The Politics of Proof: Why Good Systems Need More Than Better Data

May 18, 2026

9 min read

5 views

74%

The problem is not a lack of information

A common mistake in policy, evaluation, and social change is to assume that the hardest part is finding the truth. In practice, the harder part is deciding what kind of truth should count. A spreadsheet can be precise and still be politically irrelevant. A story can be vivid and still be methodologically weak. A network map can reveal hidden relationships and still leave decision makers unsure what to fund next.

That is the deeper tension: systems do not fail only because they lack data, they fail because they lack a shared way to recognize which data matters for which decision. Once you see that, a new question appears. What if the real infrastructure challenge is not collecting more evidence, but building the capacity to combine different kinds of evidence without flattening them into one false certainty?

This is where evaluation and systems financing unexpectedly meet. One asks, “What counts as evidence?” The other asks, “What counts as a viable system?” Together they point toward a more demanding idea: real-world change depends on the ability to triangulate truth across methods, institutions, and relationships, then fund the connective tissue that makes that triangulation possible.

Why one kind of evidence is never enough

Most debates about evidence get trapped in a false hierarchy. Quantitative data is often treated as the gold standard because it looks objective, scalable, and comparable. Qualitative evidence is treated as contextual and human, but sometimes dismissed as anecdotal. Network data is treated as elegant and structural, but sometimes seen as too abstract to guide action. Each of these reactions misses the point.

A hospital trying to reduce readmissions needs different kinds of evidence at once. The claims data might show where patients return. Interviews might reveal that discharge instructions are confusing. Network analysis might show that the same community organizations are carrying too much of the burden. No single lens tells the whole story. If you only trust the billing data, you may optimize the wrong process. If you only trust interviews, you may misunderstand scale. If you only trust the network map, you may see relationships but not outcomes.

The deeper lesson is that evidence is not just a substance, it is a coordinate system. Different forms of evidence answer different questions:

What is happening? Often answered by descriptive data.
Why is it happening? Often answered by explanatory, qualitative, or mixed evidence.
Where is the system connected or fragmented? Often answered by network data.
What is changing over time? Often answered by longitudinal and comparative data.

Policy and evaluation become more credible when they stop pretending one method can do all four jobs.

The goal is not to crown the best evidence. The goal is to assemble a more complete map of reality.

That shift matters because bad decisions often come from overconfidence in one narrow dataset. A program can look successful on paper while quietly failing the people it was meant to serve. A coalition can feel inspiring while lacking the structural connections needed to scale. A funding strategy can reward visible outputs while starving the relational work that makes outcomes possible.

Networks do not scale through force, they scale through trust

Now consider the financing side. System change networks are not ordinary organizations. They are more like living infrastructures, made of many actors, shared goals, and shifting roles. Trying to fund them as if they were single grantees is like trying to irrigate a forest with a fire hose. You may wet a few leaves, but you will miss the underground system that actually sustains the ecosystem.

Traditional funding logic tends to prefer clean lines: one applicant, one budget, one deliverable, one owner. But networked change is messy by design. It spreads risk, distributes expertise, and evolves through relationships. That makes it powerful, but also hard to fit into conventional grantmaking. Funding often ends up supporting the visible node, while the hidden coordination work goes unpaid.

This creates a structural mismatch. Networks need resources for sensemaking, alignment, brokerage, and trust building, yet these are some of the hardest activities to justify in a budget spreadsheet. A coalition may need a facilitator who never appears in a final outcome metric. It may need convenings that do not produce immediate outputs, or shared data systems that reduce duplication later. These are not extras. They are the operating system.

Think of it like aviation. A plane is not held together only by engines. It needs navigation, maintenance, air traffic control, and communication systems. If you fund only the engines, you may get a machine that can move but cannot safely fly. In social change, funding the flagship program while neglecting the network infrastructure is the equivalent of paying for thrust and ignoring steering.

This is where evidence and financing become inseparable. To fund a network well, you need data that can reveal whether the network is actually becoming more connected, adaptive, and capable. But to collect and interpret that data well, you need a network that has the time and trust to share information honestly. Evidence and infrastructure co produce each other.

The missing middle: from proof to coordination

The most important insight at the intersection of these ideas is that the purpose of evidence is not just to prove something happened, but to coordinate what happens next.

That sounds obvious until you examine how most systems operate. Evaluation is often treated as an after the fact audit. Funding is treated as a pre decision allocation. Research is treated as a separate sphere from practice. But networked change requires a different model, one where evidence is continuously used to steer a collective effort.

Here is a useful mental model: evidence has three jobs.

Legitimize: convince stakeholders that a problem is real and worthy of attention.
Orient: show where the leverage points and bottlenecks are.
Coordinate: help multiple actors decide what to do together.

Most systems are reasonably good at the first job and weak at the third. That is why so many reports are admired and then ignored. They legitimize a concern without creating a shared action frame. By contrast, useful network evidence does not just describe a field. It helps participants align their efforts, reduce duplication, and identify who should do what next.

A practical example: imagine a city trying to reduce youth homelessness. If the city only tracks shelter occupancy, it may expand beds. If it only interviews young people, it may learn about trauma, family conflict, and bureaucratic barriers. If it only maps provider relationships, it may notice that schools, landlords, and mental health services are poorly connected. The real breakthrough comes when these forms of evidence are combined to answer a coordination question: who needs to be in the same room, what should be shared, and which relationships are currently failing the system?

That is a more demanding use of evidence than “What happened?” It becomes, “How should a network reorganize itself to make better outcomes possible?”

Evidence becomes useful when it changes the shape of collaboration, not just the content of reporting.

This is why system change networks need funding models that support learning loops. If a network can test, observe, adapt, and reallocate resources quickly, it becomes more resilient. If it is forced to justify itself only through rigid annual metrics, it will often become brittle, performative, and slow.

A better model: fund the capacity to triangulate

The deepest synthesis is this: the thing worth funding is not only a program or a network, but the capacity to triangulate across forms of truth.

Triangulation is often treated as a technical method, a way to verify one fact with multiple sources. But in complex systems it is more than verification. It is a governance practice. It means creating conditions where quantitative patterns, lived experience, and relational structure can inform one another without one being prematurely subordinated to the others.

This suggests a different funding architecture with three layers:

1. Outcome layer

This is the part funders usually see. It includes service delivery, policy change, or measurable impact. It matters, but it is the top layer, not the whole building.

2. Learning layer

This includes evaluation, reflection, and data infrastructure. It answers whether the work is moving in the right direction and why.

3. Connectivity layer

This includes convening, facilitation, network weaving, relationship maintenance, and shared language. It allows the system to act as a system.

Most funding flows disproportionately to the outcome layer. But if the learning layer and connectivity layer are underfunded, outcomes become fragile and harder to sustain.

A useful rule of thumb is this: if you cannot describe how your funding will improve the system’s ability to learn, connect, and adapt, you are probably funding symptoms rather than capacity.

This is especially important in environments where the goal is not one linear intervention, but long term transformation. For example, climate adaptation, public health, educational equity, and community safety all involve multiple actors with partial authority. In such contexts, isolated grants can create islands of activity, while network funding can create a common operating rhythm.

What does that look like in practice?

Shared measurement systems that do not force identical metrics onto every partner, but allow comparability where it matters.
Regular convenings that surface bottlenecks and reassign roles.
Budget lines for coordination, not just delivery.
Dashboards that combine numbers, narratives, and network structure.
Evaluation designs that ask not only “Did it work?” but “Did the system get better at working together?”

This is not softer than conventional accountability. It is harder. It asks whether the whole ecosystem is becoming more intelligent.

Key Takeaways

Stop asking which single dataset is best. Ask which combination of evidence is needed for this decision: outcome data, lived experience, and network structure each reveal different truths.
Fund the connective tissue. Coordination, facilitation, and trust building are not peripheral costs, they are core system functions.
Treat evidence as a tool for coordination. The best data does more than prove impact, it helps multiple actors decide how to act together.
Measure system capability, not just program output. Look for signs that the network is becoming more adaptive, aligned, and able to learn.
Design for triangulation. Build processes where quantitative, qualitative, and relational evidence can inform each other instead of competing for supremacy.

The real question is not what counts as evidence

The temptation in policy and philanthropy is to keep polishing the proof machine. Better indicators. Cleaner dashboards. More rigorous evaluations. But the deeper challenge is organizational and political, not merely technical. We need systems that can hold multiple truths at once and use them to coordinate action.

That means the future of effective change depends on a more mature idea of intelligence. Not just data abundance, but evidence literacy. Not just funding generosity, but network stewardship. Not just proof of impact, but infrastructure for learning.

So the next time a report, a narrative, and a network map seem to disagree, resist the urge to choose a winner too quickly. The disagreement may be the signal. It may be showing you where the system is fragmented, where power is hidden, or where the most important work is happening off the page.

The deepest systems are not those that produce the most evidence. They are those that can turn evidence into coordinated, adaptable action. In that sense, the real question is not what counts as data. It is what kind of collective intelligence a society is willing to fund.