[!info] Related Documents - Central Hub: ../4. Indexes/00_BARSS_MOC - Core Methodology: ../5. Paper Drafts/EEDTM_Mathematical_Framework

A Shed Full of Dull Tools: Why the Sharpest Companies in Tech Keep Missing the Obvious Contract

Author: Wesley Bertil Date: February 2026 Paper Type: Cross-Disciplinary Policy Analysis / Technology Economics / Government Modernization

TARGET

• Journal/Outlet: Public Administration Review / Journal of Policy Analysis and Management (cross-submission candidate)
• Length: 25,000-35,000 words with appendices
• Format: Academic journal with embedded Statements of Work (SOWs)
• Unique Feature: Reads simultaneously as a policy paper, technology paper, and business paper
• Tagline: $170B-$285B/year in waste. $8.55B to fix it. Nobody bidding.

ABSTRACT

American government treats its citizens as middleware. In the absence of interoperable data systems, human beings physically transport information between agencies that possess the same data but cannot share it. A person drives to the DMV to prove they live at an address the IRS already knows. A mother fills out a Medicaid application using income data the Social Security Administration already holds. A 911 caller describes their location to a dispatcher while their phone broadcasts GPS coordinates to every pizza delivery app on earth.

This paper documents ten civic systems where the absence of basic data integration forces citizens into the role of human API... the biological connective tissue between government databases that should talk to each other automatically. For each system, we present the problem forensically, quantify the economic damage, and deliver a complete Palantir-style Statement of Work (SOW) with line-item budgets priced to the individual software license, labor category, and cloud hosting subscription.

We then demonstrate that these civic contracts represent objectively superior business for AI companies compared to the military and surveillance contracts they currently pursue. A national Medicaid auto-enrollment system would generate 4x to 19x more annual revenue than Palantir's entire ICE deportation portfolio. The building permits modernization for all U.S. jurisdictions costs less than one Army AI contract. The 911 modernization market is so obviously lucrative that Axon Enterprise spent $1.27 billion in 90 days acquiring two AI 911 companies.

The central finding is not that government should modernize... everyone knows that. The finding is that even from a purely Machiavellian capitalist perspective, civic modernization is better business than surveillance. More revenue, deeper lock-in, richer data, longer contracts, and zero boycott risk. The fact that AI companies chase $287 million ICE contracts while ignoring $30 billion civic infrastructure markets is not a strategic choice. It is strategic malpractice.

Keywords: civic technology, government modernization, AI procurement, NG911, Medicaid, automated enforcement, building permits, public administration, digital transformation, Palantir

I. INTRODUCTION: THE HUMAN API

1.1 The Thesis

Somewhere in America right now, a woman is sitting in a plastic chair at a government office, holding a folder of documents that prove things the government already knows about her. She has her Social Security card, which the Social Security Administration issued. She has her tax return, which the IRS processed. She has her utility bill, which the municipal water authority generated. She has her child's birth certificate, which the county vital records office filed.

She is waiting to hand these documents to a government employee who will manually enter the information into a government database. The information she is providing originated in government databases. She is the courier. She is the middleware. She is the API.

This is not a metaphor. It is the operational architecture of American government in 2026. The United States maintains hundreds of thousands of databases across federal, state, and local agencies. These databases do not talk to each other. They were built at different times, by different vendors, using different standards, running on different hardware. Some are written in COBOL. Some are running on actual mainframes from the 1980s. Some are, and this is real, stored in filing cabinets.

In the private sector, this problem was solved two decades ago. Your bank talks to your credit card company talks to your mortgage lender talks to your investment account. Amazon knows what you bought, what you browsed, what you almost bought, and what you will probably buy next Tuesday. Uber knows where you are, where you are going, and where you went last Thursday. Domino's can track a pizza from oven to door within ten feet of accuracy.

The government cannot do this. Not because the technology does not exist... the technology has existed for twenty years. Not because the legal authority is absent... Congress has repeatedly authorized data sharing between agencies. Not because the money is not there... the federal government spends over $100 billion per year on IT. The government cannot do this because nobody has built the API.

This paper argues three things:

1. The absence of civic data integration imposes a tax on every American. We quantify this tax across ten systems and find it runs into the hundreds of billions annually in direct costs, lost productivity, preventable deaths, and systemic inequity.

2. The solution is not speculative. For each of the ten systems, we present a complete, priced Statement of Work with line-item budgets down to the individual software license, labor category, and monthly cloud hosting bill. These are not white papers. They are contracts waiting for a signature.

3. Even from a purely capitalist perspective, civic AI contracts are objectively superior to military and surveillance contracts. More revenue, longer lock-in, richer data, lower political risk, and no Congressional hearings. The fact that AI companies pursue the inferior business is not strategy. It is stupidity.

1.2 The Palantir Problem

Palantir Technologies is the poster child for AI companies that chose wrong.

Founded in 2003 with CIA seed money, Palantir built its business on intelligence analysis and military contracts. By 2025, the company had accumulated $1.9 billion in federal contracts, with $970.5 million in a single year. Its flagship products... Gotham, Foundry, and the Army's Maven Smart System... are designed to find patterns in data for the purposes of surveillance, targeting, and warfighting.

Palantir's most controversial contract is with Immigration and Customs Enforcement. Over eleven years (2014-2025), Palantir received $287 million from ICE to build FALCON, ICM, and ImmigrationOS... systems designed to identify, locate, and facilitate the deportation of undocumented immigrants. The average cost per deportation using these systems: $10,854 to $17,121 per person.

Meanwhile, the cost to auto-enroll one eligible person in Medicaid using Express Lane Eligibility: $11.59.

That is not a typo. The government spends roughly one thousand times more to remove a person from the country than it would cost to enroll that same person in healthcare. And Palantir chose the deportation contract.

Consider the business math. Palantir's entire ICE portfolio generates approximately $26 million per year. A national Medicaid auto-enrollment contract would generate $100 million to $500 million per year in recurring revenue. The civic contract is 4x to 19x larger. It comes with 90% federal matching funds (the government pays 90 cents of every dollar during development). It creates permanent infrastructure dependency that no administration can easily unplug. And it does not produce front-page investigative journalism about your company helping separate families.

Palantir chose the $26 million contract with Congressional hearings over the $500 million contract with ribbon-cutting ceremonies. This is not strategy. This is institutional brain damage.

1.3 The Architecture of This Paper

This paper proceeds in four parts.

Part I: The Ten Failures documents ten civic systems where the absence of data integration forces citizens to serve as human middleware. For each system, we present the problem, quantify the damage, identify who bears the cost, and explain why the current state persists. The ten systems are:

Part II: The Ten Solutions presents complete Statements of Work for each system. The first five ("Big Five") include full SOW-level costing with line-item budgets. The remaining five include 75%-depth budget estimates. Each SOW is structured as a contract that a Palantir-type firm could present to a state governor for signature.

Part III: The Business Case demonstrates that civic AI contracts are objectively superior to military and surveillance contracts across every business metric... revenue, margin, lock-in, data value, political risk, and reputational exposure. This section is written for the Machiavelli in the room. It assumes the reader has no soul and cares only about shareholder value. Even then, the civic play wins.

Part IV: The Missing API synthesizes the ten cases into a unified theory. The common thread is not technology. The technology exists. The common thread is not money. The money exists. The common thread is that American government was built in a pre-digital era and nobody has performed the integration. The "Missing API" is the connective layer that transforms government from a collection of disconnected filing cabinets into a coherent operating system. Building it is the largest, most profitable, most defensible, and most socially beneficial AI contract in history.

And nobody is bidding on it.

1.4 Methodology

This paper employs forensic policy analysis combined with bottom-up cost engineering.

For Part I (The Ten Failures), we compiled primary data from government sources (Census Bureau, CMS, FCC, NHTSA, DOE), investigative journalism (ProPublica, KFF Health News, Chicago Tribune), academic literature (QJE, INFORMS, Commonwealth Fund), and federal audit reports (GAO, OIG, MACPAC). Each claim is sourced to a specific document or dataset. Where estimates are used, we explicitly label them as estimates and provide the derivation.

For Part II (The Ten Solutions), we employed a methodology borrowed from government contracting itself: the Statement of Work. Each proposal is priced using:

• Actual contract values from public procurement records (GSA Schedule, USASpending.gov, city council minutes, SEC filings)
• GSA CALC tool rates for labor categories
• Published vendor pricing where available (AWS GovCloud, Twilio, Salesforce, Tyler Technologies)
• Wrap rates of 2.2x-2.5x on direct labor (standard government contractor overhead)
• Industry-standard maintenance ratios of 15-22% of initial build cost per year

Where vendor pricing is not public (common in government contracting), we derive estimates from comparable contracts and label them accordingly. All dollar figures are 2025 dollars unless otherwise noted.

For Part III (The Business Case), we compare civic and military AI contracts across six dimensions: revenue scale, contract duration, lock-in depth, data richness, political risk, and reputational exposure. The comparison uses Palantir Technologies as the primary case study because it is the most visible AI company with extensive public financial data (as a publicly traded company) and because its contract portfolio spans both civic and military work.

PART I: THE TEN FAILURES

In which we document how American government converts simple database queries into multi-hour ordeals requiring the physical presence of human beings.

FAILURE 1: Traffic Enforcement and the DMV

The Problem

The American traffic enforcement system operates as if computers have not been invented.

A police officer observes a vehicle traveling 47 miles per hour in a 35-mile-per-hour zone. The officer activates lights and siren, pursues the vehicle, and initiates a traffic stop. The officer exits a climate-controlled vehicle, walks to the driver's window, requests license and registration, returns to the patrol car, manually enters the driver's information into an in-car computer, waits for a response from a state database, walks back to the driver's window, issues a citation, and returns to patrol. Elapsed time: 12 to 25 minutes.

The E-ZPass system, deployed across 19 states and 40 agencies, reads a transponder on a vehicle traveling 65 miles per hour and charges the correct toll to the correct account in under two seconds. No officer. No stop. No confrontation. No 12-minute bureaucratic ritual.

The technology to identify a vehicle, verify its registration, check its insurance status, and issue a citation... all without a human being leaving a chair... has existed for over a decade. Automatic License Plate Readers (ALPRs) scan plates at a rate of thousands per hour. Speed cameras measure velocity with mathematical precision. RFID-based vehicle identification can passively confirm registration status as a car passes a sensor mounted on a traffic light.

We have this technology. We use it to collect tolls. We do not use it to enforce traffic laws. Instead, we deploy 700,000 sworn police officers to perform the world's most expensive database lookup, at an average fully-loaded cost of approximately $300 per traffic stop when accounting for officer time, vehicle depreciation, fuel, and administrative processing.

The Numbers

The DMV side is equally absurd. The average American spends 44 minutes per DMV visit. There are approximately 230 million licensed drivers in the United States. A significant plurality of DMV transactions... address changes, registration renewals, title transfers... involve information the government already possesses. You are driving to a building to tell the government something it already knows.

FAST Enterprises, the dominant DMV modernization vendor, charges states $50 million to $280 million to modernize their DMV systems. Yet most state DMV platforms still cannot automatically update a registration when the owner files a change of address with the USPS... a federal database that already exists.

Who Pays

Everyone. The officer who spends 95% of their shift doing paperwork instead of investigating crimes. The driver who loses an afternoon at the DMV. The Black driver who is 20% more likely to be stopped and searched during a traffic stop that a camera could have handled without racial bias. The taxpayer who funds 700,000 officers to do the work of a network of sensors that would cost a fraction of the personnel budget.

And the person who dies because an officer was tied up writing a speeding ticket on Highway 9 instead of responding to the domestic violence call two miles away.

Why It Persists

Three reasons. First, traffic enforcement generates revenue. Municipal budgets in many jurisdictions depend on ticket revenue, creating a perverse incentive to maintain the manual system. Second, police unions resist automation that would reduce headcount. Third, and most fundamentally, nobody has presented the alternative as a complete, priced, ready-to-sign contract.

We will.

FAILURE 2: Courts and Pretrial Justice

The Problem

The American pretrial system operates on the following logic: a person accused of a crime, who has not been convicted of anything, is placed in a cage until a judge decides whether they should remain in a cage. The decision is made in approximately 2 to 5 minutes, based on the judge's gut feeling, the prosecutor's request, and whatever information the defense attorney (who met their client 15 minutes ago) can provide.

The result is 470,000 people sitting in local jails on any given day who have not been convicted of a crime. They are legally innocent. Many of them are in jail because they cannot afford bail amounts as low as $500. The cost to detain a single person pretrial averages $106 to $136 per day, depending on the jurisdiction. The cost of a GPS ankle monitor: $4 to $35 per day. The cost of a court reminder text message: under $1 per case.

The Public Safety Assessment (PSA), a validated pretrial risk tool developed by the Laura and John Arnold Foundation, is available to any jurisdiction in America for free. It uses nine factors from a defendant's criminal history to generate a risk score. It has been validated across 300+ jurisdictions. It does not eliminate judicial discretion... it informs it. And it is free.

Most jurisdictions do not use it.

The Numbers

COVID-19 proved that remote court hearings work. Every courthouse in America conducted hearings via Zoom in 2020 and 2021. Failure-to-appear rates dropped. Case processing times improved. Then the pandemic ended and most jurisdictions went back to requiring physical presence... for hearings that last an average of three minutes.

The scheduling problem alone is staggering. A single criminal courtroom processes 50 to 100 cases per day. Defendants, witnesses, attorneys, and officers are all required to appear at the same time, regardless of when their case will actually be called. A defendant summoned for a 9:00 AM hearing may not be called until 2:00 PM. This is not a scheduling system. It is a waiting room with a judge.

Who Pays

The defendant who loses their job because they sat in jail for three days over a $500 bail they could not post. The child whose parent does not come home because a judge set bail at 2x the family's monthly income based on a two-minute assessment. The taxpayer who funds $14 billion per year in pretrial detention for people who are legally innocent. The victim whose case is delayed because the court calendar is backed up by three months of continuances caused by a scheduling system that has not been updated since 1970.

Why It Persists

The bail bond industry is a $2 billion per year business. It employs lobbyists in every state legislature. It has successfully blocked bail reform in dozens of jurisdictions. The industry's survival depends on the current system's inefficiency... if defendants could be released on risk assessment and monitoring, the bail bondsman becomes unnecessary.

Also: judges like discretion. They resist tools that might constrain their authority, even tools that would make their decisions better. A judge who sets bail at $50,000 based on instinct does not want a computer telling them the defendant is a low risk. The computer might be right, and that is threatening.

FAILURE 3: Medicaid Enrollment

The Problem

The government will spend $287 million to find people to deport but cannot spend a fraction of that to find people to insure.

There are 6.4 million Americans who are eligible for Medicaid but not enrolled. These are not people who have been denied. They are people who have never applied, or who applied and were lost in the paperwork, or who were enrolled and then "churned" off during a renewal they did not know was happening.

The data to enroll them already exists. The IRS knows their income. The Social Security Administration knows their identity. The SNAP program knows their household composition. The state vital records office knows their dependents. Every piece of information on the Medicaid application originates in a government database. The application is a form that asks citizens to transcribe government data and hand it back to the government.

Express Lane Eligibility (ELE), authorized under CHIPRA 2009, allows states to use data from other agencies to determine Medicaid eligibility automatically. In Louisiana, ELE reduced the cost per enrollment from $116.48 to $11.59... an 87 to 90 percent reduction. Louisiana saved $9 to $12.9 million in Year 1 alone. Fourteen states have used ELE. Eight currently operate it.

The remaining 42 states require human beings to fill out applications to prove things the government already knows.

The Numbers

The Medicaid unwinding following the end of COVID-era continuous enrollment produced 25 million disenrollments. Sixty-nine percent of those were "procedural"... meaning the person was likely still eligible but failed to complete a renewal form they may not have received, may not have understood, or may not have been able to return because they were working two jobs and the Medicaid office closes at 5 PM.

Tennessee pays Deloitte $823 million to run a Medicaid eligibility system that produces a 69% procedural error rate in disenrollments. The entire national auto-enrollment system would cost $692 million to $2 billion to build. The system that works costs less than the single-state system that does not.

Who Pays

The 6.4 million people who are eligible for healthcare but do not have it. The hospitals that provide $50 billion per year in uncompensated care because patients show up in emergency rooms without insurance. The taxpayers who fund $12.1 billion per year in Disproportionate Share Hospital (DSH) payments to reimburse hospitals for treating uninsured patients who should have been insured. The children who lost coverage during the unwinding because their parents could not navigate a renewal form.

And, in the cruelest irony of all: the undocumented immigrant who Palantir helps ICE find and deport, who might have been the parent of a U.S. citizen child who just lost their Medicaid coverage because the state could not find them to send a renewal notice.

The government can find you to deport you. It cannot find you to insure you.

FAILURE 4: Building Permits

The Problem

In San Francisco, a building permit is reviewed by up to nine different departments. Sequentially. Building Code, Zoning, Fire, Public Utilities, Peer Structural, Engineering, Transportation, Planning, Forestry. Each department reviews the same set of plans. Each department contacts the applicant independently. Each department operates on its own timeline. A permit that could be reviewed in parallel across all nine departments in two weeks instead takes six to twelve months because each department must finish before the next one starts.

This is not a line at the DMV. This is a line at nine DMVs, back to back, where each DMV requires you to bring a different folder.

The NAHB estimates that regulations add 23.8% to the cost of a new home... $164,386 per house. For multifamily development, the regulatory burden exceeds 40% of total cost. A significant portion of this burden is not the regulation itself (which may be perfectly reasonable) but the time required to process the paperwork.

Austin, Texas deployed Archistar's AI plan review tool and discovered that approximately 80% of residential permit applications contained significant zoning deficiencies. The applications failed not because the projects were illegal, but because applicants did not understand the code. These deficiencies required an average of 1.6 resubmissions, each adding weeks to the timeline. An AI pre-screen that catches these errors before submission would eliminate the majority of rework cycles.

Honolulu deployed CivCheck's AI plan review and reduced residential code review time by 70%. Total review times dropped by up to 90%. Reviews that previously took 60 to 90 minutes were completed in 15 minutes.

The Numbers

A medium city processing 2,000 permits per year with an average 30-day unnecessary delay per permit at $670 per day: that is $40.2 million per year in construction carrying costs imposed on the private sector. The AI system that eliminates 60% of that delay costs $7 to $12 million per year. The return on investment is 2x to 3x in the first year, with benefits flowing entirely to the private sector in the form of faster housing construction.

Who Pays

The homebuyer who pays $164,386 more than they should because the permitting process adds 24% to the cost of construction. The developer who pays $45,662 per day while waiting for a permit that is being reviewed by nine departments in sequence when it could be reviewed in parallel. The family that cannot find affordable housing because the permitting bottleneck constrains supply. The construction worker who is laid off during permit delays. The city that loses tax revenue because projects are delayed or cancelled.

FAILURE 5: Water Infrastructure Monitoring

The Problem

The United States loses approximately 6 billion gallons of treated water per day to leaks in the distribution system. That is roughly 14% to 18% of all treated water, simply vanishing into the ground through cracked pipes, failed joints, and undetected breaks in infrastructure that averages 45 years old, with some systems dating to the Civil War era.

The EPA estimates the national water infrastructure investment gap at $625 billion over the next 20 years. The American Water Works Association estimates $1 trillion. The Bipartisan Infrastructure Law allocated $50 billion for water infrastructure... significant, but covering roughly 5% to 8% of the need.

IoT sensor networks can detect leaks, monitor water quality in real-time, predict pipe failures before they occur, and optimize treatment processes. The technology is mature. LoRaWAN, NB-IoT, and LTE-M connectivity options can blanket a municipal water system at a fraction of the cost of the pipe replacement itself. A basic sensor deployment for a very small water system costs $25,000 to $75,000. A comprehensive deployment for a large system runs $5 million to $25 million.

The barrier is not technology. It is the fact that America has approximately 50,000 community water systems, most of them small, most of them underfunded, and most of them running on infrastructure that predates the internet.

The Numbers

Flint, Michigan spent over $400 million responding to a water crisis that continuous monitoring would have flagged in days. The monitoring system for Flint's entire water network would have cost a few hundred thousand dollars. The ratio of response cost to prevention cost: roughly 1,000:1.

Who Pays

The resident of Flint who drank lead-contaminated water for 18 months before anyone noticed. The ratepayer who pays for 6 billion gallons of treated water that never reaches a faucet. The municipal budget that spends emergency repair dollars on catastrophic failures that predictive maintenance would have prevented. The child whose cognitive development was permanently impaired by lead exposure from corroded pipes that a $200 sensor would have flagged.

FAILURE 6: School Bus Transportation

The Problem

The United States operates the largest mass transit system in the world and optimizes it less than a Domino's delivery route.

Four hundred eighty thousand school buses transport 26 million students daily, at an annual cost of $28 billion. Most bus routes are designed manually, by a transportation coordinator using paper maps, institutional knowledge, and the same routes that have been run for decades. When students move, when schools open or close, when a new subdivision is built... the coordinator redraws routes by hand.

MIT's Operations Research Center built an algorithm for Boston Public Schools and tested it in 2017. The result: 50 buses eliminated (7% fleet reduction), $5 million in annual savings, and route optimization completed in 30 minutes instead of several weeks. This was published in the INFORMS Journal on Applied Analytics. It is peer-reviewed operations research, not a vendor white paper.

Most of the other 13,000 school districts in America have not adopted route optimization software.

The Numbers

The electric bus transition is its own argument. An electric school bus costs 3.5x a diesel bus to purchase, but the EPA Clean School Bus Program provides rebates of up to $325,000 per bus for priority districts... covering nearly the entire purchase price. The operating savings ($5,000 to $15,900 per bus per year in fuel and maintenance) accumulate over the 12-to-15-year life of the bus. And the buses can earn revenue by selling stored electricity back to the grid during peak demand hours. Beverly, Massachusetts earned $60,000 from three buses over three summers.

An electric school bus fleet is not a cost center. It is a distributed energy asset that happens to transport children.

Who Pays

The district that runs 50 more buses than it needs because nobody ran the algorithm. The student who spends 90 minutes on a bus that could take 55 minutes with optimized routing. The taxpayer who funds diesel fuel at $14,438 per bus per year when electric operation costs $8,400. The planet that absorbs the emissions from 480,000 diesel buses making 10 billion student-trips per year.

FAILURE 7: FAFSA and Financial Aid

The Problem

In 2024, the Department of Education deployed a new FAFSA form. It launched three months late (December 2023 instead of October 2023), contained 55 documented defects at launch (seven classified "critical"), could not be completed by students with at least one parent without a Social Security number (affecting millions of undocumented and noncitizen-parent households), and transmitted inconsistent tax credit data for approximately 15% of applications processed before March 30, 2024.

The call center received 5.4 million calls in five months. It answered 1.4 million. Seventy-four percent of callers were abandoned.

The cost of this disaster: Federal Student Aid spent $581 million on FAFSA modernization from FY2020 to the present. The "Next Gen FSA" program alone consumed $502 million by June 2022... already exceeding its $415 million lifecycle estimate... and the OIG found that FSA was not tracking government labor costs, so the real number is higher.

And the FUTURE Act, signed in 2019, already authorized the IRS to share tax data directly with FSA. The legal barrier was removed five years before the botched launch. Congress did its job. FSA could not do the engineering.

The Numbers

The enrollment numbers tell the story. First-year enrollment at four-year public institutions dropped 8.5%. At institutions with high proportions of Pell Grant recipients... meaning the schools that serve the poorest students... the decline exceeded 10%. NASFAA documented that the disaster disproportionately affected low-income Black and Latino students, the populations most dependent on financial aid.

The Department of Defense received 72% of all federal AI contract dollars from FY2022 to FY2024... $4 billion. Civilian agencies received $1.6 billion. The system that processes $120 billion in annual financial aid for 17.6 million students was allocated the digital equivalent of a tip jar.

Who Pays

The student who did not enroll in college because the FAFSA form did not work. The first-generation college student who gave up after calling the helpline four times and being abandoned each time. The institution that lost 8.5% of its freshman class because FSA could not deploy a website. The taxpayer who funded $581 million for a system that shipped with 55 defects. And, compounding across four years of lost tuition per student, the estimated $4 to $8 billion in lifetime institutional revenue that evaporated because the government could not build a form.

FAILURE 8: Property Tax Assessment

The Problem

Professor Christopher Berry of the University of Chicago documented that in Cook County, Illinois, flawed property assessments shifted $2 billion from wealthy homeowners to low-income homeowners. The cheapest properties were systematically over-assessed. The most expensive properties were systematically under-assessed. The result: a regressive tax... one that takes more from people who have less.

This is not unique to Chicago. Berry's nationwide analysis found that in 97.7% of U.S. counties, homes in the bottom decile of sale price face an assessment level approximately twice as high as homes in the top decile. The pattern is universal.

The racial dimension is documented in the Quarterly Journal of Economics (Avenancio-Leon & Howard, 2022): Black residents face a 12.7% higher effective property tax burden than white residents in the same jurisdiction, holding all else constant. Brookings found that Black-owned homes are appraised 21 to 23% below their value in non-Black neighborhoods, and that 81.3% of tax assessors are white.

The inequity is structural, and it is solvable. Riverside County, California contracted C3.ai to build an AI-powered appraisal system for its 460,000 residential parcels. The result: 40% improvement in model accuracy, 30% increase in properties valued within 5% of actual sale price, and quarterly model recalibration reduced from 40 days to 4 days. The cost: $6.375 million over five years... approximately $2.77 per parcel per year.

When Cook County replaced its assessor and reformed its practices, homeowners saved $1.9 billion. The technology to eliminate assessment inequity exists, is proven, and costs less than $3 per parcel per year to operate.

The Numbers

EagleView, the dominant aerial imagery provider for assessment, covers nearly 90% of the nation's population and generates an average ROI of 900% for its government customers. The average annual imagery cost is approximately $140,000 per county. For that price, the county receives satellite/aerial change detection that catches unpermitted additions, garage conversions, pool installations, and demolished structures that ground inspectors miss.

Who Pays

The low-income homeowner who pays twice the effective assessment rate of their wealthy neighbor. The Black homeowner who pays a 12.7% tax premium for the crime of owning a home in a Black neighborhood. The taxpayer who funded $250 million in Tyler Technologies contracts that tripled in cost and delivered late. And, in the most structurally ironic outcome: the entire property tax base, which is less accurate and less equitable because we use 1990s statistical methods on $630 billion in annual revenue while spending $2.77 per parcel per year could achieve 40% better accuracy.

FAILURE 9: Special Education and IEP Management

The Problem

There are 7.9 million children in the United States with Individualized Education Programs (IEPs) under the Individuals with Disabilities Education Act (IDEA). Their teachers spend more time on paperwork than on teaching.

An IEP is a legal document. It must contain specific, measurable goals. It must be reviewed annually. The student must be re-evaluated every three years. Progress must be reported to parents at regular intervals. Prior written notice must be given before any change in placement. Every deadline is legally enforceable, and missing one can trigger a due process complaint that costs the district $50,000 to $100,000 to defend.

Special education teachers manage caseloads of 15 to 30 students. Each IEP requires 1 to 3 hours of progress monitoring documentation per reporting period. Multiply 7.9 million students by 2 hours average by 3.5 reporting periods per year: 55 million teacher-hours per year are consumed by progress monitoring paperwork alone.

Twenty-five percent of special education teachers leave their positions every year. Eighty-three percent report emotional exhaustion. The paperwork is not the only cause, but it is the accelerant.

Meanwhile, districts leave $400 million to $1.8 billion per year in Medicaid reimbursements unclaimed because the IEP documentation systems do not integrate with the Medicaid billing systems. Students with IEPs receive speech therapy, occupational therapy, physical therapy, counseling, and other services that Medicaid will reimburse... if the documentation is complete, the consent forms are current, and the billing codes are correct. In most districts, these systems do not talk to each other.

Texas lost $300 million in special education federal funding in January 2024 due to improper coding. Texas then slashed $607 million per year in Medicaid from special education programs after federal auditors found systemic problems. The state asked 573 districts to return $16 million in improperly received funds going back to 2011.

The Numbers

The platform that solves this pays for itself on the Medicaid module alone. A district with 5,000 IEP students, 60% Medicaid-eligible, claiming $1,500 per student in reimbursable services, currently capturing 70% of eligible claims... that district is leaving $900,000 per year on the table. The IEP platform costs $405,000 to $680,000 in Year 1. It is self-liquidating by month 12.

Who Pays

The teacher who leaves the profession because the paperwork consumed them. The child whose IEP goals are vague and unmeasurable because the teacher did not have time to write better ones. The district that pays $100,000 to defend a due process hearing it would have won if the documentation had been complete. The taxpayer who funds $15.5 billion in IDEA funding while districts leave $1.8 billion in Medicaid reimbursements uncollected because the forms do not connect.

FAILURE 10: 911 Emergency Services

The Problem

A Domino's Pizza app can locate a delivery driver within ten feet of your front door in real time. A 911 call from the same phone may produce only a cell tower sector covering a quarter-mile radius.

The core infrastructure of the American 911 system is 1960s-era circuit-switched telephony. Most Public Safety Answering Points (PSAPs) run on legacy selective routers, copper circuits, and proprietary call-handling hardware that cannot accept text, video, or data. Approximately 40% of PSAPs cannot receive text messages. Eighty percent of 911 calls now come from cell phones, but the system was built for landlines.

The FCC estimates that proper indoor location accuracy for wireless 911 calls would save 10,000 lives per year. Ten thousand people die annually because the system that directs emergency responders cannot figure out which floor of a building the caller is on. Every pizza delivery app on earth has solved this problem. The entity responsible for saving your life has not.

The NG911 (Next Generation 911) transition would replace the analog infrastructure with IP-based networking, enabling text-to-911, video streaming, real-time location, and AI-powered call triage. NHTSA and NTIA estimate the cost at $9.5 to $12.7 billion over ten years, depending on the implementation model. Congress has introduced bills for $10 to $15 billion in funding. None have passed.

Meanwhile, the 911 system is funded by phone surcharges that generate approximately $3.8 to $4 billion per year. States are legally required to spend this money on 911 infrastructure. New York diverts approximately 41% of its 911 surcharge revenue to the general fund... $113 million in FY2024 alone. The FCC has no enforcement authority to stop this. It can only publish a report noting the diversion.

The Numbers

Axon Enterprise spent $1.27 billion in 90 days acquiring two AI 911 companies (Prepared and Carbyne). That is more than the annual federal 911 grant program has disbursed in its entire history. The private market has already decided that civic 911 technology is a massive business. The government has not funded the transition. The private sector is not waiting for permission.

But Axon is a body camera company. Its entry into 911 means the entity controlling the call-to-case evidence chain... from the moment you dial 911 to the moment the body camera footage is uploaded to the cloud... is a surveillance hardware company. The civic modernization opportunity that Palantir-type AI firms left on the table is being claimed by the police equipment industry. The AI companies chose military contracts. The surveillance companies chose civic infrastructure.

Who Pays

The 10,000 people who die each year because the 911 system cannot locate them. The caller who texts 911 from an active shooter situation and gets no response because their PSAP cannot receive texts. The dispatcher working mandatory overtime because 25% of positions are vacant. The taxpayer in New York whose 911 surcharge was redirected to plug a budget hole. The non-English speaker who waits 40 seconds for a live interpreter to join the call while AI translation could have activated in 8 seconds.

End of Part I. Part II (The Ten Solutions) follows with complete Statements of Work.

PART II: THE TEN SOLUTIONS

In which we present ten contracts that a Palantir-type firm could hand to a governor tomorrow. Priced. Scoped. Ready to sign.

The following section presents each civic modernization proposal as a Statement of Work (SOW). The Big Five (Traffic/DMV, Courts/Bail, Medicaid, Building Permits, Water) include full line-item budgets. The remaining five include 75%-depth budget estimates with key cost drivers identified.

Each SOW follows a standard structure: System Architecture, Key Components, Budget (State Pilot and National Scale), Revenue Model for the Contractor, and Return on Investment.

SOLUTION 1: Unified Traffic Enforcement and Vehicle Services Platform

Contract Title: Automated Traffic Safety and Vehicle Management System Scope: Statewide deployment of automated enforcement, passive vehicle identification, income-adjusted fine processing, and unified DMV digital services.

System Architecture

The platform consists of six integrated layers:

Layer 1: Passive Vehicle Identification Network RFID-based vehicle identification sensors mounted on traffic infrastructure (signals, signs, highway gantries). Each vehicle receives a transponder at registration... functionally identical to E-ZPass but for law enforcement purposes. As vehicles pass sensors, their registration status, insurance status, and inspection compliance are verified in real time. No officer involvement. No traffic stop.

Layer 2: Automated Speed and Signal Enforcement Speed cameras at high-risk intersections and corridors. Red-light cameras at dangerous intersections. All violations photographed, verified by AI, and citations issued electronically. Income-adjusted fines calculated automatically using IRS income data (the Finnish model, where a speeding ticket costs proportional to income).

Layer 3: AI Violation Processing Computer vision verifies each violation. Machine learning filters false positives. Citations are generated automatically with photograph, location, timestamp, and applicable statute. Contested citations are routed to an online adjudication portal.

Layer 4: Digital DMV All DMV transactions available online and at self-service kiosks. Address changes auto-populated from USPS change-of-address database. Registration renewals processed automatically upon insurance verification and fee payment. Title transfers executed digitally. License renewals with remote identity verification (biometric).

Layer 5: Income-Adjusted Fine Engine Fines scaled to ability to pay, using IRS income data. A $200 speeding ticket for a minimum-wage worker. A $2,000 speeding ticket for a hedge fund manager. Same violation, proportional consequence. Constitutional under existing precedent (Bearden v. Georgia, 1983).

Layer 6: Real-Time Analytics Dashboard Every traffic event, violation, citation, and court outcome feeds into a unified data platform. Cities can identify dangerous corridors in real time, measure the effectiveness of speed cameras, and allocate traffic engineering resources to locations with the highest crash rates.

Budget: State Pilot (Medium State, ~5M Population)

National Scale (10-Year)

Revenue offset: Automated enforcement generates fine revenue. A single speed camera at a busy intersection generates $500,000 to $2 million per year in citations. A 500-camera state deployment generates $250 million to $1 billion annually. The system pays for itself from enforcement revenue within 3 to 5 years.

Political note on federal ban: The Trump administration has signaled opposition to traffic cameras. This is functionally irrelevant. Traffic enforcement is a state and local function under the 10th Amendment. A federal ban would require overriding state police power... the kind of big-government overreach that is toxic to the conservative base. If left to the states, it does not matter. If attempted as a federal ban, it is political suicide. Two paragraphs and we move on.

SOLUTION 2: Pretrial Justice and Court Modernization Platform

Contract Title: Integrated Pretrial Risk Assessment, Court Scheduling, and Remote Adjudication System Scope: County-to-statewide deployment of validated risk assessment, AI-optimized scheduling, remote hearing infrastructure, automated notifications, and digital case management.

System Architecture

Module 1: Pretrial Risk Assessment Engine Deploy the Public Safety Assessment (PSA) or equivalent validated tool. Integrate with court records, criminal history databases, and real-time data feeds. Present risk scores to judges at first appearance with recommended conditions of release.

Module 2: AI-Optimized Court Scheduling Constraint-satisfaction algorithm that schedules hearings based on judge availability, attorney availability, courtroom capacity, defendant location, witness availability, and case complexity. Eliminates the "everyone shows up at 9 AM" problem.

Module 3: Remote Hearing Infrastructure Secure, CJIS-compliant video conferencing for all hearing types where physical presence is not legally required. Integrated with court recording, AI transcription, and digital case files.

Module 4: Automated Notification System Text/email/voice reminders for upcoming hearings. Automated compliance check-ins for pretrial supervision. Multilingual. Failure-to-appear rates drop 26% with text reminders alone (multiple studies).

Module 5: Digital Case Management Unified case file accessible to all parties (judge, prosecutor, defense, probation). Real-time docket management. Automated deadline tracking. Integration with state court management systems.

Budget: County Pilot (500K Population)

Statewide and National Scale

ROI: Pretrial detention costs $106-$136/day. GPS monitoring costs $4-$35/day. Every defendant moved from jail to monitoring saves $71-$132 per day. A county that diverts 200 defendants from pretrial detention saves $5.2 million to $9.6 million per year in jail costs. The system pays for itself.

SOLUTION 3: Medicaid Auto-Enrollment Platform

Contract Title: Automated Medicaid Eligibility Determination and Enrollment System Scope: National platform leveraging existing federal data (IRS, SSA, SNAP) to auto-enroll eligible individuals in Medicaid, eliminating manual applications and reducing churn.

System Architecture

Core Engine: Real-time eligibility determination using MAGI (Modified Adjusted Gross Income) rules applied to IRS income data, SSA identity verification, and SNAP/TANF enrollment records. When an individual becomes eligible (income change, job loss, life event), the system generates an enrollment action automatically. The citizen receives a notification that they are enrolled, not a form asking them to prove they qualify.

Federal Data Integration: - IRS API (IRC Section 6103 compliant): Income verification - SSA API: Identity, citizenship, death records - SNAP/USDA API: Household composition, categorical eligibility - State vital records: Birth, death, dependents - Federal Data Services Hub (FDSH): Existing infrastructure, built for Healthcare.gov

Identity Resolution: Master data management across federal and state databases to link records for the same individual across systems. Resolves name variations, address changes, and SSN discrepancies.

Churn Prevention: Continuous eligibility monitoring. Instead of annual renewal forms that generate 69% procedural disenrollment rates, the system checks eligibility continuously against federal data and only initiates a renewal action when the data shows a potential change in status.

Budget: State Pilot (Medium State, ~2M Medicaid Population)

State out-of-pocket (90/10 CMS match on DDI): $3.15M-$7.1M State out-of-pocket (75/25 CMS match on M&O): $1.9M-$4M/year

National Scale

Context: Tennessee pays Deloitte $823 million for one state's broken eligibility system. The entire national auto-enrollment platform costs $825 million to $1.685 billion over ten years... roughly equivalent to what one state pays one vendor for a system that produces 69% error rates.

Revenue Model for Contractor:

Palantir makes $26 million per year from ICE. This contract generates $108 million to $500 million per year. The math is not close.

SOLUTION 4: AI-Powered Building Permit Platform

Contract Title: Parallel Review and AI-Assisted Building Permit Processing System Scope: City-to-statewide deployment of digital permitting, AI plan review, parallel departmental routing, and automated inspection scheduling.

System Architecture

Digital Intake: Unified online portal replacing in-person plan submission. Applicants upload architectural drawings, site plans, and engineering calculations. AI pre-screen identifies common deficiencies before formal submission.

AI Plan Review Engine: Computer vision scans architectural drawings against local building code, zoning ordinances, fire code, and accessibility requirements. Deterministic logic (not generative AI) ensures 97-100% accuracy. Catches the 80% of deficiencies that currently cause resubmissions.

Parallel Departmental Routing: All reviewing departments (Building, Zoning, Fire, Utilities, Structural, Engineering, Transportation, Planning) receive the plans simultaneously. A comment consolidation engine collects all feedback before returning a single, unified set of corrections to the applicant. No version conflicts. No contradictory instructions from different departments.

Automated Inspection: Mobile inspection app with GPS check-in, photo documentation, real-time upload. AI-optimized inspector routing (Salesforce Field Service-class). Predictive scheduling based on permit type and construction timeline.

Budget: City Pilot (Medium City, 200K-500K Population)

Scale-Up

ROI for a medium city (2,000 permits/year): - Current delay cost to economy: 2,000 permits x 30 extra days x $670/day = $40.2M/yr in carrying costs - With 60% reduction: saves $24.1M/yr in private sector costs - System cost: $7-$12M/yr - ROI: 2-3x in Year 1

SOLUTION 5: Smart Water Infrastructure Monitoring Network

Contract Title: IoT-Based Water Distribution Monitoring, Leak Detection, and Predictive Maintenance System Scope: Deployment of sensor networks across municipal water systems for real-time quality monitoring, leak detection, pressure management, and predictive pipe failure analysis.

System Architecture

Sensor Layer: Distributed IoT sensors at key nodes in the distribution network (treatment plant outlets, pump stations, pressure zones, dead ends, high-risk corridors). Monitors pressure, flow, turbidity, chlorine residual, pH, temperature, and acoustic signatures (for leak detection).

Connectivity: LoRaWAN for dense urban networks (low power, long range). NB-IoT/LTE-M for wider coverage. Cellular backhaul for critical nodes. Redundant connectivity at treatment plants.

Analytics Platform: Machine learning models trained on flow patterns, pressure profiles, and historical failure data to predict pipe breaks before they occur. Anomaly detection for water quality events. Digital twin of the distribution network for scenario modeling.

SCADA Integration: API bridge to existing Supervisory Control and Data Acquisition systems. Does not replace SCADA... augments it with predictive intelligence and remote monitoring capabilities.

Smart Meters (Optional): AMI (Advanced Metering Infrastructure) for customer-side monitoring. Real-time consumption data. Automated leak alerts to customers. Demand forecasting.

Budget: Single System Deployment (by Size)

National Scale

ROI: A system losing 18% of treated water saves approximately $500,000 to $2 million per year per large utility in recovered water alone. Predictive maintenance avoids catastrophic pipe failures ($500K to $5M per main break). The Flint analogy: $400 million in crisis response avoided by $200,000 in monitoring. Prevention-to-response cost ratio: 1:2,000.

SOLUTION 6: AI-Optimized School Bus Operations (75% Depth)

Contract Title: Integrated School Transportation Management System Scope: AI route optimization, GPS telematics, parent notification, predictive maintenance, electric fleet transition planning, and vehicle-to-grid energy management.

The Absurdity

The United States operates the largest mass transit system on earth... 480,000 vehicles, 26 million daily passengers, $28 billion per year... and optimizes it with the same technology FedEx abandoned in 1986. In most of America's 13,000 school districts, the transportation coordinator draws bus routes using laminated maps, colored markers, and decades of institutional memory that lives in one person's head. When that person retires, the knowledge evaporates and the district starts from scratch. MIT built an algorithm for Boston Public Schools in 2017 that cut 50 buses in 30 minutes of compute time, saving $5 million per year. The result was published in a peer-reviewed INFORMS journal. It was national news. And then approximately 12,900 of America's 13,000 school districts continued drawing routes with markers. The algorithm exists. The savings are proven. The software is commercially available. School districts buy it at the same rate that Americans adopt the metric system... which is to say, they acknowledge it is superior and then do nothing.

System Architecture

The platform consists of five integrated modules:

Module 1: AI Route Optimization Engine Constraint-satisfaction algorithm incorporating student addresses, school locations, bell times, walk zones, hazard boundaries, special needs transportation requirements, maximum ride times, vehicle capacities, and driver shift rules. The engine re-optimizes nightly as students enroll, withdraw, or change addresses. The MIT/Boston implementation demonstrated that the algorithm can reassign 30,000 students across 650 routes in under 30 minutes... a task that takes human coordinators 2 to 4 weeks. The engine also models bell-time scenarios, since staggering school start times by 15 minutes can eliminate 8 to 12% of fleet requirements without changing a single route.

Module 2: GPS Telematics and Fleet Management Real-time vehicle tracking via OBD-II telematics devices installed on every bus. Monitors location, speed, idle time, hard braking events, engine diagnostics, and fuel consumption. Geofencing triggers automated alerts when buses deviate from assigned routes. Integrates with parent notification (Module 3) to provide accurate ETAs. Samsara, the dominant telematics provider, charges $27 to $50 per vehicle per month for hardware plus SaaS. At scale, a 500-bus district pays $162,000 to $300,000 per year for complete fleet visibility.

Module 3: Parent Communication and Rider Verification Mobile app providing real-time bus location, estimated arrival time, boarding/alighting notifications, and two-way messaging with transportation office. RFID or QR-code student identification at boarding confirms the correct student boarded the correct bus. Absence notifications flow automatically to attendance systems. For districts with custody complications (which is most districts), the system enforces court-ordered transportation arrangements without relying on driver memory.

Module 4: Predictive Maintenance and Asset Management Engine diagnostic data from telematics feeds into a maintenance scheduling engine. Predicts brake wear, battery degradation, tire replacement, and engine failure based on mileage, terrain, and driving patterns. Converts maintenance from reactive (bus breaks down, students stranded, substitute bus dispatched) to predictive (bus flagged for service three days before failure). Industry benchmarks show $2,800 per bus per year in maintenance savings from predictive scheduling. For a 500-bus fleet, that is $1.4 million annually.

Module 5: Electric Fleet Transition and V2G Energy Management Fleet electrification planning module that models total cost of ownership for diesel-to-electric conversion by route. Identifies routes best suited for electric buses based on distance, terrain, and charging window. Manages depot charging schedules to minimize demand charges. For districts with vehicle-to-grid (V2G) capability, the module manages energy arbitrage... selling stored electricity back to the grid during peak demand hours when buses sit idle (roughly 18 hours per day, plus weekends, plus summer). Beverly, Massachusetts earned $60,000 from three V2G buses over three summers. At scale, V2G revenue of $5,000 to $10,000 per bus per year transforms the school bus fleet from a pure cost center into a distributed energy asset.

Budget: District Pilot (500-Bus District)

Electric Fleet Capital (500-Bus District):

National Scale (480,000 Buses, 13,000 Districts)

National savings potential:

Key Vendor Landscape

Why an AI-native entrant wins: The incumbent landscape is fragmented. Routing, telematics, parent communication, maintenance, and electrification planning are sold by different vendors who do not integrate. A transportation coordinator manages five logins, five dashboards, and five vendor relationships. The AI-native play is the unified platform... one contract, one dashboard, one data model. Transfinder does routing but not telematics. Samsara does telematics but not routing. Neither does V2G management. The district that buys the unified platform gets 15-25% fuel savings, 7% fleet reduction, and V2G revenue in a single procurement. The fragmented vendor ecosystem cannot deliver that.

Federal Funding Pathways

The EPA Clean School Bus Program alone has distributed nearly $2 billion in two funding rounds. Priority districts (high-poverty, rural, tribal) receive up to $325,000 per bus... enough to cover 81% to 93% of the purchase price of an electric bus. The remaining cost is recouped within 3 to 5 years from diesel fuel and maintenance savings. After that, the bus generates net revenue via V2G. The federal government is literally paying districts to convert to a fleet that makes money, and most districts have not applied.

SOLUTION 7: FAFSA and Financial Aid Modernization (75% Depth)

Contract Title: Automated Financial Aid Determination and Unified Student Loan Platform Scope: Auto-populated FAFSA using FA-DDX (already authorized), AI-assisted aid packaging, real-time eligibility updates, unified loan servicing data layer.

Budget Summary

Context: The national AI modernization costs less than the Accenture contract that maintains the current broken system. FSA has already spent $581M on FAFSA simplification and produced 55 defects. A clean-sheet AI build for the same money would deliver a working system.

Institutional savings from AI-assisted packaging: 20% staff efficiency gain across 5,400 institutions = $200M-$400M/yr in labor savings.

SOLUTION 8: AI-Powered Property Tax Assessment (75% Depth)

Contract Title: Machine Learning Appraisal Platform with Equity Auditing Scope: AI mass appraisal, satellite/aerial imagery integration, automated appeals triage, equity monitoring dashboard.

Budget Summary

Revenue context: Applied against $630B in annual property tax revenue, the platform costs 0.08% to 0.16% of collections. Cook County's reform saved homeowners $1.9B. The national equity dividend from accurate assessment is in the tens of billions.

Key vendors and pricing: - C3.ai (Riverside County): $6.375M / 5 years / 460K parcels = $2.77/parcel/year - EagleView (aerial imagery): ~$140K/county/year, 900% average ROI - Tyler Technologies (iasWorld): $30M for Cook County, $18M for NYC

SOLUTION 9: Unified IEP Management and Medicaid Recovery Platform (75% Depth)

Contract Title: Integrated Special Education Compliance, Progress Monitoring, and Medicaid Billing System Scope: AI-assisted IEP goal writing, automated compliance tracking, progress monitoring, parent portal, Medicaid claims integration.

Budget Summary

Self-liquidating ROI (district level): - Medicaid recovery improvement: $900K/yr (from 70% to 90% capture) - Litigation avoidance (2-4 hearings): $100K-$400K/yr - Teacher retention (2-3 saved): $30K-$75K/yr - Year 1 ROI: Platform pays for itself on Medicaid alone

National impact: - Medicaid recovery: $400M-$1.8B/yr in currently unclaimed federal funds - Litigation reduction: $100M-$200M/yr - Teacher time recovered: $850M-$1.2B/yr in wage-equivalent value - Teacher retention savings: $180M-$300M/yr

SOLUTION 10: NG911 and AI Emergency Services Platform (75% Depth)

Contract Title: Next Generation 911 Transition with AI Call Triage and Predictive Dispatch Scope: ESInet deployment, cloud-native call handling, AI triage for non-emergency calls, real-time translation, text/video 911, predictive unit positioning.

Budget Summary

Three NHTSA/NTIA implementation models:

Market validation: Axon spent $1.27B in 90 days acquiring Prepared + Carbyne. RapidSOS raised $450M+. The CAD market is projected to grow from $3.8B to $20.6B by 2036. The market is real and the money is flowing... just not from the government.

Funding gap: States collect $3.8B-$4B/yr in 911 surcharges. The transition costs $9.5B-$12.7B. The annual federal grant program provides $109M. At current federal funding levels, full national NG911 takes 15-25 years. With a $10-$15B appropriation (bills introduced but not passed), it takes 7-10 years.

PART III: THE BUSINESS CASE

In which we prove that even Machiavelli would choose civic contracts over military ones.

This section is addressed to the AI company executive who has no interest in public welfare and cares only about revenue, margin, lock-in, and shareholder value. We demonstrate that even from this perspective... especially from this perspective... civic modernization is the superior business.

3.1 Revenue: The Math Is Not Close

The combined civic portfolio generates more annual revenue than Palantir's entire federal business ($970.5M in 2025). And it does so across thousands of state and local jurisdictions, eliminating the concentration risk of relying on a handful of federal agencies that change leadership every four years.

3.2 Lock-In: The 99-Year Contract

Military contracts are awarded competitively and rebid every 5 to 10 years. A new administration can cancel a program, reduce funding, or rebid the contract to a competitor.

Civic infrastructure contracts are permanent. Once a state deploys a Medicaid eligibility system, it cannot remove it. The system becomes the government. Deloitte's Medicaid systems run in 25 states covering 53 million enrollees. Switching vendors would require multi-year data migrations, retraining thousands of state employees, and risking service disruptions that affect millions of people. No governor takes that risk. The contract, once won, is effectively permanent.

Tyler Technologies derives 86.8% of its revenue from recurring sources. It covers 97 million parcels and generates 21 million annual tax bills. The switching cost for a county to replace Tyler's assessment system is so high that Tyler can charge 18-22% annual maintenance indefinitely. That is the business model of civic infrastructure: win once, bill forever.

FAST Enterprises has built DMV systems in dozens of states. Not one state has ever replaced a FAST system with a competitor. The vendor becomes the institutional memory. The contractor's employees know the system better than the government's employees. Replacing the vendor means losing the knowledge. The lock-in is not contractual. It is cognitive.

A military contractor wins a 10-year deal. A civic infrastructure contractor wins a 99-year annuity.

3.3 Data: The Most Valuable Dataset on Earth

Palantir's military contracts give it access to intelligence data that is classified, compartmented, and legally restricted from commercial use. The data is valuable for government work. It is worthless for everything else.

Civic infrastructure contracts give the contractor access to the most valuable dataset in commercial history: the complete economic profile of every American. Income data (Medicaid), property data (assessments), location data (traffic), education data (FAFSA), health service data (IEP/Medicaid), utility data (water), and emergency response patterns (911). Properly anonymized and aggregated, this data is worth billions in adjacent markets: insurance, real estate, public health research, urban planning, and infrastructure finance.

The military contractor gets classified satellite imagery. The civic contractor gets the operating system of the American economy.

3.4 Political Risk: Zero Boycott Risk

Palantir's ICE contract generated sustained activist campaigns, employee protests (at other companies that partnered with Palantir), Congressional investigations, and investigative journalism. Peter Thiel's political associations amplify the reputational risk.

No activist has ever organized a boycott against a company that helps people enroll in Medicaid. No employee has ever staged a walkout because their company builds better school bus routes. The civic portfolio has zero reputational downside and considerable upside: the company that modernizes 911 and saves 10,000 lives per year earns a narrative that money cannot buy.

3.5 The Consolidation Proof

The market is already validating this thesis. Axon Enterprise (a body camera company) spent $1.27 billion acquiring AI 911 companies. Tyler Technologies ($2.1B revenue) dominates local government software across courts, tax, permitting, and public safety. Deloitte controls $6 billion in Medicaid IT. FAST Enterprises has quietly locked up the DMV market.

These companies are not AI companies. They are legacy technology and services firms that stumbled into civic infrastructure because nobody else was competing for it. The AI companies were busy chasing $287 million ICE contracts.

The opportunity for an AI-native firm to disrupt Tyler, Deloitte, and FAST across all ten civic domains is not theoretical. It is the single largest uncontested market in the technology industry.

3.6 The Unified Pitch

Imagine walking into the office of every governor in America with the following pitch:

"We will build the operating system for your state government. One platform that handles traffic enforcement, court scheduling, Medicaid enrollment, building permits, water monitoring, school bus routing, financial aid, property assessment, special education management, and 911 dispatch. The federal government pays 90% of the Medicaid module. The traffic cameras pay for themselves from fine revenue. The IEP platform is self-liquidating on Medicaid recovery alone. The building permits module generates 2-3x ROI for every developer in your state.

The total contract value: $500 million over 10 years. Your state's out-of-pocket after federal matching: $50 million.

For $50 million, you get a state government that works.

Or you can keep paying Deloitte $823 million for a Medicaid system that produces a 69% error rate."

No governor turns down that meeting.

PART IV: THE MISSING API

In which we add up the damage, name the pattern, and explain why the richest country on earth runs its government like a fax machine.

4.1 The Common Thread

Across all ten failures, the pattern is identical. The government possesses the data needed to serve citizens. The citizens possess the same data. The technology to connect these databases exists and has for decades. The legal authority has been granted. The funding is available. And yet, human beings physically transport information between systems that should talk to each other automatically.

The DMV asks you to prove your address. The IRS already has it. Medicaid asks you to prove your income. The IRS already has it. The FAFSA asks you to report your tax data. The IRS already has it. The property assessor drives past your house. The satellite already imaged it. The 911 dispatcher asks where you are. Your phone already knows. The school bus coordinator draws routes on paper. The algorithm already optimized them. The building department reviews plans sequentially. The AI already flagged the errors. The court schedules everyone for 9 AM. The constraint solver already found the optimal time. The IEP teacher spends three hours writing progress notes. The AI already drafted them. The water utility discovers a burst main when someone calls in a flood. The sensor already detected the pressure drop.

In every case, the information exists. The technology exists. The legal authority exists. What does not exist is the integration layer... the API... that connects government databases to each other and to the citizens they serve.

4.2 Why the API Does Not Exist

Three structural reasons explain the absence.

Vendor incentives are misaligned. The incumbents (Deloitte, Tyler, Accenture, FAST) profit from complexity, not efficiency. A Deloitte Medicaid system that requires $823 million and 25 years of change orders is more profitable than a clean, automated system that costs $50 million and works. The vendors are paid to build the maze, not to solve it.

Government procurement is structurally hostile to integration. Each agency procures its own systems independently. The DMV issues one RFP. The Medicaid office issues another. The court system issues a third. Nobody procures the integration between them because no single agency owns the integration. The API falls into the gap between bureaucracies.

AI companies chase the wrong market. The defense and intelligence community is a known buyer. The procurement process, while complex, is well-documented. The contracting officers speak the same language as the vendors. State and local government is a fragmented market with 50 states, 3,143 counties, and 19,502 municipalities, each with its own procurement rules, budget cycles, and technology infrastructure. AI companies see this fragmentation as a barrier to entry. They should see it as a moat. Once you build the integration layer for one state, you can replicate it across all 50 at marginal cost.

4.3 The $30 Billion Idiot

Palantir Technologies generated $2.9 billion in total revenue in 2024. Its market capitalization exceeds $200 billion. The company has 4,000 employees and relationships with every major intelligence agency in the Western world.

With those resources, Palantir could have built the civic integration layer for the entire United States. It could have won Medicaid contracts in 25 states (currently held by Deloitte), DMV contracts in 30 states (currently held by FAST), court management contracts in hundreds of counties (currently held by Tyler), and 911 infrastructure contracts across 5,748 PSAPs.

The combined contract value of these civic domains, based on the SOWs presented in this paper, exceeds $30 billion over 10 years. That is 10x Palantir's current annual revenue. The contracts come with federal matching funds that subsidize 75-90% of development costs. They create permanent infrastructure dependency that no administration can undo. They generate richer data than any military application. And they do not produce investigative journalism about your company.

Instead, Palantir chose $287 million from ICE.

This is not a strategic disagreement. This is a $30 billion missed opportunity. The most profitable AI contract in history is the one nobody is bidding on, and the company best positioned to win it built a deportation app instead.

4.4 The Integration Tax

Every failure documented in Part I imposes a cost. We have reported those costs individually. Now we consolidate them into a single table, because the aggregate reveals something the individual cases obscure: the total price of the Missing API is not a line item. It is a tax. An invisible, recurring, regressive tax levied on every American who interacts with government, which is every American.

Table 4.4a: The Integration Tax by Domain

Now for the aggregate.

Table 4.4b: Annual Integration Tax (Conservative Estimate)

Compare this to three benchmarks:

That last row is the one that matters. The annual cost of not building the API exceeds the annual cost of building it by roughly two orders of magnitude. The Integration Tax is $170 billion to $285 billion per year. The API costs $15 billion to $30 billion over ten years, or $1.5 billion to $3 billion per year amortized.

The United States is paying somewhere between 57 and 190 dollars in damage for every dollar it would cost to fix the problem. This is not fiscal conservatism. It is not bureaucratic caution. It is the most expensive decision the country makes every year, and it makes it by not making a decision at all.

If a private company operated this way... spending 100x the cost of a software integration on the manual processes that software would eliminate... the board would fire the CEO before lunch. When government does it, we call it "the way things are."

4.5 The Resistance Ratchet (Borrowed from EEDTM)

The EEDTM (Elite Extraction with Differential Targeting Model), a framework developed for tracing wealth extraction across historical cases from Haiti to the American subprime crisis, identifies a structural pattern called the Resistance Ratchet. The concept is simple: when one extraction mechanism is blocked, the system does not stop extracting. It shifts to another mechanism. Theta... the total elite capture rate... is preserved. Only the channel changes.

The Resistance Ratchet applies with uncomfortable precision to civic modernization failure.

Consider the DMV. Thirty-nine states now offer online registration renewal. This is progress. It is celebrated. And it is carefully circumscribed. You can renew your registration online, but you cannot transfer a title online in most states. You can change your address online, but you cannot obtain a Real ID without appearing in person. Every in-person requirement that is eliminated gets replaced, or the remaining in-person requirements are jealously preserved.

The function being preserved is not the transaction. It is the human middleware role itself.

When states implemented online Medicaid applications... partially answering the integration problem... they still required paper-based annual renewals that produced the 69% procedural disenrollment rate documented in Failure 3. The front door was modernized. The back door stayed broken. The result: people could enroll online and then get churned off by mail. The system absorbed the reform and maintained its dysfunction.

When COVID forced courts to adopt remote hearings, failure-to-appear rates dropped and case processing improved. Then the pandemic ended and approximately 70% of jurisdictions reverted to in-person requirements for hearings that last three minutes. The technology worked. The system rejected it anyway.

When the FUTURE Act authorized the IRS to share tax data directly with Federal Student Aid in 2019, eliminating the legal barrier to auto-populated FAFSA, FSA spent five years and $581 million failing to implement it. The law passed. The engineering did not happen. The barrier was not legal. It was institutional.

This is the Resistance Ratchet operating in civic infrastructure. Each time a partial reform succeeds, the system reorganizes around the remaining inefficiencies. Online DMV? Require in-person for everything else. Online Medicaid enrollment? Keep paper renewals. Remote courts? Revert post-COVID. Authorized data sharing? Fail the implementation.

The pattern explains why piecemeal reform fails. Modernizing one system while leaving the others broken does not reduce the Integration Tax proportionally. It redistributes it. The citizen who no longer waits at the DMV still waits at the courthouse, still fills out the Medicaid renewal, still navigates the FAFSA, still overpays property taxes. The total burden shifts but does not shrink, because no single reform addresses the underlying architecture.

This is why only a unified platform succeeds. The Missing API is not ten separate projects. It is one project with ten entry points. When the integration layer exists... when government databases talk to each other... the human middleware function is eliminated at the architectural level. There is no remaining mechanism for the Ratchet to shift to. You cannot preserve the plastic-chair-and-folder model when the folder's contents are already in the system.

Piecemeal reform hands the Ratchet a new channel. The unified API kills the Ratchet entirely.

4.6 The Race Dimension

Everything in Part I is worse if you are Black.

This is not rhetoric. It is documented in the data we have already presented, domain by domain, and it adds up to something that has a name in the EEDTM framework: Gamma. The differential targeting coefficient. The multiplier that describes how much more a system extracts from one population than another, holding the mechanism constant.

In the EEDTM, Theta (~0.80) measures total elite capture. Gamma measures the ratio of extraction between populations. A Gamma of 1.0 means everyone is extracted equally. A Gamma above 1.0 means Black and brown populations bear disproportionate cost. Across the ten failures documented in this paper, Gamma is never 1.0. It is always above it. Often far above it.

Table 4.6: The Racial Gamma Across Ten Civic Domains

The Gammas in this table are conservative. They measure only the documented, quantified disparity within each system. They do not capture compounding. A Black mother who is 20% more likely to be stopped in traffic, who pays 12.7% more in property taxes, whose child is 40% more likely to be misidentified in special education, who loses Medicaid during a procedural churn, whose FAFSA was delayed by a system that disproportionately failed her demographic, whose 911 call takes longer to locate and process... that mother does not experience these as ten independent 1.2x to 2.0x penalties. She experiences them as a single, compounding architecture of disadvantage.

And the compounding is multiplicative, not additive. If each system imposes an average Gamma of 1.3x across ten domains, the composite Gamma is not 1.3 x 10 = 13. The composite Gamma approaches 1.3^10, which is 13.8. A person who is 30% more disadvantaged in each of ten systems is not 300% disadvantaged overall. They are structurally excluded.

This is the racial dimension of the Missing API. The integration failure is not race-neutral. It cannot be, because the systems it fails to connect were themselves built on racially disparate foundations. The property tax system was designed during Jim Crow. The traffic enforcement system operates through officers with documented racial bias in stop rates. The Medicaid non-expansion states are former Confederate states. The water systems that fail are in Flint and Jackson and the Mississippi Delta. The FAFSA disaster hit Pell-eligible institutions hardest, and those institutions serve Black and Latino students.

The Missing API is itself a mechanism of racial extraction.

Not because anyone designed it to be. Not because a policy-maker sat in a room and said, "let us build government systems that harm Black people." But because the absence of integration preserves the human discretion layer... the officer who decides whom to stop, the judge who sets bail by gut, the assessor who drives through Black neighborhoods less frequently, the dispatcher who routes slower. When the system depends on human middleware, the system inherits human bias. When the algorithm replaces the discretion, the bias has no channel.

A speed camera does not see race. An auto-enrollment engine does not see race. An AI property assessment does not see race. A constraint-satisfaction court scheduler does not see race. A sensor network does not see race.

The unified API does not eliminate racism. But it eliminates the civic infrastructure through which racism currently operates. It closes the Gamma channel in ten systems simultaneously. No other single intervention... not diversity training, not bias audits, not community review boards... addresses the racial dimension across all ten domains at once.

The Missing API is, incidentally, the most effective civil rights technology nobody has proposed.

4.7 The International Comparison

The Missing API is not a global problem. It is an American problem. Other countries have built it.

Estonia: X-Road

Estonia, a country of 1.3 million people with a GDP smaller than that of Fresno, California, built a unified digital government backbone in 2001. X-Road is a secure data exchange layer that connects every government database in the country. When an Estonian citizen is born, they receive a digital identity. When they start school, the education system knows. When they earn income, the tax authority knows. When they need healthcare, the system knows.

Estonians do not go to a government office to prove things the government already knows. They sign documents digitally (average time: 5 seconds per signature, saving an estimated 800 years of working time per year across the population). They vote online. They file taxes in under five minutes... the system pre-populates the return because it already has the data.

The Estonian government estimates that X-Road saves 1,400 years of working time annually. The entire system cost approximately $70 million to build. The United States spends more than that every year maintaining the COBOL systems that run its Social Security Administration.

India: Aadhaar + UPI

India enrolled 1.4 billion people in a biometric digital identity system (Aadhaar) and then built a unified payment interface (UPI) on top of it. In 2024, UPI processed 16.6 billion transactions in a single month. Government benefits are disbursed directly to bank accounts linked to Aadhaar. No middleman. No paper form. No plastic chair.

The Indian government estimates that Aadhaar-linked direct benefit transfers saved $33 billion over ten years by eliminating ghost beneficiaries and intermediary leakage. India spent approximately $1.5 billion building Aadhaar. The return on investment: 22x in a decade.

A country where 200 million people lacked bank accounts in 2014 now processes more real-time digital payments per month than the United States, the United Kingdom, and the European Union combined.

Singapore: GovTech

Singapore's Government Technology Agency operates a single digital platform (LifeSG) through which citizens access 40+ government services. A birth triggers automatic enrollment in healthcare, education waitlists, and parental benefits. A death triggers automatic notifications to banks, insurers, and government agencies. Life events are processed end-to-end without the citizen serving as courier.

Singapore's approach is simple: the government has the data, the government should use the data, and the citizen should never have to tell the government something the government already knows.

Denmark: Digital Post and NemID/MitID

Denmark abolished paper government correspondence in 2014. Every Danish citizen over 15 has a mandatory digital mailbox. Government agencies communicate exclusively through this channel. Tax returns are pre-filled and require confirmation, not preparation. Denmark processes 92% of government correspondence digitally. The remaining 8% are exemptions for elderly citizens who opted out.

The Comparison

The richest country in the table is the only one without a unified digital government platform. A nation that put a man on the moon, built the internet, and operates the most sophisticated military technology stack on earth cannot connect its DMV database to its IRS database.

The standard excuse is federalism. The United States has 50 states, 3,143 counties, and 19,502 municipalities, each with independent authority over its own systems. No central government can mandate integration across all of them. This is true. It is also irrelevant.

India has 28 states, 8 union territories, 22 official languages, and 1.4 billion people. It built Aadhaar anyway. Estonia rebuilt its entire government technology stack while still recovering from Soviet occupation. Denmark convinced an entire nation to abandon paper mail. Federalism is a real structural constraint. It is not a sufficient explanation for a country that spends $100 billion per year on federal IT and operates the world's largest cloud computing infrastructure.

The real explanation is simpler. Those countries had a government that decided to build the API. The United States has not decided. Not because it cannot. Not because the technology is hard. Not because the money is unavailable. But because the vendors who profit from the current system lobby against integration, the agencies that would lose headcount resist automation, and the AI companies that could build it are busy chasing $287 million from ICE.

Federalism is both the excuse and the opportunity. It means the first state to build the unified platform becomes the proof of concept for the other 49. It means a governor, not the President, can make the decision. It means the contract can be won one state at a time, which is exactly how Tyler, Deloitte, and FAST built their empires... one state at a time, one broken system at a time, over decades.

The difference is that those companies built fragmented, siloed systems that do not talk to each other. The Missing API company builds the one that does.

V. CONCLUSION: THE CONTRACT THAT WRITES ITSELF

She is still sitting in the plastic chair.

The woman from the Introduction. She has her Social Security card, her tax return, her utility bill, and her child's birth certificate. She has been waiting for forty-five minutes. She will wait for another thirty. When she reaches the counter, a government employee will type information from government documents into a government database, and the woman will drive home having completed a transaction that a functioning API would have handled while she slept.

She does this multiple times per year. DMV. Medicaid renewal. FAFSA. Property tax appeal. IEP meeting. She is the human middleware of American government, and she has been doing this job... unpaid, uncredited, and unnecessary... for her entire adult life.

Now multiply her by 330 million.

We have documented ten civic systems where the absence of basic data integration forces citizens into the role of biological connective tissue between government databases that should talk to each other automatically. We have presented ten complete Statements of Work, priced to the individual software license, with budgets validated against actual government contracts. We have demonstrated that the civic AI market is objectively superior to the military AI market across every business dimension. And we have quantified the cost of inaction.

The numbers, consolidated:

Table V: The Missing API... Full Accounting

Every module has a proven proof of concept. Boston proved the bus algorithm. Louisiana proved auto-enrollment. Honolulu proved AI plan review. Riverside proved AI assessment. COVID proved remote courts. Estonia proved the unified backbone. India proved it scales. The research is done. The pilots are completed. The international precedents exist. The only thing missing is the company that walks into the governor's office with the unified contract.

The racial dimension makes this urgent. The Integration Tax is regressive by design and racist by operation. A composite Gamma approaching 14x means Black Americans pay an order of magnitude more for the privilege of serving as their government's unpaid data couriers. The Missing API does not merely save money. It closes the channel through which ten systems simultaneously extract disproportionate cost from Black and brown citizens. No diversity initiative, no bias training, no community review board addresses ten systems at once. The API does.

The Resistance Ratchet makes piecemeal reform futile. Online DMV preserves paper Medicaid. Remote courts revert post-COVID. Authorized data sharing fails in implementation. Each partial modernization is absorbed by the system and the total burden redistributed rather than reduced. Only the unified platform... the API that connects all ten systems at the architectural level... kills the Ratchet. There is no remaining mechanism to shift to when the integration layer eliminates the human middleware function entirely.

This paper does not argue that government should modernize. Everyone knows that. It does not argue that citizens deserve better services. Everyone knows that too. It does not argue that racial equity matters. The people who need convincing on that point are not reading academic journals.

This paper argues something more uncomfortable: that the AI industry's failure to pursue civic modernization is simultaneously a moral failing, a business failing, and a civil rights failing, and that the business failing may be the one that finally moves the needle, because the people who build AI companies respond to revenue projections faster than they respond to justice claims.

The civic market is larger than the military market. More profitable. More defensible. More durable. More legally protected. Better funded. Less politically volatile. And completely uncontested by the companies best equipped to win it.

This is not a policy paper. It is a business plan. The ten SOWs presented in Part II are not white papers. They are contracts waiting for a signature. The business case presented in Part III is not an academic exercise. It is a pitch deck. The Integration Tax quantified in Section 4.4 is not an abstraction. It is the revenue opportunity for whichever company decides to stop chasing $287 million from ICE and start bidding on the $30 billion in civic infrastructure that nobody else wants.

The woman in the plastic chair will keep waiting until someone builds the API. The country will keep paying $170 billion to $285 billion per year in integration tax until someone builds the API. Ten thousand people will keep dying from 911 location failures until someone builds the API. Black Americans will keep paying a compound Gamma penalty across ten systems until someone builds the API.

The technology exists. The legal authority exists. The money exists. The proof of concepts exist. The international precedents exist. The business case is overwhelming.

The contract is sitting on the governor's desk. It is the most profitable, most defensible, and most socially consequential AI contract in history.

All it needs is a signature.

REFERENCES

Alternative Fuels Data Center. (n.d.). Electric school bus operating cost savings. U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy. Retrieved February 2026, from https://afdc.energy.gov/vehicles/electric-buses

American Hospital Association. (2024). Uncompensated hospital care cost fact sheet. AHA. Retrieved February 2026, from https://www.aha.org/fact-sheets/2020-01-06-fact-sheet-uncompensated-hospital-care-cost

American Society of Civil Engineers. (2021). 2021 report card for America's infrastructure: Drinking water. ASCE. Retrieved February 2026, from https://infrastructurereportcard.org/cat-item/drinking-water-infrastructure/

American Water Works Association. (n.d.). Water loss control. AWWA. Retrieved February 2026, from https://www.awwa.org/resource/water-loss-control/

Archistar. (2024, October). Archistar secures contract with the City of Austin for groundbreaking AI-powered automated building permit assessment platform [Press release]. PR Newswire. Retrieved February 2026, from https://www.prnewswire.com/apac/news-releases/archistar-secures-contract-with-the-city-of-austin-for-groundbreaking-ai-powered-automated-building-permit-assessment-platform-302269544.html

Arnold Ventures. (n.d.). About the Public Safety Assessment (PSA). Advancing Pretrial Policy and Research. Retrieved February 2026, from https://www.advancingpretrial.org/about-the-psa/

Avenancio-Leon, C., & Howard, T. (2022). The assessment gap: Racial inequalities in property taxation. The Quarterly Journal of Economics, 137(3), 1383-1434. https://doi.org/10.1093/qje/qjac009

Axon Enterprise. (2025a, September 23). Axon to acquire AI-powered 911 communications company Prepared, strengthening the public safety ecosystem from call to closure [Press release]. Retrieved February 2026, from https://investor.axon.com/2025-09-23-Axon-to-acquire-AI-powered-911-communications-company-Prepared

Axon Enterprise. (2025b, November 4). Axon to acquire Carbyne, uniting cloud infrastructure and AI to redefine the 911 experience [Press release]. Retrieved February 2026, from https://investor.axon.com/2025-11-04-Axon-to-Acquire-Carbyne,-Uniting-Cloud-Infrastructure-and-AI-to-Redefine-the-911-Experience

Bearden v. Georgia, 461 U.S. 660 (1983). https://supreme.justia.com/cases/federal/us/461/660/

Bellwether Education Partners. (2024). Who pays for special education? An analysis of federal, state, and local spending by states and districts. Bellwether. Retrieved February 2026, from https://bellwether.org/publications/who-pays-for-special-education/

Berry, C. R. (2021). Reassessing the property tax (Working Paper). University of Chicago, Harris School of Public Policy. Retrieved February 2026, from https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3800536

Bertsimas, D., Delarue, A., Eger, W., Hanlon, J., & Martin, S. (2020). Bus routing optimization helps Boston Public Schools design better policies. INFORMS Journal on Applied Analytics, 50(1), 37-49. https://doi.org/10.1287/inte.2019.1015

Brookings Institution. (2024, November). Fewer freshmen enrolled in college this year following troubling FAFSA cycle. Brookings. Retrieved February 2026, from https://www.brookings.edu/articles/fewer-freshmen-enrolled-in-college-this-year-following-troubling-fafsa-cycle/

C3.ai. (2023). Riverside County, California selects C3 AI as the standard for real estate property appraisal [Press release]. Business Wire. Retrieved February 2026, from https://c3.ai/riverside-county-california-selects-c3-ai-as-the-standard-for-real-estate-property-appraisal/

C3.ai. (n.d.). Riverside County drives 40% increase in model accuracy for property appraisal [Case study]. Retrieved February 2026, from https://c3.ai/customers/riverside-county-drives-40-increase-in-model-accuracy-for-property-appraisal/

Center for Appropriate Dispute Resolution in Special Education. (2023). IDEA dispute resolution data summary for U.S. and outlying areas: 2022-23. CADRE. Retrieved February 2026, from https://cadreworks.org/resources/data-resources/national-data

Centers for Medicare & Medicaid Services. (n.d.). 42 CFR Parts 433.112, 433.116: Federal financial participation in State expenditures; enhanced FFP for eligible E&E expenditures. CMS.

Children's Health Insurance Program Reauthorization Act of 2009 (CHIPRA), Pub. L. No. 111-3, 123 Stat. 8 (2009).

Clariti Software. (n.d.). Honolulu cuts review times by 70% with CivCheck's AI plan review software [Case study]. Retrieved February 2026, from https://www.claritisoftware.com/honolulu-cuts-review-times-by-70-percent-with-civchecks-ai-plan-review-software

Commonwealth Fund. (2024, March). Automatic enrollment in health insurance: A pathway to increased coverage for people with low income. Commonwealth Fund. Retrieved February 2026, from https://www.commonwealthfund.org/publications/fund-reports/2024/mar/automatic-enrollment-health-insurance-pathway-increased-coverage

Congressional Research Service. (2024a). Funding the transition to Next Generation 911 (NG911): Considerations for Congress (Report No. R48015). Library of Congress. Retrieved February 2026, from https://www.congress.gov/crs-product/R48015

Congressional Research Service. (2024b). The Individuals with Disabilities Education Act (IDEA) funding: A primer (Report No. R44624). Library of Congress. Retrieved February 2026, from https://www.congress.gov/crs-product/R44624

EagleView. (n.d.). Manitoba Assessment realizes +900% ROI and enhances efficiency with EagleView aerial imagery [Case study]. Retrieved February 2026, from https://www.eagleview.com/casestudy/manitoba-assessment-realizes-900-roi-enhances-efficiency-and-accuracy-with-eagleview-aerial-imagery/

Empire Center for Public Policy. (2024). The 411 on New York's 911 skim. Empire Center. Retrieved February 2026, from https://www.empirecenter.org/publications/the-411-on-new-yorks-911-skim/

Endrew F. v. Douglas County School District RE-1, 580 U.S. ___ (2017). https://supreme.justia.com/cases/federal/us/580/15-827/

Federal Communications Commission. (2024). Sixteenth annual report to Congress on state collection and distribution of 911 and enhanced 911 fees and charges. FCC. Retrieved February 2026, from https://www.fcc.gov/sites/default/files/16th-Annual-911FeeReport-2024.pdf

Federal Communications Commission. (2025). Improving wireless 911 caller location [Fact sheet]. FCC. Retrieved February 2026, from https://www.fcc.gov/general/911-fee-reports

Federal Student Aid. (2024). FSA annual report, FY2024. U.S. Department of Education, Office of Federal Student Aid. Retrieved February 2026, from https://studentaid.gov/

Flock Safety. (n.d.). Pricing. Flock Safety. Retrieved February 2026, from https://www.flocksafety.com/pricing

FUTURE Act (Fostering Undergraduate Talent by Unlocking Resources for Education Act), Pub. L. No. 116-91, 133 Stat. 1189 (2019).

GovConWire. (2025, August). Accenture Federal Services secures $936.7M TIVOD contract from Education Department. GovConWire. Retrieved February 2026, from https://www.govconwire.com/articles/accenture-federal-services-contract-award-department-education-student-aid

GovWin IQ. (2025). Federal artificial intelligence spending trends, FY 2022-2024. Deltek GovWin. Retrieved February 2026, from https://iq.govwin.com/neo/marketAnalysis/view/Federal-Artificial-Intelligence-Spending-Trends-FY-2022-2024/8401

Grand View Research. (2024). Computer-aided dispatch (CAD) market size, share & trends analysis report. Grand View Research. Retrieved February 2026, from https://www.grandviewresearch.com/industry-analysis/computer-aided-dispatch-cad-market

Individuals with Disabilities Education Act (IDEA), 20 U.S.C. Sections 1400-1482.

Infrastructure Investment and Jobs Act (Bipartisan Infrastructure Law), Pub. L. No. 117-58, 135 Stat. 429 (2021).

KFF. (n.d.-a). A closer look at the remaining uninsured population eligible for Medicaid and CHIP. KFF. Retrieved February 2026, from https://www.kff.org/uninsured/a-closer-look-at-the-remaining-uninsured-population-eligible-for-medicaid-and-chip/

KFF. (n.d.-b). Medicaid enrollment and unwinding tracker. KFF. Retrieved February 2026, from https://www.kff.org/medicaid/medicaid-enrollment-and-unwinding-tracker/

KFF Health News. (2024a, June 24). Medicaid for millions in America hinges on Deloitte-run systems plagued by errors. KFF Health News. Retrieved February 2026, from https://kffhealthnews.org/news/article/medicaid-deloitte-run-eligibility-systems-plagued-by-errors/

KFF Health News. (2024b, September 10). Errors in Deloitte-run Medicaid systems can cost millions and take years to fix. KFF Health News. Retrieved February 2026, from https://kffhealthnews.org/news/article/deloitte-run-medicaid-systems-errors-cost-millions-take-years-to-fix/

Koerth, M., & Denvir, D. (2017-2018). The tax divide [Investigative series]. Chicago Tribune and ProPublica Illinois. Retrieved February 2026, from https://www.propublica.org/series/the-tax-divide

Medicaid and CHIP Payment and Access Commission. (2024, November). State reported Medicaid unwinding data brief. MACPAC. Retrieved February 2026, from https://www.macpac.gov/publication/state-reported-medicaid-unwinding-data/

National Association of Home Builders. (2021, May). Government regulation in the price of a new home: 2021. NAHB. Retrieved February 2026, from https://www.nahb.org/-/media/NAHB/news-and-economics/docs/housing-economics-plus/special-studies/2021/special-study-government-regulation-in-the-price-of-a-new-home-may-2021.pdf

National Center for Education Statistics. (n.d.-a). Fast facts: Students with disabilities (64). NCES. Retrieved February 2026, from https://nces.ed.gov/fastfacts/display.asp?id=64

National Center for Education Statistics. (n.d.-b). Fast facts: Transportation (67). NCES. Retrieved February 2026, from https://nces.ed.gov/fastfacts/display.asp?id=67

National Emergency Number Association. (n.d.). 9-1-1 statistics. NENA. Retrieved February 2026, from https://www.nena.org/page/911statistics

National Highway Traffic Safety Administration. (2024). Overview of motor vehicle traffic crashes in 2023 (Research Note DOT HS 813 705). NHTSA. Retrieved February 2026, from https://crashstats.nhtsa.dot.gov/Api/Public/Publication/813705

National Highway Traffic Safety Administration & National Telecommunications and Information Administration. (2023). Next Generation 911 cost study. 911.gov. Retrieved February 2026, from https://www.911.gov/projects/next-generation-911-cost-study/

National Multifamily Housing Council & National Association of Home Builders. (2022). Regulation: 40.6 percent of the cost of multifamily development. NMHC. Retrieved February 2026, from https://www.nmhc.org/research-insight/research-report/nmhc-nahb-cost-of-regulations-report/

National Student Clearinghouse Research Center. (2024). Final fall enrollment trends, fall 2024. NSCRC. Retrieved February 2026, from https://nscresearchcenter.org/final-fall-enrollment-trends/

Office of the Assistant Secretary for Planning and Evaluation. (2013). CHIPRA mandated evaluation of Express Lane Eligibility: First year findings. ASPE, U.S. Department of Health and Human Services. Retrieved February 2026, from https://aspe.hhs.gov/reports/chipra-mandated-evaluation-express-lane-eligibility-first-year-findings-0

Office of the Assistant Secretary for Planning and Evaluation. (2015). CHIPRA mandated evaluation of Express Lane Eligibility: Final findings. ASPE, U.S. Department of Health and Human Services. Retrieved February 2026, from https://aspe.hhs.gov/reports/chipra-mandated-evaluation-express-lane-eligibility-final-findings

Ownwell. (2026, February). Ownwell raises $50M, launches national service to streamline property tax appeals and make home ownership more affordable [Press release]. PR Newswire. Retrieved February 2026, from https://www.prnewswire.com/news-releases/ownwell-raises-50m-launches-national-service-to-streamline-property-tax-appeals-and-make-home-ownership-more-affordable-302692103.html

Palantir Technologies. (2025). 2024 annual report (Form 10-K). U.S. Securities and Exchange Commission. Retrieved February 2026, from https://investors.palantir.com/files/2024%20FY%20PLTR%2010-K.pdf

Perry, A. M., Rothwell, J., & Harshbarger, D. (2018). The devaluation of assets in Black neighborhoods: The case of residential property. Metropolitan Policy Program, Brookings Institution. Retrieved February 2026, from https://www.brookings.edu/wp-content/uploads/2018/11/2018.11_Brookings-Metro_Devaluation-Assets-Black-Neighborhoods_final.pdf

Pierson, E., Simoiu, C., Overgoor, J., Corbett-Davies, S., Jenson, D., Shoemaker, A., Ramachandran, V., Barghouty, P., Phillips, C., Shroff, R., & Goel, S. (2020). A large-scale analysis of racial disparities in police stops across the United States. Nature Human Behaviour, 4(7), 736-745. https://doi.org/10.1038/s41562-020-0858-1

Prison Policy Initiative. (2017, February 7). Pretrial detention costs $13.6 billion each year. Prison Policy Initiative. Retrieved February 2026, from https://www.prisonpolicy.org/blog/2017/02/07/pretrial_cost/

Prison Policy Initiative. (2025). Mass incarceration: The whole pie 2025. Prison Policy Initiative. Retrieved February 2026, from https://www.prisonpolicy.org/reports/pie2025.html

RapidSOS. (2025, November 19). RapidSOS secures $100 million as it crosses one billion emergencies supported [Press release]. PR Newswire. Retrieved February 2026, from https://rapidsos.com/blog/rapidsos-raises-100m-to-scale-emergency-response-ai/

Research Nester. (2024). Computer-aided dispatch (CAD) market size & share, growth report 2037. Research Nester. Retrieved February 2026, from https://www.researchnester.com/reports/computer-aided-dispatch-cad-market/5014

Sances, M. W., & You, H. Y. (2024). How the property tax system harms Black homeowners and widens the racial wealth gap. Brookings Institution. Retrieved February 2026, from https://www.brookings.edu/articles/how-the-property-tax-system-harms-black-homeowners-and-widens-the-racial-wealth-gap/

State of Surveillance. (2025). Palantir $287M 2025: ICE immigration tracking machine. State of Surveillance. Retrieved February 2026, from https://stateofsurveillance.org/articles/surveillance/palantir-immigration-machine-287-million/

Tax Policy Center. (n.d.). How do state and local property taxes work? Urban-Brookings Tax Policy Center. Retrieved February 2026, from https://taxpolicycenter.org/briefing-book/how-do-state-and-local-property-taxes-work

Texas Public Radio. (2024, January 11). Texas school districts lose $300 million in federal special education funding. TPR. Retrieved February 2026, from https://www.tpr.org/education/2024-01-11/texas-school-districts-lose-300-million-in-federal-special-education-funding

Texas Tribune. (2024, October 28). Texas is slashing $607 million in Medicaid funding from program for students with disabilities. The Texas Tribune. Retrieved February 2026, from https://www.texastribune.org/2024/10/28/texas-shars-medicaid-special-education/

Tyler Technologies. (2015, October 5). Cook County, Illinois, signs $30 million contract with Tyler Technologies for integrated property tax solution [Press release]. Business Wire. Retrieved February 2026, from https://investors.tylertech.com/news/news-details/2015/Cook-County-Illinois-Signs-30-Million-Contract-with-Tyler-Technologies-for-Integrated-Property-Tax-Solution-10-05-2015/

Tyler Technologies. (2025). 2024 annual report. Tyler Technologies, Inc. Retrieved February 2026, from https://s201.q4cdn.com/385562537/files/doc_downloads/2025/03/Tyler-2024-Annual-Report_Digital_FINAL.pdf

U.S. Census Bureau. (n.d.). Building permits survey, annual data. U.S. Census Bureau. Retrieved February 2026, from https://www.census.gov/construction/bps/

U.S. Department of Education. (2024). Emergency FAFSA funding allocation, FY2024. U.S. Department of Education. Retrieved February 2026, from https://www.ed.gov/

U.S. Environmental Protection Agency. (2023). EPA's 7th drinking water infrastructure needs survey and assessment. EPA. Retrieved February 2026, from https://www.epa.gov/dwsrf/epas-7th-drinking-water-infrastructure-needs-survey-and-assessment

U.S. Environmental Protection Agency. (n.d.). Clean School Bus Program. EPA. Retrieved February 2026, from https://www.epa.gov/cleanschoolbus

U.S. Government Accountability Office. (2020). Medicaid information technology: Effective CMS oversight and states' sharing of claims processing and information retrieval systems can reduce costs (GAO-20-179). GAO. Retrieved February 2026, from https://www.gao.gov/products/gao-20-179

U.S. Government Accountability Office. (2023). Department of Education: Federal Student Aid system modernization project should better estimate cost and schedule (GAO-23-106376). GAO. Retrieved February 2026, from https://www.gao.gov/products/gao-23-106376

USASpending.gov. (n.d.). Contract to Palantir Technologies Inc. USASpending.gov. Retrieved February 2026, from https://www.usaspending.gov/award/CONT_AWD_70CTD022FR0000170_7012_GS35F0086U_4730

Vera Institute of Justice. (2017). The price of jails: Measuring the taxpayer cost of local incarceration. Vera Institute. Retrieved February 2026, from https://www.vera.org/publications/the-price-of-jails-measuring-the-taxpayer-cost-of-local-incarceration

Ward, J., & Eltagouri, M. (2025, April 6). Tyler Technologies Inc. contracts cost Illinois taxpayers $250M. Injustice Watch and Chicago Tribune. Retrieved February 2026, from https://www.injusticewatch.org/judges/court-administration/2025/costs-triple-delays-mount-in-tyler-technologies-contracts/

World Resources Institute. (n.d.). Electric school buses and vehicle-to-grid (V2G) revenue potential. WRI. Retrieved February 2026, from https://www.wri.org/

Author: Wesley Bertil Affiliation: Reparations Finance Lab Date: February 2026 Word Count: [To be calculated upon completion] Status: First Draft

This paper was produced using forensic policy analysis methodology developed as part of the BARSS (Bertil's Analytics Research Sciences & Sorceries) research program. The SOW costing methodology employs bottom-up cost engineering validated against public procurement records.