesign regulation, which scores a product across disassembly (fastener type, layer count, specialized tool requirements), documentation (availability and completeness of repair manuals and diagnostic tools), spare-parts availability (availability period and price ratio), and software support. The design discipline that produces repairability is design for repair or design for serviceability, part of the "design for X" tradition in engineering, which treats the repair scenario as a design constraint alongside manufacturing cost, performance, and reliability. The political economy of repairability is inseparable from the property: manufacturers have systematic incentives to minimize repairability (shorter replacement cycles, captive service revenue, software-locked serialized parts) that are countered by right-to-repair regulation, independent-repair-access requirements, and eco-modulation of extended-producer-responsibility fees that reward higher repairability scores. The property is specific to the geometry and material continuity of physical artefacts — it does not transfer as a mechanism to software, biological tissue, or social institutions, where the operations of disassembly and component replacement do not apply with the same content.

Structural Signature

Sig role-phrases:

  • the physical artefact — a device whose geometry and material continuity permit disassembly and component-level replacement
  • the accessible disassembly — fasteners removable without destructive force (screws, clips, pull-tabs) rather than adhesive bonding, welding, or potting, with a low layer count and ordinary tooling
  • the modular architecture — standardized interfaces letting individual faulty components be identified, removed, and replaced without disturbing unrelated parts
  • the documentation — service manuals and diagnostics that let a technician localize the fault
  • the spare-parts availability — parts stocked over a guaranteed horizon at a price ratio low enough that repair is economically rational versus replacement
  • the software support — the absence of locks on serialized parts, so a swapped component will actually function
  • the target repair population — the parameter indexing who does the repair (independent shop, consumer, field technician), since the same device can be repairable by one and not another
  • the composite index — the scored vector (disassembly, documentation, parts, software support) rolled into one auditable repairability score
  • the repair-versus-replace verdict — the outcome read off the scores: repair economically rational, or replacement forced
  • the anti-repair counter-incentive — manufacturer incentives (shorter replacement cycles, captive service revenue, software-locked parts) that depress the scores, countered by right-to-repair regulation and eco-modulated producer-responsibility fees

What It Is Not

  • Not durability. Durability is resistance to failing in the first place; repairability governs recovery after failure. The two are independent — a device can be highly durable and barely repairable (a rugged glued enclosure), or fragile and trivially repairable (a flimsy clip-together case). Conflating them hides that a long-lived product may still be unrecoverable once something breaks.
  • Not fault tolerance. Fault tolerance keeps a system running through a failure without intervention (redundancy, graceful degradation); repairability is intervention-based recovery after the system has stopped working. One avoids the need to act; the other governs how cheaply someone can act once action is required.
  • Not modularity alone. A modular architecture is necessary but not sufficient: repairability requires modularity plus accessibility, spare-parts availability, and documentation. The load-bearing case is the modular board sealed behind glue — high on decomposition, low on disassembly — which is modular yet unrepairable. Reading modularity as enough for repairability is the characteristic error the property guards against.
  • Not a single design choice or a vague "build quality." Repairability is a conjunction of scored sub-criteria — disassembly (fastener type, layers, tooling), documentation, spare-parts availability and price ratio, software support — made auditable by the repairability index. "Well made" or "user-serviceable" names nothing measurable; each anti-repair choice (a proprietary screw, a software-locked serialized part) registers as a deliberate debit on a specific scored axis.
  • Not a substrate-spanning mechanism. Repairability is built on the geometry and material continuity of physical artefacts; the operations "disassemble" and "replace a component" do not apply with the same content to software (you edit source or ship a patch), tissue, or institutions. "Institutional repairability" is metaphor unless it actually imports the disassembly–replacement–reassembly structure; the portable skeleton is the parent set — modularity, maintenance, accessibility, design-for-lifecycle-adaptability — not this named property.

Scope of Application

Because repairability is a scored design property of physical artefacts — whether a failed instance can be cheaply restored by component replacement — rather than a causal mechanism, it applies wherever the artefact's geometry and material continuity permit disassembly and replacement. The product classes below are genuine instances of the same property (one substrate at many scales); it does not transfer to software, tissue, or institutions, where "disassemble/reassemble" has no matching content — those are metaphor, and the portable skeleton there is the parent set (modularity + maintenance + accessibility + design-for-lifecycle-adaptability).

  • Design for X / serviceability — the home discipline, with design-for-disassembly, modular product architectures, and fastener/joint selection treating repair as a design constraint.
  • Consumer electronics — iFixit teardown scoring, Fairphone's modular phone, and the right-to-repair movement, where the repairability index is most visible.
  • Environmental and circular-economy policy — the EU Right to Repair Directive, France's repairability index, and eco-modulated extended-producer-responsibility fees rewarding higher scores.
  • Automotive — independent-repair access (Massachusetts Right to Repair), OBD-II diagnostic standardization, and third-party parts/service ecosystems.
  • Aerospace and medical devices — line-replaceable units, field-replaceable modules, and MRO economics, where replacement is uneconomic so repairability is engineered in.
  • Public infrastructure — roads, bridges, and water systems designed for inspection and repair as a lifecycle-cost lever.

Clarity

Naming repairability collapses a diffuse and unfalsifiable design conversation — "good build quality," "user-serviceable," "long-lasting" — into a conjunction of measurable sub-criteria: disassembly step count, fastener type, spare-parts price ratio, documentation completeness, software support period. The repairability index is exactly this move made operational: it forces the vague virtue to resolve into scores a regulator can audit and a buyer can compare. The sharper consequence is that anti-repair design stops being invisible. A glued enclosure, a proprietary screw, a software-locked serialized part — once repairability is a named and scored property, these read not as neutral engineering choices but as deliberate optimization against repair, which is precisely what makes the political economy of captive service revenue legible as a design decision rather than an accident of manufacturing.

The concept also sharpens repairability against the family of lifecycle properties it is constantly conflated with. Durability is resistance to failing in the first place; repairability is what governs recovery after failure — a device can be highly durable and barely repairable, or fragile and trivially repairable. Fault tolerance keeps a system running through failure without intervention; repairability is intervention-based recovery. Modularity is the architectural decomposition into swappable parts, but repairability requires modularity plus accessibility, parts availability, and documentation — a modular board sealed behind glue is modular and unrepairable. Holding these apart lets a designer ask the load-bearing question — not "is this well made?" but "when a component fails, can the target repair population economically restore function?" — and ties the answer directly to lifecycle cost, e-waste, and right-to-repair obligations that the diffuse vocabulary could never reach.

Manages Complexity

Whether a failed product is worth repairing rather than discarding depends, in full, on a tangle of incommensurable particulars: the joining method at every interface, the layer count and disassembly path to each component, the tools a given repair population owns, the price and stocking horizon of each spare, the completeness of diagnostics, the software locks on serialized parts — and these vary product by product, component by component, and repairer by repairer. Adjudicating it case by case means re-deriving the whole repair scenario for every device and every fault. Repairability compresses that tangle by naming a single design property and resolving it into a fixed, scored set of sub-criteria — disassembly (fastener type, layers, special tools), documentation, spare-parts availability and price ratio, software support — that the repairability index makes into one auditable composite. The regulator, designer, and buyer then track a small vector of scores, or even one rolled-up number, instead of the unbounded physical-economic detail of each repair, and read the qualitative outcome — repair economically rational, or replacement forced — off that compressed measure. The compression is what makes the otherwise-invisible legible as a small set of decisions: each anti-repair choice (glued enclosure, proprietary screw, locked serialized part) now registers as a movement on a specific scored axis rather than a neutral detail buried in the bill of materials, so "optimization against repair" becomes a readable pattern in the scores rather than a suspicion. The same compression sorts repairability cleanly against the lifecycle properties it is constantly fused with, via a small branch structure rather than case analysis: durability is failure-resistance (whether the artifact fails), repairability is post-failure recovery (whether it can be restored), fault tolerance is uninterrupted operation through failure (no intervention), and modularity is mere architectural decomposition — so a device can sit at any combination of these, and the designer reads which property is at issue from which axis is in play (a modular board sealed in glue scores high on decomposition and low on disassembly, and is therefore unrepairable despite being modular). And because the property is scored, the downstream couplings the diffuse vocabulary could never reach become functions of the same small vector: lifecycle cost, e-waste volume, and right-to-repair compliance all read off the disassembly-documentation-parts-software scores. So a per-device, per-fault adjudication spanning physical geometry, parts economics, and service politics reduces to tracking a handful of scored axes, from which the repair-versus-replace outcome, the anti-repair design pattern, and the lifecycle and regulatory consequences are all read rather than reconstructed.

Abstract Reasoning

Repairability licenses reasoning moves a product designer, regulator, or repair technician runs on any physical artefact, all conducted on a small vector of scored axes — disassembly, documentation, spare-parts availability and price ratio, software support — that stands in for the unbounded physical-economic detail of an actual repair.

The economic-rationality move runs from the scored axes to a repair-versus-replace verdict for a given fault. The reasoner asks not "is this well made?" but "when a component fails, can the target repair population economically restore function?", and computes the answer from the axes: the disassembly path and tooling determine labor, the spare-parts price ratio against the cost of a new unit determines whether repair is cheaper than replacement, the documentation determines whether the fault can even be localized, and the software locks determine whether a swapped serialized part will function. The prediction is binary — repair economically rational, or replacement forced — and it reads off the compressed measure rather than a fresh case analysis, with the target repair population a parameter: the same device may be repairable by an independent shop with specialized tools and unrepairable by a consumer, so the verdict is indexed to who is doing the repair.

The diagnostic move reads anti-repair intent off the scores. Once repairability is a named and scored property, the reasoner treats a glued enclosure, a proprietary screw, or a software-locked serialized part not as a neutral engineering detail buried in the bill of materials but as a deliberate movement on a specific scored axis — a debit on disassembly, on parts availability, on software support. This licenses an inference from the pattern of low scores to optimization against repair, which in turn makes the political economy legible: a product that scores low precisely on the axes that gate independent service is read as engineered for captive service revenue or shortened replacement cycles, not as an accident of manufacturing. The move is to infer the manufacturer's incentive from where the scores were sacrificed.

The boundary-drawing move sorts repairability against the lifecycle properties it is constantly fused with, by asking which axis is in play, and it guards a characteristic error. The reasoner distinguishes durability (resistance to failing in the first place — a property of whether the artefact fails) from repairability (recovery after failure), fault tolerance (running through failure without intervention) from repairability (intervention-based recovery), and modularity (architectural decomposition into swappable parts) from repairability (which requires modularity plus accessibility, parts, and documentation). The load-bearing case is the modular board sealed behind glue: it scores high on decomposition and low on disassembly, so the reasoner predicts it is unrepairable despite being modular — the error the distinction prevents is reading modularity as sufficient for repairability. A device can occupy any combination of these properties, and the reasoner identifies which one is at issue from which axis the question touches.

The interventionist move treats each axis as a lever and predicts the downstream effect of moving it. Because lifecycle cost, e-waste volume, and right-to-repair compliance all read off the same disassembly-documentation-parts-software vector, the designer or regulator forecasts the consequence of a design change as a movement on a specific axis: replacing adhesive with a pull-tab raises the disassembly score, which is predicted to raise independent-repair rates and extend service life; publishing the service manual raises documentation; shortening the guaranteed spare-parts horizon lowers parts availability and is predicted to force replacement once stocks lapse. The reasoning runs from a scored-axis change to its predicted effect on repair behavior and on the coupled lifecycle and regulatory outcomes, so the policy lever (eco-modulation of producer-responsibility fees rewarding higher scores, mandated parts availability) and the design lever (fastener choice, interface standardization) are reasoned about as operations on the same small vector rather than as separate, incommensurable interventions.

Knowledge Transfer

Within engineering design-for-X and its adjacent product and policy fields the property transfers as mechanism, and what carries is the whole apparatus: the scored-axis vector (disassembly, documentation, spare-parts availability and price ratio, software support), the repair-versus-replace verdict indexed to the target repair population, the diagnostic that reads anti-repair intent off where the scores were sacrificed, and the interventionist treatment of each axis as a lever with downstream lifecycle and regulatory consequences. The precondition is a physical artefact whose geometry and material continuity permit disassembly and component replacement, and across product classes each is a genuine instance of the same property rather than a likeness — consumer electronics (iFixit scoring, Fairphone's modular phone), automotive independent-repair access (OBD-II standardization, parts catalogues), aerospace and medical line-replaceable units and MRO economics, and public infrastructure designed for inspection and repair as a lifecycle-cost lever. The transfer is mechanism-preserving because these are one substrate — physical artefacts with the disassembly/replacement operations — at many product scales, so the index, the verdict, and the levers carry intact, not by analogy.

Beyond physical artefacts the report is twofold, and the boundary is sharp because the property is built on physical structure. (1) The named property does not transfer as a mechanism to software, biological tissue, or social institutions: the operations "disassemble" and "reassemble" do not apply there with the same content — software "repair" is editing source, deploying a patch, opening an API, a structurally different operation — so calling those "repairability" is metaphor, and "institutional repairability" is metaphor unless it actually imports the disassembly–replacement–reassembly structure rather than just the vibe. Such uses should be marked as analogy. (2) The genuinely portable content is one level up and is already owned by parent primes that travel across substrates in their own right: the general skeleton — engineer the artefact so that intervention-to-restore-function is locally accessible and cheap — decomposes into modularity (decomposition into independently swappable parts), maintenance (sustained intervention that keeps intended function intact against degradation), accessibility (the cost of reaching the point of intervention), and design_for_lifecycle_adaptability (planning for change over the lifecycle). Stripped of physical-artefact context, repairability is "modularity for replacement plus easy access for intervention," so the cross-domain lesson should carry those parents, not the name "repairability," whose distinctive cargo (the repairability index and iFixit scoring, fastener-type tables and parts-availability schedules, and the manufacturer/independent-shop/consumer political economy of right-to-repair and post-warranty access) is mechanical-engineering-and-consumer-protection furniture that does not and should not travel. Mechanism within physical-artefact design (one substrate at many product classes); metaphor beyond, with the genuine portable skeleton resident in the parent primes (modularity + maintenance + accessibility + design-for-lifecycle-adaptability) it instantiates rather than in this named property. This is exactly the boundary Structural Core vs. Domain Accent draws.

Examples

Canonical

France's indice de réparabilité, mandatory since January 2021, is the defining operationalization. A product (initially five categories, including smartphones and laptops) is scored across five criteria, each worth 20 points for a 100-point total that is then divided by 10 to yield a 0–10 label: documentation availability, ease of disassembly (fastener type, tools, steps), spare-parts availability period, spare-parts price ratio, and a category-specific criterion (for phones, software-update support). A smartphone might score documentation 12/20, disassembly 15/20, parts availability 18/20, parts price 8/20, and software 14/20 — summing to 12 + 15 + 18 + 8 + 14 = 67, hence 6.7/10, printed at the point of sale so buyers and regulators can compare.

Mapped back: The five criteria are exactly the composite index rolling up the accessible disassembly, the documentation, the spare-parts availability, and the software support into one auditable number. The low parts-price sub-score (8/20) is a visible debit on one axis — the kind of anti-repair counter-incentive the scoring is designed to expose — and the 6.7 result feeds the repair-versus-replace verdict a prospective repairer would compute.

Applied / In Practice

The Fairphone line is the clearest field deployment of design-for-repair. The repair community iFixit awarded the Fairphone 3 a perfect 10/10 repairability score: its back cover comes off by hand, the battery is user-swappable without tools, and major modules (display, camera, USB port, speaker) are held by standard Phillips screws and clearly labeled, replaceable individually from Fairphone's own parts store. This lets an ordinary owner, not just a specialist shop, restore a cracked screen or dead port at modest cost — extending service life against the shorter replacement cycles typical of sealed, glued flagship phones.

Mapped back: The phone is the physical artefact engineered for accessible disassembly (hand-removable cover, Phillips screws, no adhesive) and a modular architecture of individually swappable modules, backed by spare-parts availability through the maker's store. Crucially the target repair population is widened to the consumer — the perfect score means the repair-versus-replace verdict comes out "repair rational" even for a non-technician, the direct inverse of the anti-repair counter-incentive.

Structural Tensions

T1: Composite index versus conjunctive reality (the rolled-up number hides the binding axis). Rolling disassembly, documentation, parts, and software support into one auditable score is the property's great legibility win — a regulator can audit it and a buyer can compare it. But the repair-versus-replace verdict is effectively conjunctive, gated by the weakest axis: a phone that scores well overall yet locks serialized parts in software, or prices spares above a new unit, is unrepairable in practice no matter how high the composite. An additive index (France sums five 20-point criteria) treats the axes as substitutable when they are not, so a strong total can mask the one debit that actually forces replacement. The scalar that makes repairability comparable is precisely what can conceal the constraint that decides an individual repair. Diagnostic: Does the composite score survive inspection of its lowest axis, or is a single gating debit (parts price, software lock) being averaged out of view?

T2: A property of the artefact versus a relation to the repairer (the verdict is indexed to who). Repairability is presented — and printed at the point of sale — as a scored property of the device itself. Yet the repair-versus-replace verdict is explicitly indexed to the target repair population: the same phone can be repairable by an independent shop with specialized tools and unrepairable by a consumer, or repairable only by the OEM holding the software keys. So what looks like a fixed attribute of the object is really a relation between the artefact and a particular repairer's tools, skills, and access, and a single label collapses that relativity into one number that is true for some populations and false for others. Diagnostic: Repairable by whom — is this score computed for the OEM, an independent shop, or the end consumer, and does the intended population actually clear every axis?

T3: Anti-repair intent versus innocent engineering tradeoff (reading motive off a low score). Making anti-repair design legible is one of the property's sharpest contributions: a glued enclosure or a proprietary screw reads as a deliberate debit engineered for captive service revenue rather than a neutral detail in the bill of materials. But the scores measure outcome, not motive, and many repairability-lowering choices are genuine constraints — adhesive for water resistance and thinness, welds for structural strength, integration for performance or cost. Inferring intent from where the scores were sacrificed conflates constraint with malice; yet refusing that inference lets deliberately captive design hide behind the language of engineering necessity. The diagnostic's political-economy power and its capacity to misattribute motive are the same move. Diagnostic: Is this low-scoring choice traceable to a hard performance or safety constraint, or does it depress exactly the axes that gate independent service while sparing those that do not?

T4: Design-for-repair versus the objectives it competes with (repairability is one design-for-X among rivals). The property treats the repair scenario as a design constraint — but the tradition it belongs to sets it alongside manufacturing cost, performance, reliability, and durability, and these genuinely conflict. Accessible disassembly (removable fasteners, service seams) can cost water resistance, ruggedness, thinness, and unit cost; the very seams that let a technician in are the ones water and dust get through. So "engineer in repairability" is not a free lifecycle virtue but a trade against the primary product, and maximizing the repairability score can degrade the thing being made repairable. Even the lifecycle payoff is contestable: a durable sealed unit may generate less e-waste than a repairable-but-flimsy one that keeps needing repair. Diagnostic: Does raising the repairability axis here cost a primary objective (sealing, strength, cost, weight), and does the lifecycle math still favor repair once that cost is counted?

T5: Design-time score versus time-decaying, firm-controlled repairability (the property does not hold still). Repairability is scored as though it were fixed by the artefact's geometry at design. But a large part of the vector is neither geometric nor permanent: the guaranteed spare-parts horizon lapses, software-update support sunsets, serialized-part authorization can be revoked, and documentation can be pulled offline — all post-sale decisions the manufacturer controls after the score is printed. A phone that scored 10/10 at launch can become effectively unrepairable years later with no change to a single screw, because the parts and software that the disassembly relied on were withdrawn. So a "design property" is in practice a time-varying outcome partly held hostage to the maker's continuing choices. Diagnostic: Is this repairability sustained by the physical design alone, or does it depend on parts stocking, software authorization, and documentation the manufacturer can withdraw after purchase?

T6: Autonomy versus reduction (a scored engineering property or its parent primes). Repairability is a specific, operationalized property with its own cargo — the repairability index and iFixit scoring, fastener-type tables, parts-availability schedules, and the right-to-repair political economy — genuinely useful for diagnosing a physical artefact in situ, and it does not transfer as mechanism to software, tissue, or institutions, where "disassemble and replace a component" has no matching content. But its portable skeleton is one level up, owned by parents that travel in their own right: modularity, maintenance, accessibility, and design_for_lifecycle_adaptability. Stripped of physical-artefact context, repairability just is modularity-for-replacement plus cheap access for intervention, so "institutional repairability" is metaphor unless it actually imports the disassembly–replacement–reassembly structure. Diagnostic: Resolve toward the parent primes (modularity, maintenance, accessibility, design-for-lifecycle-adaptability) when carrying the lesson past physical artefacts; toward the named property when scoring a real device's repair scenario in situ.

Structural–Framed Character

Repairability sits in the mixed band of the spectrum — anchored to structure by a genuine physical substrate (the geometry and material continuity that let a component be swapped) but pulled toward frame by being a scored design desideratum wrapped in a political economy, not a fact of nature. On evaluative_weight it leans mildly framed: the underlying measure is descriptive (disassembly step count, spare-parts price ratio, fastener type), but the concept is constituted as a virtue to be maximized — the repairability index exists to reward and penalize, and calling a device "low-repairability" is close to a debit filed against it, read (per T3) as anti-repair intent engineered for captive service revenue. It is not the pure verdict of a fallacy, but it is not the flat neutrality of an allometric slope either; the score carries a normative pull. On human_practice_bound it is genuinely mixed: the operations "can you get the fasteners out without destroying the enclosure" and "is the architecture modular" are observer-independent facts of the artefact's physical structure that hold whether or not anyone scores them — but the property's payoff quantity, how cheaply and reliably a failed instance can be restored by component replacement, folds in economics (price ratio, spare-parts horizon) and a target repair population (independent shop, consumer, OEM) that are constituted by human repair practices and dissolve without them. Half the vector is geometry; half is a relation to a repairer and a market.

Institutional_origin points firmly framed: the entry as an operationalized, comparable property is an artifact of a tradition and its instruments — the France-2021 indice de réparabilité, EU ecodesign regulation, iFixit teardown scoring, the design-for-X discipline, and the right-to-repair legal apparatus — a scored metric someone invented, layered on top of the physical facts, not a regularity found in the world. Vocab_travels points framed: the operative vocabulary — the composite index, fastener-type tables, parts-availability schedules, software-locked serialized parts, eco-modulated producer-responsibility fees — is pinned to physical-artefact design and consumer-protection policy and does not float free. And import_vs_recognize is bimodal and unusually sharp: within physical artefacts (consumer electronics, automotive, aerospace line-replaceable units, infrastructure) reuse is genuine mechanism-recognition — one substrate at many scales, the index and verdict carrying intact — while beyond physical artefacts (software "repair," tissue, "institutional repairability") the operations have no matching content, so reuse is flatly metaphor, and the entry insists it be marked as such.

The portable structural skeleton is engineer the artefact so that intervention-to-restore-function is locally accessible and cheap — modularity-for-replacement plus cheap access. That skeleton genuinely travels — but it does not make "repairability" itself portable, because it is exactly what the entry instantiates from its parent primes, modularity, maintenance, accessibility, and design_for_lifecycle_adaptability, not what makes the scored property travel: the cross-domain reach belongs to those umbrellas, while repairability's own distinctive cargo — the index, the fastener tables, the parts-price schedules, the right-to-repair political economy — is mechanical-engineering-and-consumer-protection furniture that stays home on physical artefacts. Its character: structural at its physical core — a real, substrate-anchored fact about whether matter can be non-destructively disassembled and its parts swapped — but constituted as an invented, mildly value-laden design score wrapped in an economics-and-policy overlay that pins it to physical-artefact design, leaving it mixed rather than a neutral natural mechanism or a free-floating prime.

Structural Core vs. Domain Accent

This section decides why repairability is a domain-specific abstraction and not a prime, and it also carries the case for why it is domain-specific — so it is worth being exact about what could lift off physical artefacts and what stays pinned to them.

What is skeletal (could lift toward a cross-domain prime). Strip the physical-artefact context and a thin relational structure survives: engineer the thing so that intervention-to-restore-function is locally accessible and cheap — decomposition into independently swappable units plus low-cost access to the point of intervention. The portable pieces are abstract — a decomposable structure, a cheap path to the failed part, sustained intervention against degradation, and design planned for change over a lifecycle. That skeleton genuinely travels, which is exactly why the entry decomposes into modularity (decomposition into swappable parts), maintenance (intervention that keeps intended function intact against degradation), accessibility (the cost of reaching the point of intervention), and design_for_lifecycle_adaptability (planning for change) — parents that hold across substrates in their own right. Stripped of physical structure, repairability just is modularity-for-replacement plus cheap access. But it is the core repairability shares with those parents, not what makes it distinctive.

What is domain-bound. Everything that makes the concept repairability in particular is built on the geometry and material continuity of physical artefacts and their political economy: accessible disassembly (fasteners removable without destructive force, versus adhesive bonding, welding, potting; layer count; tooling); spare-parts availability over a guaranteed horizon at a workable price ratio; documentation and diagnostics; software support on serialized parts; the target repair population (OEM, independent shop, consumer); the composite repairability index (France's indice de réparabilité, iFixit teardown scoring); and the right-to-repair, EU-ecodesign, and eco-modulated-producer-responsibility apparatus. These are the worked vocabulary, instruments, and empirical cases (Fairphone's 10/10, a glued flagship's debits) of mechanical-engineering-and-consumer-protection practice. The decisive test the entry itself insists on: the operations "disassemble" and "replace a component" have no matching content in software (you edit source or ship a patch), tissue, or institutions, so remove the physical substrate and "institutional repairability" is metaphor unless it actually imports the disassembly–replacement–reassembly structure rather than just the vibe.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose transfer is recognition of the same mechanism, not analogy. Repairability's transfer is bimodal and its boundary is unusually sharp. Within physical artefacts it travels as mechanism across consumer electronics, automotive, aerospace and medical line-replaceable units, and public infrastructure — one substrate at many scales, so the index, the repair-versus-replace verdict, and the design levers carry intact, not by likeness. Beyond physical artefacts the operations lose their content entirely, so the reuse is flatly metaphor, and the entry insists it be marked as such. What genuinely reaches other substrates is not repairability but the parent set — modularity, maintenance, accessibility, design_for_lifecycle_adaptability — that it instantiates. So the cross-domain reach belongs to those umbrellas; repairability clears the domain-specific bar for physical-artefact design, while its portable skeleton is already carried, in more general form, by the primes it composes, and its distinctive cargo (the index, the fastener tables, the parts-price schedules, the right-to-repair political economy) should stay home.

Relationships to Other Abstractions

Local relationship map for RepairabilityParents appear above the current abstraction, mutual partners to the right, and children below. Node labels state whether each abstraction is prime or domain-specific; colors identify relation types.RepairabilityDOMAINPrime abstraction: Design for Lifecycle Adaptability — is part ofDesign for Life…PRIMEPrime abstraction: Maintenance — is part ofMaintenancePRIMEPrime abstraction: Modularity — is part ofModularityPRIME

Current abstraction Repairability Domain-specific

Parents (3) — more general patterns this builds on

  • Repairability is part of Design for Lifecycle Adaptability Prime

    Repairability contains lifecycle adaptability because the original design must preserve a feasible path for later diagnosis, access, replacement, and support.

  • Repairability is part of Maintenance Prime

    Repairability contains Maintenance as the intervention that restores intended function after degradation or component failure.

  • Repairability is part of Modularity Prime

    Repairability contains Modularity because a failed component must be independently identifiable, removable, replaceable, and reconnectable without replacing the whole artifact.

Hierarchy paths (5) — routes to 4 parentless roots

Not to Be Confused With

  • Reliability. The probability that an artefact keeps functioning over a period without failing — a statement about the rate and likelihood of failure, quantified as MTBF or failure-in-time. Repairability is silent on whether a device fails and governs only recovery once it does; a highly reliable unit that rarely fails can still be unrepairable when it finally does. Tell: is the quantity how often/likely the artefact fails (reliability) or how cheaply it is restored after a failure (repairability)?
  • Maintainability / serviceability. The broader RAMS attribute covering all upkeep — scheduled inspection, preventive maintenance, calibration, and fault diagnosis, not just component replacement after a breakdown. Repairability is the post-failure, replace-the-faulty-part slice of this, scored by disassembly, parts, documentation, and software. Maintainability is the whole upkeep envelope; repairability is the corrective-recovery part of it. Tell: is the concern all sustained upkeep including preventive service (maintainability), or specifically restoring function by swapping a failed component (repairability)?
  • Upgradability. The ability to improve or extend an artefact by swapping in better components (more RAM, a faster module). It rides on the same accessible-disassembly-plus-modularity substrate, but its goal is enhancement, not restoration of lost function — an upgrade replaces a working part with a better one; a repair replaces a failed part with an equivalent. Tell: is the swap aimed at making a functioning device better (upgradability) or returning a broken one to its original function (repairability)?
  • Recyclability / design for disassembly (for recycling). Designing the artefact so that at end-of-life its materials can be separated and recovered. It shares the disassembly operation with repairability but points at material recovery after the product is discarded, whereas repairability aims to keep the product in service. A design optimized for shredding-and-sorting need not permit non-destructive, component-preserving repair. Tell: is disassembly in service of recovering raw materials at end-of-life (recyclability) or of restoring and preserving the working artefact (repairability)?
  • Right to repair. The legal-and-policy movement — mandated parts access, independent-repair provisions, anti-tethering rules — that pressures manufacturers to enable repair. It is the political-economy apparatus that rewards and enforces the property, not the scored design property itself; a device's repairability score is a fact about the artefact, whereas right-to-repair is regulation acting on the firm. Tell: is this a rule or entitlement compelling access (right to repair) or the measured design attribute of the device (repairability)?
  • The lifecycle-design primes it instantiates (modularity, maintenance, accessibility, design_for_lifecycle_adaptability). The substrate-free skeleton — decomposition into swappable units plus cheap access for intervention over a lifecycle — that repairability composes from its parents. These travel across substrates in their own right (software, organizations); repairability is the physical-artefact-pinned specialization, its cross-domain reach owned by those umbrellas and treated more fully there. Tell: strip away physical geometry, the fastener/parts/software index, and the right-to-repair economics — if a substrate-free "make restoration accessible and cheap" pattern remains, that is the parent set, not repairability (a use like "institutional repairability" is those parents by metaphor).

Neighborhood in Abstraction Space

Repairability sits in a sparse region of the domain-specific corpus (97th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Software Decay & Debugging (10 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-07-12