Policy design depends on knowledge. The conventional account of what knowledge matters focuses on expert analysis: economic modeling, epidemiological data, engineering assessments, legal precedent. These are real and often indispensable. But they are not the only kind of knowledge that determines whether a policy actually works when it meets the world it is meant to affect. There is another category of knowledge — contextual, tacit, and rooted in direct experience — that is systematically harder to transmit through formal channels, and that civic infrastructure has historically served to carry into policy processes.
When civic infrastructure weakens, this experiential knowledge does not simply fail to reach decisionmakers. It disappears from the process entirely, because the organizational mechanisms that gathered it, verified it against collective experience, and translated it into legible policy input are no longer present. The result is not that policy proceeds with incomplete information — that could be corrected by better data collection. The result is that policy proceeds without the kind of knowledge that data collection, as conventionally practiced, does not capture.
Two Kinds of Knowledge
James C. Scott’s analysis in Seeing Like a State introduced a useful distinction between what he called episteme and metis. Episteme is formal, codified, generalizable knowledge — the kind that can be written down, taught in classrooms, and applied across contexts. Metis is contextual knowledge: the practical understanding of how to navigate specific situations that can only be acquired through direct experience, that resists formalization, and that often cannot be transmitted verbally at all but only through participation.
Scott’s concern was with how modern states systematically undervalue metis in favor of episteme — producing policies and plans that look coherent from a distance but fail in specific conditions their designers did not understand. The classic illustrations are large-scale agricultural collectivization schemes that ignored local knowledge of specific soils and microclimates, or urban planning projects that replaced functional if informal neighborhoods with designed environments that failed to work because they lacked the properties that made the originals livable. In each case, the formal knowledge systems that informed design could not capture what the people living with the policy knew and what would have predicted the failure.
This is not a problem unique to authoritarian planning. It operates wherever policy is designed at a remove from the conditions it will affect, which includes much of normal democratic governance. The question is whether there are institutional mechanisms to bring experiential knowledge into the design process — and civic infrastructure has historically been one of the primary such mechanisms.
What Experiential Knowledge Looks Like
It is worth being concrete about what this kind of knowledge consists of, because the abstraction can make it sound either trivial or impossibly vague.
Consider housing vouchers. The policy logic of housing vouchers is that direct subsidies to tenants, allowing them to rent in the private market, are more efficient and provide more choice than public housing projects. This logic is defensible from a housing economics perspective. But the actual experience of navigating the voucher system involves a specific, concrete set of conditions that modeling does not capture: the administrative complexity of finding landlords willing to accept vouchers in areas with accessible employment, the time pressure of voucher expiration deadlines that may not align with housing market conditions in a given city, the inspection process that can void a lease weeks into a placement, and the cumulative effect of these conditions on families managing simultaneous employment and childcare constraints. The person who knows these things is not an economist or a housing administrator. It is the person who has been through the process.
That knowledge matters for policy design because it identifies failure modes that are not visible from a distance. If vouchers expire before recipients can successfully place them in tight housing markets, the program’s effectiveness is compromised regardless of its theoretical soundness. If landlord acceptance is concentrated in neighborhoods far from jobs or with poor school performance, the choice logic of the program is not actually operative. These are empirically correctable problems — but only if someone brings them to the attention of people who can change the program’s design.
Or consider healthcare prior authorization — the process by which insurers require clinicians to obtain advance approval for certain treatments before coverage is provided. The policy logic is utilization management: ensuring that expensive treatments are medically appropriate before they are covered. The experience of a patient managing a chronic condition through prior authorization involves repeated administrative barriers, delays in time-sensitive treatments, paperwork that falls on clinicians rather than payers, and denials that require appeals that require documentation that the patient’s condition has already prevented them from organizing. The patient with that experience knows things that an actuarial analysis of utilization rates does not capture — specifically, that the friction itself has medical consequences, including treatment delays, medication non-adherence, and the substitution of less-effective but easier-to-access alternatives.
Or consider school funding formulas. The distributional effects of property-tax-based school funding are well-documented at a policy level. What is less visible in the analysis is what the formula actually produces in a specific underfunded district: the condition of specific facilities, the turnover rate of specific positions that can’t be competitively compensated, the specific extracurricular programs that have been cut, and the downstream effects on student outcomes that aggregate statistics can measure but that the families and teachers in that district understand with a granularity that the statistics cannot replicate.
How Civic Infrastructure Captured This Knowledge
The historical function of civic infrastructure was not just organizational or political. It was epistemological: it created channels through which experiential knowledge could be aggregated, verified, and transmitted into policy processes.
Union grievance processes are a well-documented example. The formal grievance mechanism in a union contract is typically described as a dispute resolution process. But it also functioned as a knowledge system. Grievances identified specific points at which management practice diverged from worker experience in ways that affected working conditions, safety, and productivity. Organized grievances, tracked over time, revealed patterns that neither management data systems nor government labor statistics would capture. The union organization that processed those grievances had a detailed, specific, verified understanding of how work actually functioned on the shop floor — and that understanding fed into both contract negotiations and regulatory advocacy in ways that shaped labor policy.
Community organizing in local civic associations performed an analogous function at the neighborhood level. Organizations with sustained neighborhood presence — meeting regularly, connecting with residents across a range of issues, maintaining continuity of organizational knowledge over years — understood the conditions of specific places in ways that neither municipal data systems nor periodic surveys captured. They knew which building’s landlord was systematically avoiding code enforcement, which intersection had produced multiple accidents without being flagged in official statistics, which school’s principal was effective and which was not, and which city services were actually being delivered as described. That knowledge was a resource for policy engagement that did not depend on formal expertise.
Elinor Ostrom’s research on polycentric governance documented comparable dynamics in natural resource management. Ostrom’s work showed that communities managing shared resources — fisheries, irrigation systems, forests, groundwater — frequently developed governance rules that were far better adapted to local conditions than either market mechanisms or centralized state regulation, because those rules were informed by generations of accumulated local knowledge about how the specific resource behaved under specific conditions. That knowledge was not written down in ways that external experts could easily access; it was embedded in community practice and transmitted through the organizations that sustained that practice.
The relevance extends beyond natural resources. What Ostrom identified as a general principle — that locally embedded knowledge is irreplaceable in the governance of complex systems, and that organizations with sustained local presence are necessary to capture and apply it — applies to urban governance, social service delivery, public health, and the full range of policy domains that affect how people actually live.
What Replaced These Channels
The organizational landscape that has emerged as traditional civic infrastructure has declined consists primarily of what Theda Skocpol and others have described as professionally managed advocacy organizations — groups that represent constituencies rather than organizing them directly. These organizations have genuine capabilities: legal expertise, media access, coalition-building, research capacity. What they do not have is the systematic channel through which the experiential knowledge of their nominal constituencies flows into their work.
The distinction is meaningful. An advocacy organization working on housing policy may employ policy analysts who are deeply expert in housing economics, HUD regulations, and federal housing programs. What it typically cannot do is what a sustained community organization in a specific housing market can do: aggregate the specific experiences of hundreds of residents over years, identify patterns in those experiences that reveal how programs are actually functioning, and translate that pattern-recognition into policy-relevant knowledge. The advocacy organization speaks on behalf of affected communities. The community organization, when it works well, transmits knowledge from within them.
This distinction — between representation and transmission — is not a criticism of advocacy organizations, which do important work. It is a description of what they cannot substitute for. Policy processes informed by advocacy representation alone are missing a category of knowledge that historically entered through more direct channels. The absence is not visible in the same way that missing data is visible; it shows up downstream, in the gap between how a policy is expected to perform and how it actually performs in specific conditions that its designers did not understand.
The Policy Cost of Missing Knowledge
The policy cost of this missing knowledge is not primarily errors that are obvious before the fact — those would be caught by the expert analysis that policy processes do conduct. The cost is errors that are only obvious after implementation, in specific contexts, to people with direct experience. And it is errors that recur, because without the institutional channels to transmit experiential knowledge back to decisionmakers, they cannot learn from implementation in the ways that would prevent repetition.
This problem is structurally related to the broader argument about civic infrastructure and policy outcomes, addressed in the collective wisdom framing developed elsewhere on this site. The argument there is that democratic governance depends on a kind of distributed knowledge that no expert system, however sophisticated, can fully replicate — because the problems that governance addresses are embedded in specific conditions, and the knowledge of those conditions is distributed among the people who live with them.
The specific mechanism by which civic infrastructure contributes to collective wisdom is through organizational channels that aggregate and transmit experiential knowledge. When those channels are present, policy can be designed and revised based on how it actually lands. When they are absent, policy proceeds on expert analysis alone — which is capable of great sophistication about the theoretical structure of problems but is systematically limited in its knowledge of implementation conditions.
The Reconstruction Problem
Rebuilding these knowledge channels is not simply a matter of creating feedback mechanisms — surveys, public comment periods, town halls. Formal feedback mechanisms tend to capture responses to pre-formulated questions, not the unstructured, specific, contextual knowledge that makes up metis. What civic infrastructure provides is not just a mechanism to ask people questions; it is an organizational environment in which experiential knowledge is continuously aggregated through shared activity, identified as relevant through collective discussion, and eventually transmitted through organizational channels to decision-making processes.
This is why the organizational form of civic infrastructure matters, and why its absence is not easily compensated by technological or procedural fixes. An app that allows residents to report service delivery failures captures a different kind of information than a community organization that meets regularly and develops a collective understanding of a neighborhood’s condition over years. Both have value; they are not substitutes.
Early-stage projects like America’s Plan have begun to think through what rebuilt civic infrastructure could look like in current conditions — but the honest assessment is that this question is not resolved. The organizational models that effectively captured and transmitted experiential knowledge historically were developed over long periods, were adapted to specific conditions, and often had functions — social, fraternal, economic — that extended well beyond their policy-engagement role. Replicating their epistemological function in a different organizational environment is a genuinely difficult problem, and the research base for what works is not yet strong.
What is established is what is lost when these channels are absent. Policy design proceeds without the knowledge that would identify its failure modes before they become expensive and entrenched. Implementation problems go undetected until they produce visible failures that could have been predicted by anyone with sustained experience of the conditions the policy was meant to affect. And the gap between policy intent and policy outcome persists, cycle after cycle, in ways that erode confidence in governance without identifying the structural source of the problem.
This article was researched and drafted with AI assistance under human review. See our full AI and editorial practices.