Capabilities, Dependencies and Freedom of Action

Rethinking resilience and innovation in complex socio-technical systems

This article builds on a discussion recently initiated at an Images & Réseaux technology conference dedicated to the resilience of systems of systems. The topic naturally invites a technological perspective: how can increasingly interconnected systems remain resilient as their components and infrastructures evolve?

This is why Cobelty proposed broadening the question.A system of systems is not merely a combination of technologies. Over time, its resilience also depends on people, organisations, knowledge, skills, standards, intellectual property, open-source communities, industrial relationships, financing mechanisms and institutions. Systems of systems are also complex socio-technical systems whose components, relationships and dependencies evolve on different timescales.

Discussions during and after the conference highlighted two particularly revealing reactions. The first concerned the complexity—sometimes described as “unprecedented”—that emerges when we attempt to consider these different dimensions and their interdependencies simultaneously. The second took the form of a question: “What does this mean for our collaborative innovation models?” This question is not only about how collaboration between actors should be organised. It also concerns the objectives these arrangements are intended to serve and the mechanisms through which the necessary resources are allocated.
What, exactly, are we seeking to fund? A project? A technology? An infrastructure? An industrial capability? A skill we want to preserve? The reduction of a dependency? Or an option whose future relevance we cannot yet precisely anticipate?
These objectives do not necessarily call for the same arrangements, time horizons or financing models.
Resilience therefore becomes inseparable from the question of resource allocation: which capabilities should we develop, acquire, preserve or share? Which risks are we seeking to reduce, and which options do we want to preserve?

A technological, organisational or financial decision does not only determine what we can do today. It also changes the choices that will remain available to us tomorrow.

From learning to risk management

Approaches such as Customer Development and Lean Startup have profoundly influenced innovation practices by recognising that we begin with assumptions—about problems, customers, value propositions or business models—that need to be tested and revised. This logic remains essential. But in complex socio-technical systems, uncertainty is not limited to users and markets. We also make assumptions about which technologies will matter, which capabilities will become critical, which actors we will depend on, and even where the boundaries of the ecosystem in which we operate will lie. Most importantly, these assumptions do not require the same amount of time or resources to revise. A customer assumption can sometimes be tested quickly. Developing scientific expertise may take years. A standard can shape an industry for a long period. Rebuilding a lost industrial capability or organisational competence can be both time-consuming and costly.

The question is therefore no longer only:
“Which assumptions should we test?”
It is also:
“Which risks should we accept, reduce, share or hedge—and which options should we preserve if our assumptions prove wrong?”

This also invites us to reconsider what we are funding when we fund innovation.
Funding a project with a clearly identified expected outcome is not the same as funding a capability. Preserving a competence that is rarely used but difficult to rebuild may appear inefficient if assessed solely on the basis of its immediate utilisation, while still carrying significant value from a resilience perspective. The appropriate financing logic therefore depends on the objective being pursued. The appropriate financing logic therefore depends on the objective being pursued.

This perspective also means that innovation should not be reduced to the organic development of new capabilities, or even to collaborative innovation.

An organisation may develop a capability through its own R&D, build it with partners, access it through licensing, pool it with others, recruit the necessary skills, invest in an organisation that possesses it, or acquire that organisation.

Developing, collaborating, licensing, investing and acquiring are different possible responses to the same capability challenge.
The choice depends on the desired combination of time, cost, risk, control, dependency and reversibility.

Innovation is therefore also a matter of financial engineering: how should capital be allocated across different ways of building, acquiring, preserving or sharing capabilities, given the risks we are prepared to accept and the options we want to retain?

From technologies to capabilities

This perspective shifts a recurring question in R&D:
“What product or service will come out of this project?”
towards another:
“What will this project enable us to do?”

An R&D activity may generate knowledge that proves useful across several future products. It may enable an organisation to assess an external technology, integrate future components, participate in standardisation, or simply understand a field well enough to decide what to do next.

A product can therefore be viewed as one temporary expression of a combination of capabilities accumulated over time.
A capability consequently has option value: it keeps open a future decision that we are not yet able to specify today.
Expertise may therefore have value even when a technology is purchased from a supplier, because it enables an organisation to assess it, negotiate, consider alternatives or recognise when another architecture becomes preferable.

Conversely, abandoning a capability is not simply a matter of reducing costs. It may also eliminate options.
The question then becomes:
“Which decisions would become impossible, slower or entirely dependent on a third party if we lost this capability?”
A capability architecture is therefore not a detailed plan for the future. It is a representation of the capabilities we believe we need in order to understand, decide, act and learn, while accepting that this representation itself will have to evolve.

The ecosystem cannot be taken for granted

No organisation innovates alone. Suppliers, partners, research organisations, open-source communities, standardisation bodies, investors and customers all contribute, to varying degrees, to the capabilities it can mobilise.
But representing this environment as an ecosystem contains an implicit assumption: that the actors and domains that will matter tomorrow can already be found somewhere on our map today.
Yet a scientific field that is peripheral today may become central tomorrow. A technology originating in another industry may transform ours. An open-source community may redefine an architecture. A regulatory, economic or geopolitical development may turn a previously secondary capability into a strategic one. This is where the notion of an exosystem becomes useful. An exosystem is not simply a larger ecosystem. It refers to the space from which transformations may emerge that challenge the very boundaries of the ecosystem we currently consider relevant.

The ecosystem helps us act within the environment we know. The exosystem reminds us that the environment itself can change.Strategic intelligence then becomes a capability in its own right: not only observing what is changing, but regularly questioning our own representation of what matters.

Control, access, influence

Once a capability is considered important, a practical question arises: what kind of relationship should we establish with it?
Three relationships can be distinguished: control, access and influence.
Some capabilities need to be sufficiently controlled because they are difficult to replace or essential to our ability to understand and decide. Others may remain external provided we can access them under satisfactory conditions: suppliers, licences, partners, shared infrastructures or open-source communities. But access may not be sufficient when a capability remains external while its evolution can significantly shape our future. This is where influence becomes relevant. Influence does not mean control. It means having sufficient understanding, legitimacy and relationships to contribute to the evolution of an important capability or architecture—through standardisation, open source, scientific communities, research partnerships or industrial alliances.

There is therefore no need to control everything. We need to know what must be controlled, what we merely need access to, and what we need to be able to influence as it evolves.

Sharing to preserve

Control, access and influence do not yet tell us how these capabilities should be organised. They may be organised individually or collectively.

Sharing can first be a response to asymmetry. Some capabilities are too costly, too scarce or too complex to be sustainably maintained by a single actor. Pooling them may make it possible to share costs, access or influence. But sharing can also address a challenge of organisational resilience. A capability may appear to be under control while remaining fragile if the skills, knowledge or resources required to exercise it are excessively concentrated. Resilience therefore requires us to consider their transmission, redundancy and availability, as well as whether—and how quickly—they could be rebuilt if necessary.

This requires a distinction between assets, functions and capabilities.
Ownership of an asset determines, among other things, who holds the rights to decide how it is used.
A function organises the exercise of those rights.
A capability rests on the skills, knowledge, relationships and resources required to understand, decide and act effectively.

These dimensions do not necessarily need to follow the same architecture. Concentrated ownership can coexist with a distributed capability. Conversely, distributed ownership does not guarantee that the corresponding capability will be resilient.
The question is therefore not to choose abstractly between centralisation and distribution, but to understand what needs to be preserved, at what level, and through which architecture.
This becomes particularly important for capabilities that are only used intermittently but would be lengthy, costly or uncertain to rebuild.
A capability can be too important to lose, yet insufficiently utilised to be maintained individually.

Pooling may then provide an answer. In other situations, resilience may instead require maintaining an organisation-specific capability.
Sharing to preserve therefore does not mean sharing systematically. It means organising capabilities in ways that preserve the ability to act despite asymmetries, failures and changes in the environment.

Dependencies, options and freedom of action

Whether we control a capability, access it, influence its evolution or share it, dependencies arise. This is not necessarily a weakness. It is a characteristic of complex socio-technical systems. The objective is therefore not to eliminate dependencies, but to understand their architecture and their consequences for our options.
An organisation may depend on a supplier while retaining sufficient expertise to assess its choices, identify alternatives and change architecture if necessary.
Conversely, a seemingly limited dependency can become structural if the knowledge or alternatives required to challenge it disappear. Similarly, a deliberately chosen interdependency between several actors may offer greater resilience than a fully internalised but fragile capability. Greater interdependence does not therefore necessarily mean less autonomy.

In the face of certain asymmetries, deliberately chosen interdependencies may actually reduce far more constraining dependencies.
The relationship can be summarised simply:
Capabilities shape dependencies.
Dependencies open or close options.
And options determine our freedom of action.

From technology transfer to the circulation of capabilities

This perspective also invites us to broaden the traditional conception of technology transfer. Technology transfer is often represented as a pathway: research → invention → protection → transfer → industrialisation.
But capabilities circulate in many directions. Organisations develop knowledge, absorb it from partners, recruit skills, integrate open-source technologies, acquire licences, form partnerships, invest or make acquisitions. At the same time, they disseminate capabilities through licensing, publications, software, standards, training or collaboration.

Technology transfer can therefore be understood as one of the mechanisms through which capabilities circulate, combine and recombine across organisational boundaries.
This also invites a different perspective on intellectual property, open source and standardisation. The question is not whether protection or openness is inherently preferable, but what we need to control, what we need access to, what we can share, and what we need to help shape.
Protection, openness and sharing become different ways of organising capabilities and dependencies in order to preserve options.

Strategic autonomy as freedom to adapt

This brings us back to the notion of strategic autonomy. Autonomy does not mean self-sufficiency. In a complex socio-technical system, no organisation controls all the capabilities on which it depends. Autonomy lies instead in the ability to mobilise the capabilities required to act, understand the dependencies they create, and retain enough options to change course when the environment evolves.
An organisation may own many assets while having few options. Another may rely extensively on external capabilities while retaining considerable freedom of action through its expertise, alternatives, relationships and ability to influence.
In some situations, autonomy can also be collective: several actors may pool a capability that none could preserve alone, or collectively exercise influence that would be inaccessible to them individually. The objective is therefore not independence, but to choose and organise the interdependencies that best preserve the ability to act.

The goal is not to own everything, internalise everything, share everything or predict everything.
It is to understand which capabilities and options we cannot afford to lose before we even know exactly when we will need them.

These questions belong neither to technology intelligence alone, nor to R&D strategy alone, nor to intellectual property alone, nor to finance alone. They lie precisely at their intersections: understanding how the environment is evolving, identifying capabilities that may become critical, characterising dependencies and asymmetries, assessing the associated risks, and informing choices about which capabilities to build, acquire, share, preserve or influence.

These are precisely some of the issues that Cobelty Strategic Intelligence aims to investigate and inform. Not to predict which trajectory will prevail, but to better understand the options available, those at risk of disappearing, and the decisions that can preserve freedom of action over time in an uncertain environment.

This capability-based perspective finally opens another question. We have previously proposed that changes in technological architectures are shifting the relevant unit of analysis and management of innovation, and that innovation can consequently be considered through the lens of a capability architecture. Thinking about resilience invites us to extend this shift to the way resources are allocated to building, acquiring, maintaining or sharing those capabilities.

This is not only about public funding for research and innovation. Funding a capability means, more broadly, mobilising and allocating capital to make that capability possible or preserve it, whether through internal R&D, industrial investment, partnerships, pooling arrangements, equity investment, acquisitions or public funding. Some of our frameworks for funding and managing research and innovation nevertheless rely on categories such as fundamental research, industrial research and experimental development. These categories primarily describe the nature of the activities being funded. But if the objective is also to build, acquire, preserve and maintain over time the capabilities required for our freedom of action, another perspective becomes possible. Should we only ask which research and innovation activities we want to fund, or also which fundamental, industrial or experimental capabilities we want to make possible, preserve or be able to mobilise over time?

The distinction is more than semantic. It could lead us to look differently at where we choose to allocate capital, why we do so, for how long, through which arrangements—individual or collective—and in support of which options.