Building Systems That Learn

Why I am naming the method behind my work

For most of my career, I have worked in systems full of committed people trying to do better. There have been strong policy intentions, capable professionals, and people with lived experience pushing hard for change.

And yet, the same problems persist.

Services remain fragmented.
Roles are unclear.
Responsibility is distributed, but not aligned.
People, particularly people with disability, are still too often excluded from the systems that shape their inclusion, safety, and wellbeing.

These outcomes were not simply the result of individual failure, lack of care, or lack of knowledge. They were being produced through how activity was organised across systems, across sectors, organisations, roles, tools, and levels of governance.

If we want different outcomes, we have to work with that reality. We have to engage the system as it is actually enacted in practice.

Naming the method

My work first took shape in inclusive education contexts. For over 15 years, it has developed further in inclusive emergency management, disability-inclusive disaster risk reduction, and, increasingly, climate adaptation, where outcomes depend on coordination across many different systems.

Over nearly two decades, I have developed and applied a way of working grounded in Cultural-Historical Activity Theory and Developmental Work Research. I am now naming my extension of this work more explicitly as Distributed Developmental Work Research, or DDWR.

At its simplest, DDWR is a method for helping complex systems learn.

More specifically, it is an interventionist methodology for understanding and transforming outcomes produced across multiple interacting activity systems, rather than within a single organisation or service. It applies and extends Developmental Work Research into settings where responsibility, knowledge, and action are spread across sectors, organisations, and levels of governance.

This matters because many of the outcomes I care about, inclusion, safety, continuity of support, meaningful participation, are never produced by one actor alone. They emerge through the interaction of many parts of a system, each carrying different roles, constraints, assumptions, and forms of authority.

Developmental Work Research gave me a foundation for understanding how change happens when people can recognise contradictions in their own work, reflect on how those contradictions have developed, and begin to reorganise activity around a shared object. That foundation has stayed with me. What changed over time was the scale and setting of the work.

The systems I was working in did not sit neatly inside one workplace, one team, or one organisation. The learning had to travel. It had to hold across difference. It had to keep moving.

That is where DDWR comes from.

What “distributed” means

In DDWR, “distributed” means more than many people being involved.

The problem is distributed. Responsibility is distributed. Knowledge is distributed. Authority is distributed. Contradictions are distributed. The capacity to change the system is distributed too.

No single actor holds the whole picture. No single organisation can solve the problem alone.

This is foundational to how I think about method. It connects closely with ideas from distributed cognition, where understanding is not treated as something contained within one mind, but as something held and managed across people, tools, representations, and environments.

That has been deeply important in my own work. In complex systems, people cannot act on what they cannot hold. When knowledge is fragmented across actors and settings, it has to be gathered, interpreted, synthesised, and made usable. It has to be held through tools, shared frameworks, representations, and relationships that help people work across networks of actors operating in very different ways.

When responsibility for outcomes is distributed across systems, transformation requires a distributed method of learning.

Why I needed a different way of working

Many approaches to systems change simplify complexity in order to act. They focus on one organisation, one program, one policy setting, or one stakeholder group at a time. Sometimes that is useful. Often it leaves the main dynamics untouched.

In the areas where I work, outcomes such as inclusion and safety are produced through the interaction of many systems. Emergency management, disability services, health, community networks, local governance, policy, and lived experience all shape what happens. Each part works with its own rules, tools, timeframes, obligations, and blind spots.

From inside one part of the system, it can be hard to see the whole.

What looks like failure in one place may be a rational response to pressures elsewhere. What appears to be a gap in practice may actually be produced by a deeper misalignment in roles, tools, or expectations across the system.

That is why I have worked consistently with activity systems rather than isolated actors. My focus has been on how work is actually organised across:

• people

• roles

• tools

• rules

• organisations

• relationships

• and shared goals

I have always been more interested in enacted practice than formal design. Policy matters. Organisational structures matter. Intentions matter. But outcomes are shaped in the doing.

Making systems visible

A central part of this work has been making systems visible to themselves.

I do not begin by arriving with conclusions. I work to understand practice in context and to construct representations of how the system is functioning as people actually experience and enact it. That draws on lived experience, practice knowledge, research evidence, observation, dialogue, and sustained engagement over time.

Within these representations, I surface contradictions.

I use that term in a specific way. I do not mean general tension or disagreement. I mean structured misalignments within and between elements of activity systems as they are enacted in practice. Misalignments between policy and implementation. Between roles and expectations. Between tools and outcomes. Between responsibility and authority. Between what one part of the system assumes and what another part can realistically do.

These contradictions are often historically embedded. They have developed over time. They shape the conditions in which people work, often without being fully visible to those inside them.

My research revealed the same pattern.

Each part of the system may be acting logically within its own role.
And still, the system as a whole fails to produce the outcomes it claims to value.

That is hard to grasp when you are inside the work and managing your piece of it. It becomes easier to see when the system is reflected back in a form people can recognise.

Mirror data, and what it makes possible

This is where Developmental Work Research has remained a strong influence on my research practice.

A central mechanism in my work has been the use of mirror data. I do not use mirror data as simple feedback. I use it to reveal contradictions in how activity is organised and experienced across systems.

Mirror data can take many forms. Stories. Forum findings. Interview material. Cross-study synthesis. Lived experience accounts. Structured summaries of policy and practice. Developmental tools. Representations of how different parts of the system interact around a shared outcome.

The form matters less than the function.

What matters is that the material is deliberately structured so people can recognise themselves and the system they are part of. Recognition is crucial. Without it, people tend to defend their role or dismiss the analysis as belonging to someone else. With it, something opens.

People begin to see how their own work sits within a broader pattern. They can see why actions that make sense locally may still contribute to fragmentation at the system level. They can also see that the issue does not belong to one organisation alone.

In Developmental Work Research, mirror data forms part of what Engeström describes as double stimulation. People encounter material that makes a contradiction visible, then work with concepts, tools, or structures that help them respond to it differently.

That logic has been central to my work from the beginning. Over time, I carried it into more complex and distributed settings, where the contradictions were spread across systems rather than contained within a single workplace.

From bounded intervention to distributed learning

Classic Developmental Work Research has often been used in more bounded organisational settings. My work has retained its contradiction-driven and interventionist core, while extending it into contexts where participants do not share a single workplace, chain of command, or organisational object.

In these contexts, mirror data cannot stay local. It has to move.

Insights need to circulate across communities, services, emergency systems, governance settings, education spaces, partnerships, and policy environments. They need to remain intelligible across contexts. They need to support learning among actors who bring different responsibilities, different languages, and different degrees of power.

That changes the methodological task.

It is no longer enough to facilitate reflection within one setting. The work involves building forms of understanding that can travel across distributed systems without losing meaning. It involves carrying learning from one context into another. It involves helping people recognise that they are working on a shared object even when they stand in very different places in relation to it.

Learning across boundaries

This work depends on learning across boundaries because the problems themselves are produced across those boundaries.

Throughout my work, I have deliberately brought together:

• people with lived experience

• practitioners and service providers

• organisations

• community actors

• policy and decision-makers

I have done this because these perspectives are all necessary if we want to understand how outcomes are produced and what it will take to shift them.

This creates the conditions for two forms of learning.

The first is horizontal learning, across sectors, services, professions, and organisations. People begin to understand one another’s roles, constraints, and contributions. They can see where assumptions break down and where coordination becomes possible.

The second is vertical learning, between practice and governance. Insights move from practice toward policy, and from policy back into practice. Those responsible for governance gain a clearer view of implementation realities. Those working in practice gain a stronger understanding of the wider system shaping what they can and cannot do.

Neither form of learning happens automatically. It takes structure, translation, and time. It also takes work to build enough shared understanding for participation to be meaningful.

That has been especially important in disability-inclusive disaster risk reduction. People with disability are supported to understand emergency systems and risk contexts. Emergency managers and policymakers are supported to understand disability, inclusion, and lived experience. Shared inquiry becomes possible because participants are able to enter the work as informed contributors.

The ecology and the engine

Over time, I have come to think of this work as involving both an ecology and an engine.

The ecology describes the distributed system through which learning and action occur. It includes forums, workshops, partnerships, advisory structures, education programs, tools, governance processes, communication channels, and ongoing relationships. It shows where insight was generated, how it moved, where it was taken up, and how different parts of the work connected.

The ecology explains where things happened and how they connected.

But it does not fully explain how the system moved.

The engine is the analytical, relational, and interventionist work that enables the ecology to function and evolve. This is where my methodological expertise and intentional design sit. It includes:

• analysing activity systems

• identifying contradictions

• synthesising diverse forms of knowledge

• constructing mirror data

• revealing system dynamics in ways participants can recognise

• structuring spaces for collective reflection

• designing mediating tools

• carrying learning across contexts

• and sustaining movement over time

This part of the work is often less visible from the outside. It can look as though insight simply emerged and travelled. It did not. It had to be assembled, interpreted, shaped, and reintroduced into the system in ways people could use.

DDWR operates through the generation and circulation of insight, but also through the deliberate revelation of contradictions, the active synthesis and mediation of knowledge, and sustained leadership over time. That is what helps systems recognise themselves, build shared understanding, and progressively transform their own practice.

Tools that carry the work forward

A defining feature of my work has been the co-design of tools as mediating artefacts within activity systems.

These tools matter because they help people think, interpret, coordinate, and act. They do practical work in the present, and developmental work over time.

They support people to make sense of what they are seeing. They help stabilise shared understanding. They structure reflection. They guide action. They make room for self-assessment, progression, and ongoing learning. In distributed systems, they also help hold knowledge across settings and over time so that understanding does not disappear once a workshop ends or a project closes.

This is part of what I think of as a shared knowledge infrastructure.

In my work, continuity has never depended on facilitation alone. It has also depended on creating tools, syntheses, and structures that allow the outputs of collective work to remain accessible, interpretable, and usable over time. Knowledge can then be revisited, extended, and taken up in new contexts.

That matters for transformation. It also matters for practice right now.

The most useful tools in this kind of work do two things at once. They help people engage more effectively in the work in front of them by clarifying roles, improving coordination, and supporting better decisions. At the same time, they support longer-term development by embedding reflection, self-assessment, and progression within practice itself.

That dual function has been central to how I design.

Why time matters

This work takes time. More than most systems are set up to allow.

Expansive learning does not happen because people are exposed to information once. It develops as participants encounter representations of their own practice, recognise contradictions, make sense of how those contradictions have developed, and begin to reorganise activity around shared outcomes.

Even then, systems do not move all at once.

Some actors are ready earlier than others. Some can see the contradiction but do not yet have the authority or tools to respond. Some parts of the system resist. Disasters and crises expose what had previously remained hidden. Tools evolve through use. Insights deepen as they are tested.

There is nothing smooth about this.

I have never experienced this work as a neat sequence of intervention, uptake, and resolution. It has involved setbacks, uneven readiness, repeated explanation, and a great deal of persistence. It has also involved carrying learning forward when projects ended, when funding shifted, or when systems were not yet ready to act on what had become visible.

That longitudinal commitment has shaped the method as much as any theory has. I do not move on once an intervention is complete. I continue to support reflection, test new approaches, build tools, and work with actors and systems that are often moving at different speeds.

Without continuity, systems forget quickly. With continuity, they begin to accumulate understanding.

My role in the work

As I name this method more explicitly, I also need to be clearer about the role I have played within it.

My role has never been limited to generating knowledge. I have worked to help systems recognise themselves.

That has involved analysis, of course. It has also involved a great deal of epistemic, relational, and interventionist labour. Listening across fractured perspectives. Synthesising complexity without flattening it. Deciding what contradictions to surface and when. Structuring environments where people can engage difficult truths without collapsing into blame or defensiveness. Building enough shared understanding for meaningful participation. Creating continuity across projects and contexts. Designing tools that can carry the work beyond my direct involvement.

None of that has been incidental. It has been part of the method.

I say this carefully, because I do not want to overstate my role. But I also do not want to disappear the labour involved in helping distributed systems learn. Work like this requires intentional design and sustained stewardship, especially in the early stages when a shared object is still emerging and the system cannot yet carry the learning on its own.

Why I am articulating this now

Although this work matured in inclusive emergency management and disaster risk reduction, I believe it has relevance wherever outcomes are produced across fragmented systems striving to collaborate for inclusion.

For much of this work, the method has remained implicit.

The people I work with have not needed theory language in order to participate. They have needed support to understand systems, reflect on practice, work across boundaries, and change how things are done.

But over the years, I have been asked the same question again and again.

How did you do that?

That question has stayed with me. It feels like the right time to share this, as the work has developed to a point where its patterns and effects are clear.

What I am doing now is making this methodological practice more explicit. Not because it has only just come into being. It has been enacted over many years, across many contexts, through sustained work. It has reached a point where it can be named, examined, taught, and adapted by others.

Importantly, I want to articulate my methodological insights to show how exclusion is produced across systems and to translate those system contradictions into coordinated change toward inclusion.

I am also articulating it now because the effects of this work are visible in practice. Over time, it has contributed to tools being taken up, capability being built across sectors, and policy and governance changes that begin to embed inclusion more durably within systems. I will unpack these specific developments more fully in later writing.

This piece is a starting point. There is more to share in what follows including the:

• role of mirror data

• movement from contradiction to double stimulation

• design of tools as mediating artefacts

• function of shared knowledge infrastructures

• relationship between distributed cognition and system learning

• long arc of intervention across distributed systems.

For now, what matters is this:

Naming the method makes it possible to see it more clearly.
And once we can see how a system learns, we are better placed to help it change.

---

Preferred Citation: Villeneuve, M. (2026). Building systems that learn. Sydney: Centre for Disability Research and Policy, University of Sydney. https://vilicusrecusandae242061.substack.com/publish/post/203808475

Learn more about my research at Collaborating4Inclusion