The case for open source tools in local public decision-making.
In brief
Local governments and school districts make decisions affecting thousands of people without the analytical groundwork that a comparable private organization would treat as routine. The data often exists. The comparative layer that makes it useful does not.
This paper describes where civic analytics sits within the broader civic tech landscape, explains the structural forces that keep the gap open, and argues that open source tools are the mechanism that can close it — extending to small and suburban bodies the analytical capacity that existing programs have demonstrated only at city scale.
Here's a pop quiz. It's open book.
First topic: sports. Who leads the World Cup in goals scored, and who wins the tournament outright? Will the Yankees catch Tampa Bay? Can they do it without their star player, Aaron Judge?
Second topic: local schools. What will it cost your district to electrify its bus fleet? What about the district next door? What books are being taught in each grade, and how has that changed over the past generation? How many teachers have left each year? How many bias, harassment, or discrimination incidents were reported last year — and is that number going up or down?
You may not be up on the sports questions above, and you may have no particular insight into who wins. But you're not short on resources to check. Each major sport supports a veritable stat industry. Your local newspaper, if you still have one, runs a robust sports section. Nearly every American city has a radio station devoted to talking sports twenty-four hours a day; if you're unsatisfied with the take on New York's WFAN, you can stream Tampa Bay's WDAE instead. All that data and interest adds up — are wagered on sports in the U.S. every year.
But the second set of questions is arguably more consequential to you than the first. It is also, for most readers, unanswerable on the spot — you likely wouldn't know where to begin. If you live in a place like New York City, you might have an inkling: the city is blanketed by multiple media outlets, and it has assembled an impressive data infrastructure of its own over the last twenty years. If you live just outside it, in a suburb with its own independent school district, you are more often on your own. The answer exists somewhere — in a board packet, a state filing, perhaps on the state Education Department's website — but nobody has assembled it, and you would have to do that yourself.
Here's the irony: people say they care more about it, too. Roughly seven in ten Americans are at least somewhat interested in news about local elections, and eight in ten in local laws and policies — . It's not that people don't care. It's that local government is harder to follow: politics ranks below sports in how easy people say it is to stay informed, and satisfaction with the coverage they do get is the lowest of any local news category surveyed.
There may also be a guilty-pleasure bias at work. As former Chief Justice Earl Warren about his news priorities: "The sports page records people's accomplishments; the front page has nothing but man's failures."
There's a , too. The play is the thing; there's nothing else unseen by the viewing audience. Even the decisions behind the play get explained on the record — a losing coach still takes questions at the post-game press conference. Public service is different: the work is more consequential, and almost nobody wants their own job livestreamed decision by decision. So a story gets told instead — packaged, defensible, presentable — and wherever a story has to be crafted, a competing story follows close behind. That friction is most of what people mean by "politics." Policing sits right at the fault line, which is why body cameras have proven so contentious: policing is public-facing and consequential at once, and putting a camera on it doesn't just record the work, . More objective data doesn't remove the need for narrative and judgment. But it can substitute for some of it — giving residents real accountability without requiring that every public employee's every decision be filmed.
That gap, between the questions a resident can actually ask and the analytical apparatus available to answer them, is where this paper starts.
Take the bus fleet question from the quiz above. It breaks into two distinct parts. The first — procurement mechanics, vendor relationships, grant applications — is administrative. The second — how many buses, on what timeline, which routes first, and what comparable districts are doing under the same mandate — is analytical. In practice the two collapse together: a consultant is engaged, a recommendation arrives, and whatever the existing data might have suggested gets folded into the engagement rather than examined before it begins.
This is rarely a failure of competence. The people making these decisions are often thoughtful and well-intentioned. The problem is structural.
There is a second failure layered on top. Government bodies now publish more raw information than ever before — meetings are streamed to YouTube, where AI tools can render those into searchable text minutes after being posted. By the formal measures of openness, things have never been better. And yet a resident who wants to understand on what analytical basis a district needs to double its school resource officers, or whether the enrollment projections used to justify closing an elementary school held up three years later, is rarely better served by all this raw availability than before it existed. Formal openness and genuine comprehension are not the same thing. The gap between them is where public trust erodes.
By civic analytics I mean something narrow and practical: applying the ordinary analytical methods used routinely in business and finance — comparison, peer benchmarking, trend analysis — to the decisions local governments make and to the work of the residents and organizations trying to understand those decisions. Analytical infrastructure for understanding, built for people rather than agencies, designed to be used by those the decisions affect.
Open source tools are what make this durable. When the tools are free and openly available, analysis stops being something a public body occasionally pays for and becomes something it can simply do — and the comprehension gap that currently substitutes raw availability for genuine accountability can begin to close.
Several civic tech platforms address parts of this problem, though don't necessarily fill the specific gap civic analytics addresses.
Vendors like CivicPlus, Granicus, and GovWell, alongside government-backed initiatives like Code for America and the US Digital Service, have built the infrastructure for digital government transactions — permit applications, license renewals, benefit claims, public meeting access. The work is valuable and consequential. It is also categorically different from analytics: it moves existing government processes online. It does not examine whether those processes produce the right outcomes, how a decision compares to neighboring jurisdictions, or what the data suggests before a board votes.
Tools like GoVocal, Decidim, and Pol.is allow residents to submit ideas, vote on priorities, and participate in specific decisions. (These tools are considered in our Participatory Governance Framework and associated scorecard) GoVocal's partnership with St. Louis — where the city used the platform to engage more than 7,000 residents in allocating a $250 million NFL settlement — demonstrates what well-resourced deployment looks like. The platforms are capable of sustained use. The pattern of deployment tends to be different: most engagements are tied to a specific decision, a grant cycle, or, as in St. Louis, a one-time funding event that creates both the resources and the political motivation to ask. The analytical question — what do the outcomes of similar decisions elsewhere actually show? — is usually never posed. #disparate-taxonomy
Data portals, FOIA systems, and legislative databases make raw data technically available without analyzing it. This is a necessary precondition, not a sufficient one. The gap between what a government publishes and what a resident can use to understand a decision is precisely where civic analytics works. #stranded-data
Two organizations have done the most substantial work at the city scale.
The Harvard Ash Center for Democratic Governance and Innovation has operated the Civic Analytics Network (CAN) since 2015 — a peer network of Chief Data Officers from large American cities. Their 2018 policy guidelines articulated six principles for city-level analytics, the most relevant of which holds that the public is the constituency for city analytics, and that the best insights emerge when government data use and civic engagement converge.
The Bloomberg Center for Government Excellence at Johns Hopkins (GovEx), through the What Works Cities program, has produced openly licensed documentation — benchmarking guides, performance management frameworks, community engagement playbooks — and worked directly with mid-sized cities on analytics capacity. Their benchmarking framework, which distinguishes internal benchmarking from external benchmarking against peer jurisdictions, is the most direct intellectual precedent for the cross-jurisdictional comparison this paper argues for.
Both have established the intellectual framework for analytics at city scale. Neither has produced a publicly accessible comparative dashboard showing how the cities in their programs stack up against each other — the Ash Center's public tool is a searchable catalog of individual case studies, and GovEx offers a self-assessment framework. This is evidence of how structurally difficult the comparison problem is: even the leading programs in the field have not solved it for their own domain.
The existing programs have established what analytics can do at city scale. The comparison problem remains unsolved even there.
Analytics as a practice is mature and well-resourced at the federal level. The intelligence community, the Department of Defense, the Centers for Disease Control, and the Centers for Medicare and Medicaid Services have operated sophisticated analytical programs for decades — established institutional functions with dedicated staff, procurement pipelines, and accumulated practice. State analytics is uneven: some states have capable data offices and infrastructure that informs policy; others lag significantly. The infrastructure at least exists as a recognized function worth building.
At the local level, particularly in small municipalities and independent school districts, analytics is largely absent as an institutional concept. The gap between a state government's data team and a 40-person school district administration is not primarily technological — the tools that support federal analytics are increasingly accessible. It is structural and financial.
The scale of the local tier makes this gap consequential. There are roughly 90,000 local government units in the United States. Of approximately 14,000 public school systems, about 12,900 are independent school districts — separate governments with their own elected boards, taxing authority, and staff, legally distinct from the municipalities they share geography with.
How these bodies are organized varies by region. The South and mid-Atlantic favor county-level school administration — Maryland runs every system at the county level. The Midwest and West lean toward districts independent of city and county government. Hawaii is the limiting case: a single statewide district. In the Northeast and mid-Atlantic, school districts tend to follow town and township lines geographically but are legally independent of the municipalities they resemble.
In New York State, school districts operate as fully independent governments with their own taxing authority and elected governance, rooted in statutes dating to 1812. The exceptions are the five dependent city districts — New York City, Buffalo, Rochester, Syracuse, and Yonkers — which fall under mayoral oversight. Every other district is a separate government.
School districts are not merely one more layer in the count. They carry the largest single claim on local property tax bills — typically 60 to 70 percent in New York. The decisions made with that money directly affect the community's children and draw the most engaged civic participation. New York's BOCES structure provides some shared services across districts — transportation, special education, vocational programs — and establishes a precedent for inter-district coordination that civic analytics infrastructure could extend.
The programs that have done the most for local government analytics were not designed to reach this landscape. The Harvard Civic Analytics Network serves large city Chief Data Officers. The What Works Cities program recently extended its eligibility threshold to municipalities of 30,000 or more — which still excludes most of Westchester's 45 bodies. The International City/County Management Association, writing in May 2026, explicitly named this gap: the bodies most in need of open analytical tools are those too small to build the capacity to develop or sustain them.
There is a further dynamic specific to inner-ring suburban counties. Westchester was historically the first suburban area of its scale in the world to develop, establishing a structural condition now common across the country: high fragmentation, strong local identity, and civic orientation pulled toward the adjacent principal city. Civic tech philanthropy has concentrated on that principal city. New York City has one of the most sophisticated municipal analytics operations in the country; the suburban municipalities next door are structurally absent from that conversation.
The Regional Plan Association, the independent nonprofit that has served the New York metropolitan region since 1922, offers one model for what civic analytical capacity looks like across a multi-jurisdictional region — independent experts, no governmental authority, producing research that individual municipalities cannot produce for themselves. Civic analytics tools extend that model to the day-to-day operational decisions RPA's long-range planning work doesn't reach. #lateral-blindness
The state infrastructure that should partially fill the local gap is real but limited in a specific way. New York's Education Department publishes data on every district — enrollment, assessment results, expenditure per pupil, staff qualifications — through its . The Office of the State Comptroller runs a Fiscal Stress Monitoring System that flags districts heading toward financial difficulty and conducts performance audits. These are genuine public investments in accountability.
What they are not is decision support. NYSED's data infrastructure was built for compliance reporting — districts submit data upward to the state; the state publishes some of it back. It tells you what your district's test scores were. It does not tell you whether your grade configuration makes sense against comparable districts, how your literacy outcomes look against demographically similar peers, or what the available data suggests about a curriculum before the contract is signed. The OSC catches fiscal distress after it develops; it does not inform the decisions that lead there. #stranded-data
The practical result is that the comparison most useful to a parent, a board member, or a journalist — how does our district's approach compare to others facing the same question? — requires either a consultant engagement, an extraordinary individual effort, or an act of the legislature.
That last option is not hypothetical. New York State mandates Holocaust education. When questions arose about whether districts were complying, the remedy was to pass a law directing the Education Department to produce a report. The department produced attesting that yes, all districts had complied — as of that moment, for that question. A binary compliance check on a single state mandate required legislative action and produced a one-time document with no ongoing mechanism. The effort required to answer a question that should be routine is itself evidence of the gap. #lateral-blindness
The state's data on Bias, Harassment, and Discrimination incidents follows the same pattern at the data layer. Districts are required to report incidents to NYSED. The data is collected. It becomes available more than twelve months after the reporting period, accessible only by downloading . The analysis — what does my district's incident rate look like compared to similar districts? has it improved? — is left entirely to whoever downloads the file. The data exists in the formal sense. The comparative comprehension does not. #stranded-data
The bus electrification mandate makes this visible at scale. The state passed legislation in 2022 requiring all school buses to be zero-emission by 2035, with all new purchases zero-emission by 2027. In December 2024, to the president of the New York State Energy Research and Development Authority requesting a detailed list of districts that had completed fleet electrification studies and preliminary data on costs incurred. The question implied its own answer: a comprehensive district-by-district cost model did not readily exist.
A mandate with potential implementation costs running into the billions was handed to more than 700 districts to figure out individually, with different consultants, producing non-comparable outputs, and no shared analytical baseline.
The equity dimension in all of this is not primarily about redistribution. It is about access to the comparative question itself. A parent in a wealthier Westchester district and a parent in a poorer one both lack routine access to cross-district comparison. The wealthier district can commission the answer; the poorer one typically cannot. But neither parent can simply ask how their district compares to its neighbors and receive a reliable, current answer without it costing someone something. What civic analytics makes available is that the comparative question becomes answerable as a matter of routine — not as the output of a procurement. #digital-equity
If the gap were only a matter of money, more funding would close it. Several forces hold it open simultaneously. None requires bad faith. Together they make the absence of analysis the path of least resistance.
No urgency. The absence of analysis is invisible. A decision made without it looks identical, on the night of the vote, to one made with it. Boards rotate, institutional memory is thin, and no one is accountable for a question that was never asked.
Consultant dependency. A cottage industry of local consultants has grown to fill the gap, and their role has genuine value. A consultant brings independent expertise with no institutional stake in the outcome — no past decisions to defend, no staff relationships to protect. They have worked on similar problems across multiple clients and can apply methods that a single district administrator rarely has the time or experience to develop. For decisions requiring community engagement, stakeholder interviews, and political navigation, an outside expert is often the right answer.
The structural problem is not the consultant but the dependency — and the absence of the analytical step that should precede the engagement. A district that has already established what comparable districts show, and what the state's own data suggests, arrives at a consulting engagement with a scoped question rather than a blank canvas. Without that prior pass, the consultant defines the question, and the engagement tends to be broader and more expensive than necessary. The method also stays with the firm: findings from one district don't travel to the next. #local-practice
Incentives against scale. A consultant has little reason to productize a method or share it. Every district is a separate engagement, and a method that travels freely is one that earns less. The economics reward keeping the work bespoke.
Fear of findings. Analysis that surfaces uncomfortable comparisons — between districts, between demographic groups, between what was promised and what was delivered — can be politically inconvenient. Not commissioning it is easier than answering for it.
Grant-funded distortion. Poorer districts can sometimes reach state grants for consulting. But the political economy of local vendor networks tends to produce fewer bids, insider awards, and weak accountability for what the money buys. The information that would expose this pattern — consulting contracts with stated goals and deliverables — is not currently required to be publicly reported at the local government level in New York State. #unknown-backlogs
The comparison problem. Because each district's consultant uses its own methods, the resulting reports cannot be compared across districts. The single comparison that would most help residents and policymakers — how does our choice look against what others have done? — is the one the current model makes nearly impossible. #lateral-blindness
Formal openness without comprehension. The vote on the YouTube record is typically the endpoint of a decision process that happened largely out of view: the framing of the question, the selection of the consultant, the informal alignment before the public session. By the time the agenda is posted, the decision has usually been shaped. Raw availability at the endpoint does not substitute for analytical access at the earlier stages where it would most change the outcome. #archive-failure
The system does not lack information. It lacks information that arrives early enough, in a form that citizens can use.
The natural civic question about a local decision is inward-facing: how does our district teach this subject, handle this incident category, manage this transition? That question is available — you can attend the board meeting, read the minutes, file a FOIL request. What is not routinely available is its outward-facing counterpart: how do other districts approach the same thing?
That comparative question is not exotic. It is the first thing a competent analyst asks in any other domain. It is also, structurally, the question that civic analytics tools are built to answer — not as a step in a decision-making sequence, but as standing infrastructure that makes the comparison available before anyone decides to commission it. The question that changes is not what the district does internally. It is what residents, journalists, and board members can see from outside. #lateral-blindness
Consolidation is sometimes suggested as the solution: the research literature on fragmented school districts often cites it as the structural remedy. But it's not always a practical one. Civic analytics does not require consolidation. It requires shared analytical infrastructure, which is a much lower political bar. Some of the benefits consolidation promises — reduced duplication of analytical cost, availability of cross-district comparison — can be partially achieved through shared tools without any district merging with any other. That is not an argument against consolidation as a long-term structural reform. It is an argument that the benefits do not require waiting for a political outcome that has proven consistently out of reach.
Open source has a consistent track record: wherever the underlying material is information, the open option eventually . The analytical work civic bodies need — comparison, trend analysis, structured public data — is information work of exactly this kind. When a capable tool costs nothing, the question shifts from can we afford to find this out? to what do we actually want to understand?
This framing builds directly on what the Ash Center and GovEx have demonstrated. Their work shows what city-scale analytics looks like when well-resourced. Neither has built a comparative dashboard across their member cities — evidence that even organizations committed to the principle have not solved the comparison problem through peer networks and case studies alone. Open source civic analytics asks the same questions for bodies that can't afford a Chief Data Officer, and answers them with tools rather than consulting relationships.
The organizations that benefit are those priced out of analytical work: small municipalities, school boards, nonprofits, community advocates, and local journalists. The Regional Plan Association has operated as an independent analytical body serving the New York metropolitan region since 1922 — producing research that no individual municipality has the capacity or neutrality to produce itself. That governance model — independent, non-profit, multi-jurisdictional, civic rather than governmental — is the right precedent. Civic analytics tools make it available at the operational decision level, not just the generational planning level RPA addresses.
The model is proven at larger scales and in adjacent domains. Each of the following demonstrates a different dimension of what comparative civic analytics infrastructure looks like when someone does the work.
The Stanford Education Data Archive (SEDA) is the most complete cross-district academic comparison tool available for US public schools. Built by researchers led by Sean Reardon, the Opportunity Explorer lets any resident, journalist, or policymaker generate detailed charts and maps of achievement data for virtually every school district in the country — filterable by demographics, geography, and school type, with comparisons across years. It demonstrates that standardized cross-district comparison is technically achievable at national scale. It also marks the boundary of what academic research infrastructure has reached: SEDA covers outcome data — test scores, achievement gaps — not the operational decisions that lead to those outcomes. The choice of curriculum, the configuration of grade levels, the handling of incident reports: these remain uncomparable.
Open States tracks legislation across all 50 state legislatures in standardized, cross-jurisdictional format — bill text, status, sponsors, votes — queryable by any resident. Originally a Code for America brigade project, now maintained by Plural, it demonstrates that policy comparison across many jurisdictions is buildable as standing civic infrastructure rather than a commissioned synthesis. The gap it leaves visible: state legislation is one layer up from the local ordinances, resolutions, and board decisions where most residents' daily experience of government actually lives.
Allegheny County, Pennsylvania built a data warehouse enabling cross-agency data sharing for human services delivery — linking records across health, housing, and social services to allow analysis that no single agency could perform on its own. Some of the resulting tools are public; see Allegheny County Analytics for the publications and dashboards released so far. The Civic Analytics Network documented its replication in member cities as one of the clearest examples of analytical infrastructure that travels. It demonstrates county-scale coordination among government bodies that would otherwise each work in isolation.
The IRS requires nonprofits to file Form 990 returns — detailed financial and programmatic disclosures. The data is technically public. ProPublica built Nonprofit Explorer, which makes those filings searchable and comparable across organizations, turning a compliance archive into a navigable civic resource. The architecture is the direct application of the #stranded-data insight to a specific federal dataset: the data existed; the comparative layer had to be built separately by someone willing to do it once, for everyone.
If civic analytics tools became standing infrastructure for Westchester's municipalities and school districts, several things would shift.
Analysis would happen more regularly — not only when a decision is pending, but as ongoing civic infrastructure a board can consult between decisions. When consultants are engaged, they would arrive into a problem that shared tools had already partially scoped, with baselines established and comparisons available, so their time goes to the judgment and community engagement that only they can provide.
Cross-district comparison would become a norm rather than a rarity. The apples-to-apples problem — that each district's bespoke report cannot be set against any other — dissolves when districts use shared, standardized tools. The comparison that currently requires commissioning a regional study becomes a default feature of the analytical landscape.
Insider contracting gets harder. When the analytical baseline is public and shared, a consultant's methods and conclusions are no longer easy to keep out of view. This would be reinforced by a structural reform worth naming directly: extending New York State's Open Book reporting requirements — which currently apply to state agency consulting contracts — to local government and school district consulting contracts. Requiring vendors, stated goals, costs, and deliverable summaries to be publicly filed would close the information gap that currently makes the consultant market largely invisible to the residents funding it.
Most fundamentally, the gap between formal openness and genuine comprehension narrows. Streaming board meetings and posting minutes give residents access to the record of a decision after it is made, in a format few have time to review. What they do not provide is the analytical context that allows a resident to evaluate the decision on its merits — to ask whether the enrollment projections were reasonable, whether comparable districts made different choices, whether the consultant's recommendation matched the data. That context is what civic analytics provides. Local governance does not need to be opaque to sustain itself. The communities that understand their institutions are more likely to trust them.
None of this is settled, and it would be misleading to suggest the path is clear.
There is the adoption question: who maintains these tools over time, who helps local officials and civic organizations learn to use them, and who funds the infrastructure when no one owns it for profit. There is the data quality problem: not every small body has clean enough underlying data to support meaningful analysis, and analytical capacity built on poor data is worse than no capacity at all.
There is political resistance: making analysis more accessible also makes uncomfortable findings harder to avoid. Some of the same officials whose cooperation a civic analytics initiative requires have institutional reasons to prefer opacity. The community knowledge, lived experience, and political judgment that a tool cannot substitute for are real, and this paper argues for analytics as a first resource, not a replacement for everything that follows.
Consolidation deserves a final honest note. The research literature largely points to it as the structural remedy for fragmentation's costs. This paper has argued that civic analytics can deliver some of the same benefits at a lower political cost. That argument does not settle the consolidation question, and it is not intended to. Whether the long-run answer to the equity gaps and duplicated costs of suburban fragmentation includes structural reorganization is a legitimate political question for the communities involved — not one the availability of better analytical tools resolves.
These are reasons to build carefully. The gap is real, it is widening, and the structural forces holding it open will not close it on their own.
This paper argues that open source tools are the mechanism that can close the local analytics gap. That argument doesn't need a working tool to hold up — but building one clarifies what the argument actually demands in practice.
The Ordinizer is an open-source tool, that generates record-based scorecards from public documents: structured evaluations built entirely from what an entity has put on the record itself — an enacted ordinance, a filing, a published claim — with every score citing the passage it rests on. Unlike opinion-based scorecards (analyst rankings, college guides, advocacy report cards), a record-based scorecard doesn't ask what an expert thinks of a municipality's wetland protections; it asks what the municipality's own laws actually say, using the same questions and rubric across every entity compared.
Two realms are underway: at NYSeeds, we are applying the method to municipal conservation ordinances across Westchester County, another is for software products, Participatory Governance Framework scorecard.