The machine is enormous. The missing part fits in your hand.
That is one of civilisation’s strangest asymmetries: a small component can carry a very large consequence.
Dependency importance is not proportional to physical size.
The one-sentence answer
A spare part becomes civilisation-critical when a larger function cannot continue safely without it and replacement takes longer than the system can tolerate.
The part does not need to be expensive.
It does not need to be technologically glamorous.
It only needs to sit on the wrong dependency path.
Imagine the simplest version
A pump keeps water moving through part of a facility.
The pump itself is large, expensive and well maintained.
One seal fails.
The seal is small.
There is no suitable spare on site.
The original manufacturer no longer keeps the item locally.
Delivery normally takes six weeks.
Suddenly the important object is not the pump.
It is the missing seal.
This is the spare-part problem.
Why the visible asset misleads us
People naturally notice large assets.
Trains.
Aircraft.
Water-treatment plants.
Power stations.
Hospitals.
Data centres.
Factories.
But large assets are assemblies of smaller dependencies.
Bearings.
Filters.
Valves.
Sensors.
Relays.
Power modules.
Connectors.
Special fasteners.
Software licences.
Reagents.
Certified replacement materials.
The asset is impressive because the small things are cooperating.
A civilisation can own the machine and still lack the capability if it cannot replace what wears out.
Five things determine whether a spare part is truly critical
1. Can the system operate without it?
Some failures merely reduce convenience.
Others stop the core function.
A broken decorative panel is not the same as a failed braking component.
2. Is there a substitute?
A generic fastener may have many suppliers.
A proprietary controller or medical component may have few.
Substitution is a technical question, not a wish.
3. How long does replacement take?
A cheap component available tomorrow may be low risk.
The same component with a six-month lead time may be strategically important.
4. Does replacement require special approval or certification?
A component can be physically similar and still be unusable because the system requires tested, approved or traceable parts.
5. How much time does the system have?
If an asset can operate safely in degraded mode for months, the spare-part problem is different from a system that must be restored within hours.
Criticality is consequence × substitutability × replacement time × tolerance.
Lead time is hidden until the clock starts
Lead time is one of the least intuitive forms of fragility.
An organisation may know exactly which part it needs.
The supplier may even have the part.
But manufacturing queues, export requirements, transport, customs, inspection and installation may still put the replacement outside the system’s recovery window.
This is why spare-parts planning is really time planning.
The relevant question is not simply:
“Can we buy another one?”
It is:
Can we obtain, verify and install another one before the loss of function becomes unacceptable?
Inventory buys time—but inventory has costs
The obvious answer is to keep every possible spare.
That does not work either.
Parts cost money.
They occupy space.
Some degrade in storage.
Some become obsolete.
Some are superseded by redesigned equipment.
Some need controlled conditions or periodic testing.
So resilient spare-parts management is not maximum inventory.
It is intelligent inventory.
- Which failures stop essential function?
- Which parts have long lead times?
- Which parts are single-source?
- Which parts can be substituted safely?
- Which parts can be repaired locally?
- Which components are becoming obsolete?
- Which spares should be shared across several assets?
Obsolescence creates a different Lights Out
Sometimes the problem is not shortage.
The world has moved on.
A machine may remain mechanically sound while its controller is no longer manufactured.
A software package may reach end of support.
A sensor may be replaced by a newer model that does not fit the existing system without redesign.
A specialised supplier may leave the market.
This means maintenance can eventually become redesign.
The future can make yesterday’s spare part disappear even when nobody has physically consumed it.
Single-source does not always mean bad—but it changes the risk
Some components have only one qualified supplier because the design is proprietary, the manufacturing process is difficult, the market is small or certification is expensive.
That may be efficient during normal operation.
It also concentrates dependency.
The OECD’s 2025 Supply Chain Resilience Review emphasises diversification, agility and managing concentration risks rather than assuming global interdependence itself is the problem.
OECD — Supply Chain Resilience Review
A system therefore needs to know where it is concentrated.
Not every concentrated dependency requires duplication.
But every important concentrated dependency deserves visibility.
Standardisation can turn a rare part into an ordinary part
One reason standards matter is interchangeability.
If dimensions, materials, interfaces and performance requirements are agreed, more than one producer may be able to make a usable replacement.
That does not mean all parts should be generic.
Some systems genuinely need specialised designs.
But standardisation can reduce the number of unique things civilisation must remember how to replace.
Interchangeability is stored resilience.
A spare part also carries knowledge
Suppose the part arrives.
Can anybody install it?
Do they have the correct tool?
Do they know the torque?
Does the system need recalibration afterwards?
Does replacement require software configuration?
Does the repair have to be documented or independently checked?
The physical spare is therefore only part of repair capability.
The rest is knowledge, tools, procedures, access and verification.
The wrong substitute can be worse than no substitute
In a crisis, improvisation becomes attractive.
Sometimes improvisation is exactly what keeps a system alive.
But a component that “fits” may not meet the same pressure, temperature, electrical, sterility, fire, fatigue or reliability requirement.
This is why emergency substitution must still answer to the function and the receiver.
Can the replacement do the required job?
For how long?
Under what restrictions?
What evidence supports that decision?
Who bears the risk if the substitute fails?
The Lights Out Event changes the meaning of the warehouse
In the main Lights Out Event, external supply stops.
At first the city still possesses enormous physical capability.
But every repair begins to draw from a finite inheritance of parts.
The warehouse changes category.
Before the event, stock is a buffer until replenishment.
After permanent separation, stock becomes a shrinking inheritance.
That changes behaviour.
Parts are prioritised.
Non-essential systems may be cannibalised.
Local fabrication becomes more valuable.
Designs that use common parts become more attractive.
Repairability begins influencing what new systems should look like.
The civilisation lesson
Large systems encourage us to think at large scale.
But continuity often depends on small things.
A filter.
A gasket.
A cable connector.
A reagent.
A certificate.
A technician who knows what to do.
The larger the system becomes, the more important it is to know which small dependencies can stop it.
Civilisation is not resilient because it owns impressive machines. It is resilient when those machines can keep becoming usable again.
Continue the Lights Out branch
Previous: Civilisation Is Maintenance.
Next: Civilisation Knows More Than Any One Person.
Deep connections
- Stored Civilisation — spare inventories are stored time.
- Why Standards Let Strangers Trust Each Other — standards make substitutes safer and more interchangeable.
- The Bottleneck Problem — one critical component can become a system constraint.
- After the Flow Stops — parts are converted, repaired and repurposed under prolonged disruption.
- Repair Systems — canonical repair owner.
- Standards & MeasurementOS — technical standards owner.
Return to Civilisation — The Lights Out Event.