Why Changing Systems Requires Two Kinds of Truth at Once

Jul 28, 2026

10 min read

2 views

71%

The Hidden Problem Behind Most System Change Efforts

What if the biggest reason social change efforts stall is not that they lack money, data, or good intentions, but that they keep asking the wrong kind of question?

Most attempts to improve complex systems are trapped between two instincts. One says: prove what works. The other says: fund the people and relationships that make change possible. These instincts are often treated as separate worlds, one about evidence and the other about infrastructure. In reality, they are two halves of the same puzzle.

A public health initiative can have a beautifully designed intervention and still fail if it ignores the local relationships, trust, and political conditions that determine whether anyone adopts it. A network of organizers can be full of energy and vision and still fail if it cannot sustain the quiet, unglamorous work of coordination, learning, and adaptation. The deeper issue is that systems do not change through ideas alone or networks alone. They change when evidence and relationship capacity evolve together.

That is the paradox at the center of durable change: the more complex the problem, the less useful it is to separate “what works” from “who can make it work, where, and under what conditions.”

The First Trap: Confusing Proof with Portability

There is a seductive simplicity to the phrase “what works.” It suggests a clean, universal answer, like a medicine tested in a lab or a policy tested in a pilot. But in social systems, a result is rarely portable in the way people assume. An intervention that succeeds in one place may fail elsewhere because the surrounding system is different: incentives, norms, institutions, power dynamics, and histories all matter.

This is why the most useful kind of evidence is not just “Does it work?” but “For whom, in what context, through what mechanism, and at what cost?” That is a more demanding question, but it is also the one that separates genuine learning from naïve replication.

Think of a neighborhood violence prevention program. If it reduces harm in one city, that tells you something valuable. But if the model depends on a specific coalition of trusted leaders, municipal alignment, and years of relationship building, then copying the program without copying the conditions is like transplanting a tree without soil. You may preserve the shape, but not the life.

This is where many efforts break down. They treat success as a package that can be exported intact. In reality, success is often contextual intelligence, a living adaptation to local conditions. The unit of learning is not the intervention itself. It is the interaction between the intervention and the environment.

In complex systems, the question is not whether something works everywhere. It is whether a pattern of change can travel while still adapting to the place it lands.

The Second Trap: Confusing Networks with Mere Connection

If evidence alone is insufficient, networks are often the next answer. Build coalitions, connect leaders, convene stakeholders, create a learning community. These are all essential, but they can also become vague slogans if they are not treated as serious infrastructure.

A network is not just a group of people who know one another. A high-functioning change network is more like a nervous system. It detects signals, routes information, coordinates responses, and learns from feedback. It is a distribution mechanism for trust, interpretation, and action.

This matters because system change often fails at the point where knowledge must become coordinated behavior. One organization may know how to do something well. Another may have credibility in a community. A third may have political access. Alone, each is partial. Together, if properly connected, they can create an emergent capacity greater than the sum of the parts.

But networks do not self-organize into effectiveness by magic. They require maintenance, facilitation, shared language, and resource support. If a funder only pays for direct service and never pays for the connective tissue, the network becomes brittle. People burn out, knowledge stays local, and collaboration becomes an unfunded aspiration.

That is why funding networks is not a soft add-on to strategy. It is a recognition that coordination is a productive asset. In complex change, the ability to align actors may be as valuable as any single intervention.

The Real Tension: Evidence Needs Networks, and Networks Need Evidence

The deepest insight is that these two challenges are not separate. They are interdependent.

Evidence without networks remains inert. It may sit in reports, academic journals, or slide decks, admired but unused. Networks without evidence can become highly mobilized but directionless, mistaking momentum for progress. The real engine of change is the coupling of the two: a learning network that can generate, test, adapt, and spread practical knowledge.

This creates a useful framework for thinking about system change:

Sense: identify what is happening in the system.
Interpret: understand why it is happening in this context.
Connect: move insight across people and organizations.
Adapt: modify actions based on feedback.
Reinforce: resource the relationships and routines that keep learning alive.

If any one of these fails, the system stalls. Evidence without connection cannot travel. Connection without interpretation cannot improve. Adaptation without reinforcement cannot last.

The practical implication is powerful: change efforts should not be designed as pipelines where research flows into implementation. They should be designed as feedback ecosystems, where learning is continuously produced by the network itself.

Imagine a public health initiative working to reduce childhood asthma. A traditional model might fund a study, publish findings, and ask agencies to implement recommendations. A networked model would do more: it would bring clinicians, tenants, housing inspectors, community advocates, and local officials into an ongoing loop. Data on asthma incidents would be paired with lived experience about housing conditions. Small experiments, such as targeted inspections or landlord engagement strategies, would be tested locally, then shared across the network. The result is not just information. It is coordinated learning.

That is the key shift: from evidence as a product to evidence as a social process.

A Better Model: The Three Layers of Durable Change

To understand why some change efforts endure while others fade, it helps to think in three layers.

1. The intervention layer

This is the visible activity: a program, policy, training, or service. It answers the question, “What are we doing?”

2. The learning layer

This is the mechanism that tells people whether the intervention is working, for whom, and under what circumstances. It answers, “What are we learning, and how do we know?”

3. The network layer

This is the social architecture that moves knowledge, builds trust, and coordinates action across boundaries. It answers, “Who is able to adapt, together, over time?”

Most change strategies overinvest in the first layer and underinvest in the second and third. They buy the program but not the learning loop. They fund the initiative but not the network that keeps it alive after the grant cycle ends.

The more complex the challenge, the more important the upper layers become. In a simple system, a good intervention may be enough. In a complex one, success depends on whether the system can notice, interpret, and respond. That is a networked capability, not a standalone intervention.

This is also why some reforms look successful at first but fade later. They may have had the right design, but not the relational capacity to sustain adaptation. Systems do not change once and stay changed. They are continually reconstituted by incentives, norms, and relationships. If the network is weak, the system gradually reverts to its old shape.

Why Funding Relationships Is Not Soft, It Is Strategic

There is a persistent bias in how resources are allocated: direct outputs are easier to count than relational infrastructure. It is easier to fund a workshop than a cross-sector trust-building process. It is easier to pay for a pilot than for the coordination required to scale what the pilot reveals.

Yet many of the most important assets in system change are invisible until they are absent. Trust, credibility, shared language, and repeated contact do not show up neatly in a quarterly dashboard. But without them, implementation becomes friction-filled and learning becomes fragmented.

This is why funding networks deserves to be understood as an investment in change capacity. A network can do things no single organization can do alone:

bring together actors with different forms of authority
surface local knowledge that centralized models miss
shorten the distance between learning and action
reduce duplication by sharing tested approaches
create resilience when one organization loses funding or leadership

In other words, networks are not just a communication layer. They are a collective capability.

A useful analogy is urban infrastructure. A city does not function because of one magnificent road. It functions because roads, bridges, transit systems, water lines, and traffic signals work together. If you fund only the roads and ignore the intersections, the system jams. In social change, networks are the intersections. They are where flow either happens or breaks down.

The Hidden Question: What Are We Really Trying to Scale?

One of the most important shifts in thinking is to stop asking how to scale a program and start asking how to scale a capacity to learn and coordinate.

Programs are finite. They can be copied, but often imperfectly. Capacity is different. Capacity can spread through relationships, shared methods, and common infrastructure. A well-connected network can absorb new evidence, translate it locally, and adjust to changing conditions without starting from scratch each time.

This reframes the meaning of scale. Instead of imagining a single model expanding everywhere, we can imagine a field of connected local adaptations that share principles, metrics, and support structures. The point is not identical implementation. The point is reliable improvement across variation.

This is especially important in public health, where the most important determinants of well-being often sit outside the clinic. Housing, transportation, education, employment, and community trust all shape outcomes. No single intervention can master this complexity. But a network of actors, equipped with a realist understanding of context and a durable funding base for coordination, can keep learning across boundaries.

That is the real strategic shift: move from funding answers to funding the conditions under which better answers can emerge.

The goal is not to eliminate variation. The goal is to make variation legible, learnable, and actionable.

Key Takeaways

Do not ask only whether a program works. Ask what mechanism works, for whom, and in what context.
Treat networks as infrastructure, not decoration. Coordination, trust, and shared learning are strategic assets.
Fund the feedback loop. Direct service without learning and adaptation eventually plateaus.
Measure relational capacity, not just outputs. Track how quickly insights move, how many actors adapt, and whether collaboration becomes easier over time.
Aim to scale learning, not just models. Durable change spreads through adaptive capacity, not mechanical replication.

The Conclusion: Systems Change Is a Test of Humility

The most profound lesson here is not technical. It is philosophical.

Complex problems resist simple ownership. No single organization, method, or dataset can fully explain them. That is uncomfortable for institutions that prefer control, but it is liberating for those willing to learn. It means the path to impact is not to force reality into a prewritten model. It is to build the relationships and evidence practices that allow a model to evolve in contact with reality.

That is why the most effective change efforts do two things at once. They seek truth about what is working, and they build the network that can use that truth. They recognize that learning is social, and that social infrastructure is a form of power.

So the next time a system change effort asks for more data or more collaboration, the better response is not to choose one. It is to ask a deeper question: what would it take to make evidence and relationship capacity reinforce each other until the system itself becomes capable of learning?

That is where durable change begins.