Research and version note This is a version 0.1 research draft in Who Maintains the World? It proposes an initial framework for maintenance, visibility, and authority rather than a settled philosophy of maintenance. Later articles on infrastructure, labour, ecology, and AI will test and revise that framework.
Who Maintains the World? · Article 1
Quick read
Water arrives when a tap is opened, a lift stops at the correct floor, and an electronic payment completes in seconds. Users encounter a function, not the sampling, cleaning, calibration, patching, inspection, and handover that continually make it possible. Successful maintenance usually produces no event worth reporting. Failure has a date, a cost, and identifiable consequences. The more reliable an ordinary service becomes, the more easily it is mistaken for part of the natural background.
This invisibility is not merely inattentiveness. A new project has a launch, photographs, and a capital budget. Maintenance consists of dispersed interventions whose benefit appears as longer service life or an accident that did not occur. Budget systems may classify construction as investment while treating upkeep as recurring expenditure that can be compressed. Organisations then reward people who introduce a visible object and interpret continued availability as if nothing changed.
Maintenance is not the preservation of an original state. Water treatment must respond to changing source conditions, software patches alter code, and care must respond to a person’s changing needs. What continues is often a capability rather than unchanged material. Refusing to modify a system after its operating conditions have changed may destroy the capability that maintenance was supposed to protect.
Maintainers also acquire knowledge through repeated contact: unusual sound, smell, trends, complaints, and local differences not fully captured in a manual. Experience is not infallible, and standards are not complete copies of reality. Reliable maintenance needs records that make experience testable and procedures that remain open to field feedback.
Visibility should therefore be designed without turning every quiet intervention into a publicity exercise. Condition trends, repeat defects, staffing, deferred tasks and near misses can make maintenance needs legible. Their interpretation still needs care: a low failure count may reflect sound upkeep, weak reporting or luck. Measures become useful when they support inquiry and action, not when they merely decorate a performance dashboard.
The series begins with a limited claim. Normal operation should not be treated as evidence that upkeep is unnecessary. Institutions should make the maintained capability, responsibility chain, material risks, backlog, and exit conditions intelligible, while ensuring that people who first detect deviation have credible reporting and escalation paths. Recognition does not imply that every existing structure deserves preservation. Whether a system operates reliably and whether it should continue are distinct questions.
When nothing happens
The most persistent difficulty in understanding maintenance comes from its ordinary form of success. An inspection detects that a seal is deteriorating and the part is replaced during planned downtime; no leak follows. A database administrator repairs the backup process and disaster recovery is never invoked. A care worker adjusts routines so that a person does not become dehydrated or fall. The organisational record may contain a short work order. Everyone else experiences an uneventful day.
Accidents offer a clear narrative: normality, a moment of failure, and consequential harm. Maintenance runs against that sequence. It interrupts a path that was never completed, so its outcome is partly counterfactual. We ask what would probably have happened without inspection, cleaning, or repair. That alternative cannot be observed directly, and any avoided accident can be dismissed by saying that it might never have happened.
Maintenance outcomes are nevertheless open to evidence. Asset condition, interruption rates, time between failures, leakage, health outcomes, and user complaints can all be examined. Historical comparisons or controlled studies can evaluate some preventive interventions. The problem is not total unknowability. Evidence has to be collected over time and is rarely concentrated in a single visible achievement. It is distributed across readings that did not cross alarm thresholds, defects that did not worsen, and hours of service that were not interrupted.
The absence of an incident therefore has at least two interpretations. It may show that protection worked, or it may mean that a latent risk has not yet materialised. A mature institution cannot infer safety from quietness alone. It should be able to describe which barriers, observations, and corrective activities make ordinary quiet credible.
Why construction gains a story more easily than upkeep
A new project can usually be bounded. It has approval, design, construction, and opening. Expenditure is grouped under a project and political credit can be associated with a particular administration. Maintenance has no equally neat beginning or conclusion. A water network contains pipes installed in different decades; some components operate continuously while others wait for emergencies. Servicing, renewal, replacement, and expansion overlap.
Accounting and budget categories can intensify this asymmetry. Capital expenditure is commonly understood as creating an asset, whereas maintenance appears in annual operating costs. Those distinctions serve legitimate purposes, but they can redirect attention. Reducing this year’s maintenance may immediately improve a short-term figure. The loss appears later as a probability of failure, shortened asset life, or more expensive repair. A decision-maker compares a certain present expense with an uncertain future problem and may prefer the latter even when total public cost rises.
Victoria’s Asset Management Accountability Framework places planning, acquisition, operation and maintenance, and disposal within one asset lifecycle. It requires public bodies to establish maintenance systems, monitor asset condition, and respond to performance failure. The significance of this framework is not that it produces another set of documents. It corrects the impression that delivery completes an asset. Public value comes from a service, and service has to continue through the much longer operational phase.
Lifecycle language can itself become administrative theatre. An organisation may show that it has a strategy, register, and workflow without knowing the condition of critical assets, whether maintenance resources are sufficient, or how warnings reach people with budget authority. Formal accountability becomes meaningful only when it returns to operational action.
What is actually being maintained?
Repairing a chair appears to preserve an object. In a water system, the object is no longer a single pipe. Pipes may be replaced, pumps rebuilt, and treatment methods changed while safe supply continues. Insisting on every original component would eventually prevent the system from working.
This suggests a distinction between material and functional continuity, although “function” is often still too narrow. A public records system must do more than keep files technically readable; it should preserve evidential integrity, provenance, and intelligibility. Care cannot be assessed only through bodily function because autonomy, dignity, and social participation matter to the person receiving it. An ecosystem includes changing populations, processes, and relationships rather than one output held at a fixed level.
Any claim about maintenance must therefore identify the value that ought to continue. This is a normative question, not a specification that technicians can settle alone. Engineering can describe deterioration and ways of reducing failure. It cannot independently decide which communities a railway should serve or whether an inherited administrative procedure remains just.
Maintenance also changes its object. A security patch modifies software behaviour; accessibility work changes how a building is used; river restoration rearranges relationships between people and water. Change is not the opposite of upkeep when it serves an explicit value and remains open to later review. Under altered conditions, superficial constancy can instead amount to neglect.
Why knowledge gathers at points of contact
Part of maintenance knowledge can be codified. A manual specifies torque, water-quality guidance sets monitoring methods, and change-management systems record software versions. Standards permit coordination across shifts and sites while making judgments reviewable. Without records, experience leaves with staff. Without shared procedures, an organisation cannot readily determine whether an adjustment is safe.
Other knowledge develops through repeated contact. Operators may recognise a slight change in vibration. A cleaner sees the first leak or mould growth before a manager making an occasional inspection. A carer notices appetite or behaviour departing from an individual’s ordinary pattern. Such field knowledge is not mysterious. Its basis is frequency of observation, contextual memory, and short feedback distance. People close to an object receive a resolution that cannot be reproduced in a quarterly dashboard.
Knowledge and authority, however, are often situated in different places. A person can file an anomaly but cannot stop the equipment. Contractors may know why a fault keeps returning but be unable to alter procurement specifications. Complaints are counted without entering redesign. The organisation seems to collect information, yet it only permits information to move upwards; decisions and resources do not necessarily travel back.
Experience can be wrong. Familiarity may normalise deviation, local staff may miss interactions elsewhere in the system, and inherited practice may no longer fit new conditions. A sound arrangement does not place field experience above standards. It supports reciprocal correction. Manuals constrain arbitrary judgment, records permit comparison, and anomaly reports can initiate procedural change. Maintenance knowledge forms through this traffic rather than existing completely in any one mind.
Concentrated risk and limited authority
The benefits of a maintained system are often widely distributed while risks concentrate among people who contact faults, contamination, and work performed at night. Network crews enter confined spaces. Sanitation work can involve pathogens and toxic gases. Care workers carry physical and emotional loads. Software maintainers may respond to security vulnerabilities under pressure from users around the world. These occupations differ too much to be treated as one labour condition, yet a recurring asymmetry is visible: continuity is an organisational objective, while some costs and dangers are pushed towards the end of a contracting chain or into households.
Calling workers heroes can strengthen this arrangement. The term praises an individual for compensating for scarcity without requiring an explanation of staffing, training, equipment, or rest. Once a crisis ends, extraordinary labour is narrated as dedication and the design remains intact. Stability that continually consumes health and attention is not costless. It has translated system failure into exhaustion and injury.
Authority is equally important. Possessing local knowledge does not mean that a maintainer should determine public purposes alone. There are legitimate trade-offs among service, cost, and different kinds of risk. But warning, escalation, and stop-work paths must be credible, or the knowledge closest to a fault will disappear because of contractual status, hierarchy, or fear of retaliation.
Maintenance is not naturally just
Once undervalued upkeep is recognised, another error becomes tempting: whatever has lasted deserves care. Maintenance can prolong polluting facilities, discriminatory procedures, and unsafe business models. Skilled repair may reduce the urgency of a structural problem, and organisations can use service continuity to postpone necessary reform.
Some burdens are themselves produced by inequality. A badly designed system may require low-authority staff to correct data manually. Gaps in public services are filled by unpaid care. Products designed against repair force users to buy proprietary replacements. Celebrating ingenuity without asking why the same people repeatedly compensate for poor design turns adaptation to injustice into a virtue.
A maintenance judgment must include conditions for exit. If repair costs rise without restoring acceptable performance, risk cannot be controlled, or the object’s purpose is no longer legitimate, replacement, redesign, or cessation may be more responsible. Maintainers should participate because they know where the structure repeatedly resists correction. A final decision also needs the people who receive the service and those who would bear the consequences of withdrawal.
What a Sustenesis perspective can and cannot add
The limited Sustenesis orientation used in this series treats relative stability as something other than the absence of change. Structures operate under constraint, experience disturbance, and preserve effective coherence through correction. Maintenance can then be understood as part of how an entity continues to exist as what it is, rather than an ancillary cost that begins after creation.
This helps explain why replacement of parts need not destroy identity, and why refusing adjustment can cause a structure to lose its own function. It also directs attention to differentiated roles, feedback, and corrective capacity. None of this establishes that a system is just. Effective continuation describes how a structure persists; it cannot by itself decide whether the purpose deserves preservation.
The cases in this series are consequently not decorative proof. If some systems remain viable through spontaneous user repair without a central maintainer, the framework must account for distributed formation. Where wholesale replacement is preferable to gradual upkeep, we need to explain how continuity may cross a material break. Application should allow evidence to limit interpretation rather than making every case fit in advance.
Provisional judgment: maintenance as a discussable public activity
Recognition does not require every maintainer to become a public figure. Much work needs to be performed quietly, professionally, and continuously; excessive display can interfere with service. What should become intelligible is the maintenance structure: responsibility, state information, backlog priorities, distribution of risk, routes for warnings, and conditions that initiate redesign or retirement.
Evaluation must change as well. Counting completed work orders alone can reward a badly designed object that continually generates work. Failure counts are insufficient because under-reporting and temporary luck resemble safety. Asset condition, service outcomes, risk exposure, recurrent faults, worker reports, and user experience need to be interpreted together. The result is less tidy but closer to the capability being protected.
The provisional conclusion is not that maintenance is more important than innovation. They are not permanent opponents. When something new is created, the issue is whether capacities for observation, repair, learning, and exit were also created. When something old is retained, the question becomes whether it is allowed to respond to reality. A world remains inhabitable partly because ordinary deviations are not dismissed as background noise.
What could change this view? If later cases show that visibility reliably produces rigidity, defensive compliance, or excessive cost, the form of disclosure will need tighter limits. If field participation sometimes weakens overall safety, the relevant conditions must be stated. Version 0.1 makes only a lower-level claim: normal operation cannot prove that maintenance is unimportant, and responsibility should not remain separated from the knowledge, resources, and corrective authority required to carry it.
Primary sources and further reading
- Steven J. Jackson, “Rethinking Repair”, in Media Technologies, 2014.
- Shannon Mattern, “Maintenance and Care”, Places Journal, 2018.
- Victorian Department of Treasury and Finance, Asset Management Accountability Framework.
- Victorian Auditor-General’s Office, Compliance with the Asset Management Accountability Framework.
- Sustenesis, the canonical source for the philosophical theory used in the limited application above.
Series navigation: Who Maintains the World? series overview
Discover more from Geoffrey Chen
Subscribe to get the latest posts sent to your email.