A civilisation becomes more capable by depending on more things.
That sounds like a contradiction until we look at how modern life works.
A hospital does not manufacture every medicine, electronic component and oxygen fitting it uses. A city does not grow all its own food. A data centre depends on electricity, cooling, telecommunications and replacement hardware. A school depends on families, teachers, buildings, transport, textbooks, digital systems and public finance.
Specialisation makes civilisation powerful because no person or institution needs to produce everything alone.
The price of that power is dependency.
The important question is therefore not whether a civilisation can become independent of everything. It cannot. The useful question is: which dependencies are acceptable, which are dangerous, and what makes a dependency resilient rather than brittle?
This article uses fictional examples and systems analysis. It is not a continuity plan for a particular organisation, sector or country. OECD work on supply-chain resilience and critical-infrastructure resilience provides the external reference for diversification, redundancy, adaptability, coordination and the risks of interdependence. [1]
Dependency is not automatically weakness
If a city imports food, it is dependent on external producers and transport routes.
That dependency may increase capability rather than reduce it. The city can specialise in activities it performs well while obtaining food from regions with appropriate land and agricultural systems.
The weakness appears when the dependency has no tolerable alternative and failure can propagate faster than the system can respond.
Dependency should therefore be analysed through structure, not emotion.
Ask what is depended upon, how concentrated the dependency is, how quickly failure matters, how visible the dependency is and what alternatives exist.
Every capability chain contains hidden dependencies
Consider a fictional public library.
Readers see shelves, desks and librarians. The library may also depend on building maintenance, electricity, internet access, a catalogue system, publishers, delivery services, cleaning, accessibility equipment, payment processing and specialist software.
Each dependency can contain its own dependencies.
The catalogue system depends on servers. The servers depend on electricity and network services. The network depends on equipment and maintenance. The maintenance team depends on trained staff and replacement components.
This branching structure is why complex systems can fail in surprising places.
Start with the service, then trace backward
A dependency map should begin with an outcome rather than a supplier list.
For the fictional library, the outcome might be: readers can locate and borrow available books safely during opening hours.
Now ask what must remain true for that outcome to occur.
The building must be usable. Staff must be present. The collection must be accessible. Core records must be available. Borrowing must be recordable. Essential utilities must function.
This prevents the analysis from becoming a catalogue of everything the organisation buys. The purpose is to identify dependencies that matter to the capability being protected.
Concentration makes some dependencies more brittle
Suppose a fictional service buys a critical consumable from four independent suppliers with different production locations.
Now suppose it replaces them with one cheaper supplier.
Normal operating cost may fall. The service has also concentrated its dependency.
Concentration is not automatically wrong. A single supplier may be highly reliable, and multiple suppliers can add complexity. The question is what happens if that supplier cannot deliver.
The OECD’s 2025 Supply Chain Resilience Review notes that significant import concentration has risen in parts of global trade and emphasises agile, adaptable and aligned supply chains rather than assuming that simple relocalisation always increases resilience. [1]
Domestic does not mean independent
A factory can be located inside a country while depending on imported machinery, software, finance, energy or raw materials.
Moving the final production stage domestically may therefore change one dependency while leaving others intact.
The OECD’s resilience review finds that broad relocalisation can impose large economic costs and does not consistently improve resilience. [1]
The analytical lesson is not that domestic production is never useful.
It is that geography should not substitute for a dependency map.
Diversification reduces some risks and creates others
Multiple suppliers can reduce exposure to one supplier’s failure.
But diversification can increase coordination cost, qualification work, contractual complexity and variability.
A resilient design therefore considers the risk being addressed.
If two suppliers depend on the same upstream factory, they may look diversified while sharing the same hidden point of failure.
If two transport routes pass through the same bottleneck, route diversity may be less real than the map suggests.
Resilience depends on independence of failure modes, not merely the number of names in the supplier list.
Redundancy buys time
A spare component, backup generator, duplicate data copy or second trained operator may appear unused during normal periods.
Its value is revealed when the primary system fails.
OECD work on critical-infrastructure resilience includes redundancy, supplier diversification, backup productive capacity, rapid recovery and adaptability among resilience measures. [2]
The important quantity is often not the existence of a backup, but the time it buys.
A reserve that lasts two hours is different from one that supports the system through a two-day response time.
This connects directly to How Civilisations Coordinate Across Different Timescales.
Buffers should be connected to response time
Imagine a fictional service consumes ten units of a critical input per day and keeps thirty units in reserve.
Under the simplified assumption of no replenishment, the reserve provides three days.
If an alternative source reliably takes one day to activate, the buffer may be meaningful. If it takes ten days, the same reserve may be inadequate.
The arithmetic is simple. The judgement is not.
Real systems face uncertain demand, delivery, quality and simultaneous disruption. The example is only meant to show that buffer size and response time belong in the same analysis.
Substitution is a different kind of resilience
A system does not always need a second copy of the same input.
Sometimes it can switch to a different input, technology or process while preserving the outcome.
This is substitution.
A building may use more than one acceptable energy source. A transport network may redirect passengers across modes. A manufacturer may redesign a product around an available component.
Substitution can be more flexible than stockpiling, but it requires preparation. Standards, testing, training and regulatory acceptance may be needed before the substitute is usable.
Interchangeability must be real, not assumed
Two components can look similar while differing in specification, software compatibility, quality or certification.
A resilience plan that lists an alternative without proving interchangeability may contain a paper backup rather than an operational backup.
The same applies to people.
Two job titles do not guarantee that one person can replace another immediately. Experience, licensing, local knowledge and access permissions may differ.
Capability substitution should therefore be tested before crisis, where the consequences of discovering a mismatch are lower.
Visibility is a prerequisite for dependency management
An organisation cannot deliberately protect a dependency it does not know exists.
Modern supply chains often extend across several tiers. A direct supplier may rely on an upstream producer unknown to the buyer.
OECD work on supply-chain preparedness highlights the importance of visibility, scenario planning and accurate supply-chain data. [3]
The civilisational lesson is broader: map enough of the dependency chain to identify where failure can propagate.
Perfect visibility may be impossible. The response should be to identify uncertainty, not to pretend the unknown tiers are risk-free.
A dependency map needs criticality, not only connectivity
If a service depends on fifty suppliers, not all fifty deserve the same attention.
Ask which failure stops the outcome fastest, which input has the longest replacement time, which dependency is most concentrated and which failure can propagate to several other systems.
Criticality is therefore a relationship between dependency and consequence.
A low-cost component can be highly critical if no substitute exists and the whole service stops without it.
A very expensive input can be less critical if operations can continue for months without replacement.
Single points of failure deserve special attention
A single point of failure is a component whose loss can disable the whole system because no alternative path exists.
It may be physical, digital, institutional or human.
One bridge may connect an isolated area. One software identity service may control access to many applications. One expert may hold undocumented knowledge of a legacy system.
Removing every single point of failure may be unaffordable.
The useful task is to identify the ones whose failure would be unacceptable and decide whether redundancy, substitution, repair capability or contingency is justified.
Repair capacity is resilience after prevention fails
Some resilience strategies try to prevent failure.
Others make recovery faster.
Spare parts, diagnostic tools, trained repair staff, documentation and authority to act all shorten the duration of failure.
This is important because not every disruption can be prevented.
A civilisation can be resilient not because nothing breaks, but because breakage does not remain unrepaired long enough to destroy the wider capability.
Dependencies cross sector boundaries
Electricity depends on fuel, generation assets, grid infrastructure, communication and trained operators.
Telecommunications depend on electricity.
Finance depends on telecommunications and electricity.
Transport depends on finance, communications and energy.
These relationships mean failure can cross administrative boundaries faster than governance structures do.
OECD analysis of critical-infrastructure resilience explicitly notes the complexity created by functional interdependencies across sectors. [2]
The private sector can operate public dependencies
A civilisation’s critical functions are not necessarily owned by the state.
Utilities, logistics, communications, finance and technology may involve private, public and mixed ownership.
That means resilience cannot be designed solely inside government.
OECD work notes that a large proportion of critical infrastructure can be privately owned or operated, which makes coordination among public and private actors important. [2]
The governance question is how responsibilities, information and emergency authority are aligned across those boundaries.
Dependency can create bargaining power
If many users depend on one provider and switching is difficult, the provider may gain leverage.
This can be commercial, political or strategic.
The problem is not unique to international trade. It can exist within one organisation or local market.
A resilience review should therefore examine not only failure risk but negotiation risk: who can impose new terms because the other side lacks alternatives?
Diversification can reduce this leverage even if no physical disruption occurs.
Interdependence can also create stability
Dependency is not only vulnerability.
Mutual dependence can create incentives for cooperation.
Trade partners may avoid actions that would damage both sides. Shared infrastructure can motivate common standards and conflict-resolution mechanisms.
The challenge is to distinguish balanced interdependence from one-sided dependence.
The earlier article How Civilisations Compete, Cooperate and Merge explores how repeated exchange can create larger shared systems.
Resilience is not maximum stockpiling
Keeping enormous inventories of every input would be expensive and sometimes wasteful.
Some goods expire. Some technologies change. Storage itself requires space, maintenance and management.
The OECD Supply Chain Resilience Review describes resilience through a combination of redundancy, flexibility and responsiveness rather than one universal stockpiling strategy. [1]
The appropriate mix depends on the type of risk and the speed with which alternatives can be activated.
Resilience is not maximum diversification either
Adding suppliers indefinitely creates its own complexity.
Each supplier may require qualification, contracts, monitoring and integration.
A civilisation can therefore become less manageable while trying to become safer.
The objective is sufficient diversity for the critical failure modes, not diversity as an abstract score.
Local capability can be a strategic backup
A system may normally rely on external supply because it is cheaper or better.
Maintaining some local production, repair skill or technical knowledge can create an option during disruption.
This does not mean reproducing the entire global supply chain locally.
It can mean preserving the minimum capability required to bridge, adapt or restart.
Local capability is most valuable when it is connected to a clearly defined dependency risk.
Standardisation can increase substitutability
If equipment uses common interfaces, alternative components may be easier to source.
If data uses open or well-documented formats, systems may be easier to migrate.
If professional qualifications are recognisable across regions, trained people may move where they are needed.
Standards therefore influence resilience indirectly by changing how expensive substitution becomes.
Overly proprietary systems can increase lock-in even when they perform very well under normal conditions.
Dependency risk changes over time
A supplier that was one of many can become dominant.
A technology can become obsolete.
A safe transport route can become politically unstable.
A once-common skill can become rare.
Dependency mapping is therefore not a one-time audit.
The civilisation must observe concentration, lead times, technological change and the health of alternatives.
Scenario planning reveals dependency chains
Ask what happens if one input disappears for a day, a week or a month.
Which service stops first?
Which backup activates?
Which other sector becomes constrained?
Which assumption turns out to be false?
The purpose is not to predict the exact future. It is to discover dependencies that ordinary operation hides.
This connects to How Strategic Foresight and Scenario Planning Work.
A fictional dependency stress test
Suppose a fictional urban service has three critical inputs: A, B and C.
A has two independent suppliers and a five-day reserve. B has one supplier and a thirty-day reserve. C has three suppliers but all depend on one upstream component.
Which is safest?
The answer cannot be read from supplier count or reserve size alone.
We need failure probability, response time, upstream concentration, substitutability and consequence.
This is why resilience analysis needs a profile rather than a single number.
Do not confuse resilience with refusal to change
A resilient system can adapt its structure.
It may redesign products, switch suppliers, change transport routes or alter operating procedures.
Preserving function can require changing form.
This is the same principle developed in How Civilisations Decide What to Preserve and What to Change.
Dependency management is a portfolio problem
For each critical capability, a civilisation can combine several tools:
diversification;
buffers;
substitution;
repair capacity;
local capability;
contractual safeguards;
shared standards;
monitoring;
scenario planning;
international cooperation.
No single tool is always sufficient.
The Civilisation Atlas dependency test
For any important capability, ask:
What does it depend on?
Which dependencies are internal and external?
Which are concentrated?
Which share the same upstream failure point?
How long can the service continue without each input?
How quickly can an alternative be activated?
Is the alternative genuinely compatible?
Who maintains the repair capability?
What information would reveal deterioration early?
Which dependency gives another actor excessive leverage?
What would be the cost of reducing the dependency?
Which resilience measure creates new complexity?
Final answer
How do civilisations manage dependencies without becoming fragile?
They accept that specialisation creates dependency and stop treating self-sufficiency as the only definition of safety.
They map critical capability chains.
They identify concentration and hidden common points of failure.
They combine diversification, buffers, substitution, standards, local capability and repair capacity according to the actual risk.
They maintain visibility as dependencies change.
And they preserve enough flexibility to adapt when the disruption is not the one they expected.
A strong civilisation is not independent of everything. It is dependent in ways it understands, can survive and can change.
Sources and scope
[1] OECD, OECD Supply Chain Resilience Review: Navigating Risks, 2025. Used for the discussion of concentration, the limits of indiscriminate relocalisation, and the importance of agility, adaptability and alignment in resilient supply chains.
[2] OECD, Governance of Critical Infrastructure Resilience, 2021. Used for cross-sector interdependencies and the range of resilience measures including redundancy, supplier diversification, backup productive capacity, rapid recovery and adaptability.
[3] OECD, Bublu Thakur-Weigold and Sébastien Miroudot, Promoting Resilience and Preparedness in Supply Chains, 2024. Used for visibility, scenario planning, preparedness and the distinction between routine disruption and extreme disruption. The fictional examples and Atlas dependency test are original teaching applications.
Continue through the Civilisation Atlas
Return to the Civilisation Atlas, connect dependency to How Civilisations Govern Shared Resources, or examine system stress through How Civilisations Fail, Fragment and Recover.