THE MACHINERY OF MASS INCARCERATION
A Structural Account of How the United States Built the World's Largest Carceral System — and Why It Stays Built
All factual claims in this work are sourced from the accompanying research dossier. Claims not present in the dossier are marked [NEW CLAIM — requires verification]. Dossier confidence levels [VH/VM/C/U/S] are noted parenthetically for contested or significant claims.
AUTHOR'S NOTE ON METHOD
This work does not argue for a particular policy outcome. It describes documented structural incentives. The reader is invited to form their own conclusions about what those incentives mean and what, if anything, should be done about them.
The tone occasionally runs sardonic. It earns the right to do so by staying factual. The irony in these pages is not manufactured — it arises, almost inevitably, from the documented contradictions between what institutions say they are for and what the evidence shows they actually do.
No prisoners are mocked here. No victims are dismissed. The people who built and maintain this system are not cartoon villains; they are responding to incentives that real institutions created and that real legislatures sustained. Understanding how something works is not the same as excusing it. But neither is condemning it the same as changing it.
We begin with a census.
CHAPTER ONE: A CENSUS OF THE MANAGED
"Numbers, when they get large enough, stop feeling like people."
On any given day in 2023, approximately 1.9 million people were confined in some form of detention in the United States. This figure comes from the Prison Policy Initiative's synthesis of Bureau of Justice Statistics data, federal Bureau of Prisons counts, and facility-level reporting across immigrant detention, youth facilities, and civil commitment centers.
To put 1.9 million people in a room — if such a room could exist — you would need a space that dwarfs the city of Philadelphia. You would have gathered more people than live in the entire state of Wyoming, Nebraska, or West Virginia. You would be managing a population roughly equivalent to the combined populations of San Francisco and Denver.
This is the population that the United States describes, in official terminology, as "under correctional control" through confinement. If you expand the aperture to include the 3.77 million adults on probation or parole — the Bureau of Justice Statistics' yearend 2023 figure — you arrive at a supervised population of approximately 5.6 million people, moving through their days under the legal jurisdiction of the criminal justice system. That number is larger than the population of Ireland. It is larger than the populations of New Zealand, Singapore, or Denmark.
Expand further still. The FBI maintains records — by its own count — for approximately 73.5 million Americans, a figure that works out to roughly one in three adults. This number requires immediate qualification, because the database is messy and the definition is contested. It includes everyone arrested on a felony charge regardless of whether they were convicted. It includes misdemeanor records when state agencies bother to report them. It does not consistently track unique individuals across states, meaning that a person arrested in Ohio, Texas, and Florida could theoretically occupy three separate records. But even discounted heavily for definitional slippage, the number that remains is strikingly large. Some portion of these 73.5 million Americans have served time; a larger portion have conviction records; a still larger portion carry arrest records that affect their ability to find housing, employment, and professional licenses even though they were never convicted of anything.
The Bureau of Justice Statistics does not publish a clean national count of Americans with felony convictions. The most rigorous attempt to calculate this figure comes from researchers Couture, Mauer, and Shannon, who published in the journal Demography in 2016 using life-table methods applied to BJS data. Their estimate: as of 2010, approximately 8 percent of all U.S. adults — roughly 19.8 million people — had ever been convicted of a felony. That figure had grown from approximately 3 percent in 1980. No reliable updated national estimate exists; the 2010 cutoff is a consequence of the methodological difficulty of counting a population that no single agency tracks comprehensively.
These numbers have names: Mississippi's imprisonment rate — the highest of any U.S. state — stands at 847 per 100,000 adults as of 2023. Massachusetts, the lowest, stands at 118. The ratio between them is roughly 7 to 1. Two jurisdictions, governed by the same federal constitution, applying versions of the same criminal code, with access to the same body of social science evidence, producing outcomes that differ by a factor of seven. This is not a small variance. This is not noise. This is a structural signal.
The U.S. overall incarceration rate — counting all facilities, using World Prison Brief methodology applied to 2022 BJS data — is approximately 541 per 100,000 people. This is not close to the international norm. It is not in the neighborhood of comparable wealthy democracies. Norway incarcerates at approximately 75 per 100,000. Germany at 76. The Netherlands at 69. The United States incarcerates at a rate roughly seven times higher than any of these countries. This fact is so well documented and so frequently cited that it risks becoming numbing. It should not become numbing. It is one of the most significant data points in American public policy, and its explanation is not obvious.
The incarceration rate has, it should be noted, declined substantially from its 2008 peak of approximately 760 per 100,000. The state and federal prison population in spring 2024 was approximately 13 percent below 2019 levels, according to the Vera Institute. This matters. It suggests that incarceration levels are not immutable — they rise and fall in response to policy choices, economic conditions, and political will. But the decline from 760 to 541 still leaves the United States at more than seven times the German rate. The direction of travel changed; the destination remains extreme by any international standard.
This chapter is a census. It counts. In the chapters that follow, we will ask what is being counted, why it was built this way, and — the most structural question of all — who benefits from the counting.
CHAPTER TWO: THE SCALE PROBLEM
On the difficulty of seeing something this large clearly.
The problem with describing mass incarceration in the United States is one of scale. Not scale in the sense of difficulty — the data is available, the numbers are public, the Bureau of Justice Statistics publishes annual reports. The problem is cognitive scale: human brains are not naturally equipped to think about 1.9 million people, $445 billion in annual expenditure, or a 346 percent real increase in corrections spending since 1977. These numbers are large enough to stop feeling like consequences and start feeling like weather.
So let us try several different scales.
The money scale. State and local governments spent $87 billion on corrections in 2021, adjusted for inflation to 2021 dollars. In 1977, the same inflation-adjusted calculation yields $19 billion. The increase — $68 billion in real spending — is a 346 percent expansion. Over the same period, policing spending grew from $47 billion to $135 billion (189 percent). Courts spending grew 65 percent. These are not incremental budget adjustments. They represent a sustained, deliberate, decade-by-decade reallocation of public resources toward the criminal legal system. The total 2025–2026 estimate for all criminal justice spending — policing, courts, corrections, immigration enforcement — is approximately $445 billion annually, according to the Prison Policy Initiative's February 2026 report. For comparison, the U.S. Department of Education's total discretionary budget in recent years has been approximately $79 billion. The criminal justice system costs, in one year, what six Department of Education budgets cost.
The employment scale. Approximately half of all correctional spending — roughly half of $87 billion in corrections alone — goes toward staff compensation. Correctional employment is not a side effect of mass incarceration; it is one of its primary economic functions, particularly in rural areas where few large employers remain. The Bureau of Labor Statistics reported a median annual salary for correctional officers of $53,300 in 2023 — about 10.9 percent above the median for all occupations. In California, that figure reaches $93,160. This is a career. These are mortgages. These are pensions. These are children in local schools. The political economy of correction begins right here, at the salary line.
The historical scale. The U.S. incarceration rate was not always this high. In 1977, the corrections spending base was $19 billion (2021 dollars). The prison population was a fraction of today's. Something happened between then and now. That something was not a sudden explosion of human wickedness. Crime rates — including violent crime — have fallen dramatically from their early 1990s peak. The murder rate in 2023 was lower than it was in the 1960s. Crime and incarceration, which might intuitively seem to move together, diverged sharply: incarceration continued to rise long after crime rates fell, and corrections spending continued to increase even as incarcerated populations declined.
This divergence is the first structural clue. Systems that grow independent of the problem they were designed to solve are exhibiting a phenomenon that public choice economists have a name for: they have developed their own constituencies, their own internal logic, their own reasons for perpetuating themselves that are independent of external conditions. Understanding mass incarceration requires understanding that it is not primarily a response to crime. It is also an economic arrangement, an employment system, a revenue mechanism, and a political structure — and it behaves accordingly.
The human scale. Behind every BJS statistic is a person who woke up this morning inside an institution. In 2024, 69 percent of the 657,500 people in local jails had not been convicted of anything. They were waiting — for hearings, for trial dates, for plea offers — unable to afford bail. The median pre-incarceration income of people in jail who could not meet bail was $16,233 (2020 dollars, from BJS survey data). They were not, as a rule, dangerous people awaiting trial for serious violent crimes. Many were there because a two-digit bank account balance met a three-digit bail requirement, and the math didn't work. The bail system — a structural feature, not an individual failure — produced this population. It keeps producing it.
Understanding the scale of American incarceration is not primarily a moral exercise, though moral conclusions are available to those who want them. It is an analytical prerequisite. A system this large, this expensive, this durable, does not persist because it is failing. It persists because, for significant constituencies, it is working exactly as designed — whether or not the design was ever explicit.
The chapters that follow are an attempt to make the design visible.
CHAPTER THREE: MILLION-DOLLAR BLOCKS
On the geography of public investment.
In 2006, two researchers produced a set of maps that reframed how policy analysts thought about incarceration and urban neighborhoods. Laura Kurgan, of Columbia University's Spatial Information Design Lab, and Eric Cadora, of the Justice Mapping Center, combined two datasets that had never been overlaid: the home addresses of incarcerated people (from state prison records) and the annual expenditure required to incarcerate them. Then they mapped the results by census block.
What appeared on those maps was a phenomenon they called "million-dollar blocks." In dense urban neighborhoods across the five cities they examined, there were individual city blocks — sometimes just a few hundred feet of sidewalk, two dozen row houses, a corner store — from which the state was spending more than $1 million every year to incarcerate residents. Not to educate them. Not to provide them healthcare. Not to maintain their streets or fund their schools. To incarcerate them, largely in facilities hundreds of miles away.
The maps were eventually exhibited at the Museum of Modern Art. They documented something that decades of policy debate had rendered invisible: the geography of public investment in communities experiencing the highest incarceration rates was dominated not by schools or hospitals or job training programs, but by the criminal justice system. Kurgan and Cadora's project description noted that in these neighborhoods, the criminal justice system had become "the predominant government institution." Not a presence among many — the predominant one.
This framing matters because it reorients the standard description. Incarceration is frequently discussed as an absence — these neighborhoods lack investment, lack resources, lack opportunity. The million-dollar block analysis demonstrates that they are not being ignored by government; they are receiving substantial government investment. It simply flows outward, toward institutions in other counties, other parts of the state, other economic ecosystems. The money does not circle back to fund the library, the vocational program, the mental health clinic. It travels to wherever the prison is.
The fiscal implication is significant. When researchers document that high-incarceration neighborhoods also have underfunded schools, overcrowded emergency rooms, and weak economic infrastructure, part of what they are documenting is a budget allocation decision, made repeatedly at the state level, about where public money goes. The million-dollar block is not a place that government forgot. It is a place that government priced carefully, counted precisely, and invested in heavily — via incarceration rather than any other mechanism.
Cadora's organization, the Justice Mapping Center, has extended this analysis to additional cities. The UCLA Million Dollar Hoods project adapted the methodology for California. The consistent finding: in high-incarceration zip codes, criminal justice expenditure is not marginal — it is the dominant form of public investment in the neighborhood.
A practical implication follows, one that is rarely examined in policy debates: reducing incarceration does not automatically redirect the savings to the communities from which the incarcerated population came. State corrections budgets and local school budgets are different appropriations, controlled by different bodies, subject to different political pressures. The savings from closing a prison wing in the state capital do not automatically become classrooms in the city neighborhoods that generated the prison's population. This is not a cynical observation — it is a structural one. The fiscal architecture of American government does not have a pipe that runs from the corrections department to the school district. Building such a pipe, if it were politically feasible, would require legislation.
The geographic pattern has a second dimension: direction. If money flows out of high-incarceration urban neighborhoods toward distant prisons, those prisons are located somewhere. They are located, overwhelmingly, in rural areas. The flow of state incarceration funding — which amounted to $63.6 billion in 2023 from state corrections budgets alone — travels from urban areas to rural counties, where prisons have become, in many cases, the primary public employer.
This creates a fiscal transfer of remarkable scale, running almost entirely beneath the level of public political debate. Urban taxpayers and the communities experiencing high incarceration rates are generating corrections revenue that is systematically invested in rural jurisdictions. Neither side of this transfer is uniformly aware of it. Neither side has politically organized around it as a fiscal equity question, though the raw numbers would seem to invite that conversation.
The original million-dollar block data is from 2006 — nearly two decades old. No comprehensive national update using modern census tract data has been published. This is itself a data gap worth noting: one of the most analytically powerful lenses on incarceration geography sits on 2006 data in a country that has changed substantially since then. We know the general pattern holds from subsequent state-level studies. We do not have a current national map.
But the structural logic is durable even where the current data is thin. Dollars spent on incarceration go somewhere. They do not go to the block. They go to the county that holds the prison.
CHAPTER FOUR: HOW PRISONS BECAME EMPLOYERS
On the moment a rural county stopped worrying and learned to love the cellblock.
The story of how prisons became economic anchors in rural America is, at its core, a story about deindustrialization arriving faster than alternatives. Through the 1960s, 1970s, and 1980s, manufacturing employment that had sustained rural and small-city economies began leaving — to suburbs, to the Sunbelt, and eventually overseas. What replaced it was often nothing, or nothing equivalent: service work that paid less, part-time schedules, no pensions, no unions.
State governments, during the same period, were building prisons. Lots of prisons. The political will to expand corrections was substantial — driven by genuine crime concerns, by political calculations, and by a federal funding architecture (particularly 1994 crime bill funds) that made expansion easier. These facilities needed locations. Rural counties, desperate for economic anchors, competed for them. A prison is, from a local economic development perspective, an attractive institution: it provides stable government employment with benefits, it doesn't outsource, it doesn't move to Mexico, and it runs 24 hours a day through recessions.
The employment numbers solidified into political facts. A prison with 300 correctional officer positions, in a county with few other large employers, becomes the county's economy. Those 300 officers vote. They have families who vote. Their union — where unions exist — has political interests in that facility's continued operation and staffing levels. The county commissioner whose predecessor lobbied hard to site that prison is not going to campaign on a platform of closing it. The state legislator from that district will not introduce a bill to reduce its population without understanding what happens to constituent employment.
This is not corruption. It is rational political behavior in response to real economic dependency. But the consequence is a constituency structure that systematically favors keeping prisons open and staffed, independent of whether those prisons are achieving any criminological purpose. The Bureau of Labor Statistics documented a national median correctional officer salary of $53,300 in 2023 — ranging from $35,040 in Mississippi to $93,160 in California. These are not minimum-wage jobs. They are career-track, benefit-laden positions. The prison, in a rural county with few other options, may be the best job a person without a college degree can get.
The structural incentive this creates is recognizable to any student of public choice economics: when a government agency's employees form a concentrated, organized interest group, and when those employees' economic well-being is directly tied to the agency's size, the agency will tend to grow or resist shrinkage independent of its functional performance. The California Correctional Peace Officers Association has spent millions on ballot initiatives and political campaigns. It is not unusual among correctional unions in this regard — it is simply larger and better documented. Correctional officer unions are among the more consistent opponents of sentencing reform, early release programs, and prison closures, for the straightforward reason that these policies reduce employment in their sector.
The dynamics compound when prison payrolls grow independent of population. The Prison Policy Initiative's 2026 expenditure report documents that correctional spending increased 27 percent between 2017 and 2025 even as the incarcerated population shrank by 15 percent over the same period. This means the per-prisoner cost of incarceration went up substantially. It means that reducing the prison population did not produce proportional reductions in corrections budgets. It means that approximately half of all correctional spending — payroll and benefits — has a kind of structural stickiness that does not respond readily to population decreases.
Part of this is mechanical: correctional facilities have security requirements that scale with the facility, not purely with its population. A 1,500-bed prison running at 900 beds still needs its perimeter staffed, its tiers patrolled, its medical unit operational. But part of it is political: the employees of that facility have organizational capacity to resist staff cuts even when the operational logic might justify them. Overtime budgets increase when facilities are understaffed. Per-prisoner costs rise. The budget line does not fall.
Prison siting also created an economic transformation in rural areas that was, from the community's perspective, genuinely beneficial, at least in the short term. Public health data from counties that gained prisons in the 1980s and 1990s shows improved employment figures, increased retail activity, and population stabilization compared to similar counties that did not. These benefits are real. They are also, in a structural sense, the reason closure or downsizing of those facilities is politically difficult in a way that the closure of an equivalent-sized factory in the same county would not be: a factory's closure might generate community pressure on state government to provide economic aid or replacement industry. A prison's closure generates community pressure on the state government not to close the prison, because the prison is state government.
The rural prison economy is, in this sense, a self-sealing system. Communities that depend on it have political incentives to sustain it. Legislators who represent those communities have electoral incentives to protect it. The prison budget — already insulated from population changes by payroll dynamics — is further insulated by its function as rural economic infrastructure. Understanding this is necessary to understanding why corrections spending grew 346 percent in real terms between 1977 and 2021, and why significant reductions remain politically difficult even in an era of reform.
CHAPTER FIVE: THE FINE AND FEE EXTRACTION MACHINE
On the discovery that the justice system could also be a revenue center.
In 2015, the Department of Justice conducted an investigation of the municipal court system in Ferguson, Missouri, following civil unrest after a police shooting. The investigation found something that legal scholars already suspected but that the general public largely did not know: the city of Ferguson had organized its court system primarily as a revenue mechanism for local government. Traffic stops, court fees, late payment penalties, and failure-to-appear charges had been layered into a system that generated substantial income for city coffers, drawn primarily from low-income residents who could least afford to pay and who faced escalating consequences — including incarceration — for nonpayment.
The Ferguson case became famous because it was investigated and documented by the federal government. The structural question it raised was whether Ferguson was an outlier or an example. The evidence suggests it was an example.
State and local governments collected a combined $13 billion in revenue from fines, fees, and forfeitures in 2021, according to U.S. Census Bureau data analyzed by the Urban Institute's Tax Policy Center. This is not the total criminal justice budget — that's $445 billion. This is the revenue side: the money that flows into government from the operation of the justice system, rather than the money appropriated to fund it.
This $13 billion comes from several sources. Court filing fees — charged not just to defendants but sometimes to all parties in court proceedings. Supervision fees — monthly payments required from people on probation or parole, whose supervision by the state they are required to fund themselves. Public defender fees — charged to indigent defendants who received a constitutionally mandated attorney they cannot afford. Drug testing fees — paid by people on probation who must be tested regularly. Electronic monitoring fees — paid by people on home confinement for the ankle bracelet the state requires them to wear.
There is also civil asset forfeiture, which operates on a different and considerably more contested legal basis. Forfeiture allows law enforcement agencies to seize property — cash, cars, equipment, bank accounts — that they allege is connected to criminal activity. "Allege" is the operative word: civil asset forfeiture does not require a criminal conviction. It does not, in many jurisdictions, even require criminal charges. The property itself is, legally, the defendant. The standard for seizure is lower than the standard for conviction. And in 32 states, law enforcement agencies can retain 80 to 100 percent of the proceeds directly.
The Institute for Justice's "Policing for Profit" report — now in its third edition, covering data from 17 million records across 45 states plus D.C. — documented that since 2000, state and federal governments together have forfeited at least $68.8 billion in property. The researchers acknowledge this is a significant undercount because not all states provided full data. At the federal level alone, the Treasury Forfeiture Fund reported $1.619 billion in FY 2023 revenue; the DOJ Assets Forfeiture Fund generates approximately $2 billion annually.
Half of all currency forfeitures, the Institute for Justice found, are worth less than $1,300. This is not the profile of drug cartel kingpin seizures. This is the profile of cash in a car, spending money in a wallet, a bank account with a month's rent in it. The research finds that forfeiture rates increase when local economies contract — suggesting that financial pressure on law enforcement budgets influences enforcement decisions in ways that are structurally predicted and empirically observed.
The 1984 Comprehensive Crime Control Act's equitable sharing provisions created an additional mechanism: local law enforcement agencies can turn over seized property to federal agencies, receive a share of the proceeds, and thereby circumvent state-level limits on forfeiture that their own legislatures have enacted. An NBER working paper documented that this arrangement changed police incentive structures, with agencies reallocating enforcement effort toward drug crimes when forfeiture proceeds supplemented budgets. The equitable sharing program, in other words, is partly a system by which federal policy can override state-level democratic decisions about how law enforcement should be funded and incentivized.
The aggregate consequence of all this revenue generation is an incentive structure that public administrators might describe as "misaligned." When a court system's operating budget depends on fines, judges face implicit pressure to maintain fine revenue. When a police department keeps forfeiture proceeds, officers face implicit pressure to make seizures. When a county's probation department charges supervision fees, it has an interest in keeping caseloads high. None of these pressures need to be explicit, consciously applied, or corrupt in any conventional sense. They are simply the predictable behavioral consequences of tying institutional revenue to institutional outputs.
As of 2021, at least $27.6 billion in fines and fees is owed across the country — accumulated debt that people with criminal records are legally required to pay. This debt follows people out of incarceration. It affects their credit, their licensing eligibility, and in many states their voting rights. It generates additional legal exposure if unpaid — because failure to pay can constitute a probation violation, which can result in reincarceration, which generates new incarceration costs, which are partly offset by the fine and fee revenue that the reincarcerated person is now, again, required to generate.
The system has, in places, a circular quality that appears to sustain itself.
CHAPTER SIX: THE PLEA BARGAIN MACHINE
On how a constitutional right became a statistical anomaly.
The Sixth Amendment to the United States Constitution guarantees the accused "the right to a speedy and public trial, by an impartial jury." This right is real. It is also, in practice, exercised by approximately 2 to 3 percent of people charged with federal crimes.
In 2023, approximately 97 to 98 percent of federal criminal convictions resulted from guilty pleas, according to the American Bar Association's Plea Bargain Task Force report. The figure for state courts is somewhat lower but runs between 90 and 97 percent. The federal rate has climbed steadily: 84 percent in 1984, 94 percent by 2001, approaching 98 percent today.
The United States Supreme Court, in Missouri v. Frye (2012), offered a frank assessment: plea bargaining is "not some adjunct to the criminal justice system; it is the criminal justice system."
What happened to jury trials? Several things, in sequence and combination.
First, sentencing. The federal sentencing guidelines, established in the 1980s, created mandatory sentencing ranges that reduced judicial discretion and dramatically increased the potential consequences of going to trial. State legislatures, through the 1980s and 1990s, added mandatory minimums — statutory floors for particular offenses that removed the possibility of a judge imposing a lighter sentence even when circumstances warranted it. The result was an enormous gap between the sentence available through a negotiated plea and the sentence that would follow a trial conviction. Prosecutors call this "the trial penalty." Defense attorneys call it the same thing, with different emphasis.
Second, resource asymmetry. Federal prosecutors have resources — investigators, expert witnesses, discovery capability, time — that most defense attorneys, particularly public defenders, cannot match. Public defenders in many jurisdictions carry caseloads that the American Bar Association considers incompatible with effective representation. In this environment, advising a client to fight charges through trial is sometimes a counsel of hope against odds, rather than a counsel of strategy.
Third, pretrial detention. The Vera Institute documented in its 2022 report that pretrial detention — being held in jail before trial because bail cannot be met — increases the likelihood of pleading guilty by approximately 46 percentage points. This is a striking effect size. Incarcerated people awaiting trial lose their jobs, their housing, their family stability. They are physically present in a jail. The calculus of "take the plea and go home today with probation" versus "fight the charges, remain in jail for months, and risk five years" is not a balanced calculation, especially when the home situation — dependent family members, pending rent, a job that will not be held — is deteriorating outside.
The structural result is a conviction assembly line. The ABA Task Force in 2023 described the current plea system as one in which "innocent defendants are being induced to plead guilty." A 2024 study in the American Political Science Review found that under a range of modeled scenarios, innocent defendants are more likely to enter guilty pleas than guilty defendants — particularly when facing risk-averse accused people who prefer certain light punishment to uncertain potentially heavy punishment.
The incentive structure on the prosecution side is also worth examining. In many jurisdictions, prosecutors are elected, and conviction rates are trackable, public, and politically salient. A prosecutor who goes to trial frequently and loses frequently has a problem. A prosecutor who manages their caseload through pleas maintains a high conviction rate, processes cases efficiently, and can credibly claim to be running an effective office. The system rewards efficiency. It rewards high conviction rates. It does not, structurally, reward going to trial to defend a principle.
There is a quality-control function in jury trials that disappears when 97 percent of convictions happen outside of them. Juries introduce friction into the system — an opportunity for a community to evaluate the government's case, to exercise judgment about facts and proportionality, to acquit when the charge seems disproportionate to the conduct. When that friction is nearly eliminated, the quality-control mechanism disappears. The government's case no longer needs to persuade twelve citizens. It needs to persuade one person under conditions of informational asymmetry and time pressure that fighting would probably make things worse.
This is not a moral indictment of prosecutors or defense attorneys or judges. It is an account of a system that has evolved to produce one output — pleas — with an efficiency that the designers of the constitutional jury trial system would find remarkable, and possibly concerning.
CHAPTER SEVEN: ARIZONA — WHEN THE SYSTEM BREAKS
On fourteen years, $2.5 million in fines, and the moment a federal judge ran out of options.
On February 20, 2026, U.S. District Judge Roslyn Silver signed a 128-page order placing the healthcare operations of all ten Arizona state-run prisons under federal receivership. The order ended — or perhaps more accurately, escalated — fourteen years of litigation over whether the Arizona Department of Corrections, Rehabilitation, and Reentry was providing constitutionally adequate medical care to the approximately 34,000 people in its facilities.
The word "receivership" is doing significant work in that sentence. A federal receivership is not a strongly worded letter. It is not a consent decree. It is not a finding of violation with a compliance timeline. It is an order stripping a state agency of operational control over a function that the court has determined — after exhausting every lesser remedy — the agency is incapable of performing constitutionally. The receiver, once appointed, will have authority to hire and fire healthcare staff, renegotiate or terminate ADCRR's existing $300 million contract with its private healthcare vendor NaphCare, set salaries, reconfigure facilities, and override ADCRR administrators. The receiver answers to the court, not to the state.
Judge Silver's language in the order was specific and measured in a way that made it more damning, not less. She wrote that ADCRR had "unreasonably" misread her directives, had gone to "great lengths to exploit any ambiguity to the maximum extent possible," and that the standard remedies had produced fourteen years of failure. The key passage merits quotation at length:
"After nearly 14 years of litigation with defendants having not gained compliance, or even a semblance of compliance with the injunction and the Constitution, this approach has not only failed completely but, if continued, would be nothing short of judicial indulgence of deeply entrenched unconstitutional conduct."
And:
"Plainly, only the imposition of the extraordinary can bring an end to this litigation and the reasons it was brought. An end to unconstitutional preventable suicides. An end to unconstitutional preventable deaths. An end to unconstitutional failures to treat those in severe pain. The Motion for a Receiver will be granted."
Let us trace the timeline to understand how a court reaches that language.
The original lawsuit was filed in 2012. The plaintiffs were incarcerated individuals alleging that ADCRR's medical care violated the Eighth Amendment's prohibition on cruel and unusual punishment — specifically, the "deliberate indifference to serious medical needs" standard established by the Supreme Court in Estelle v. Gamble in 1976. The legal standard is intentionally high; "deliberate indifference" requires more than negligence. It requires a finding that officials knew of and disregarded a substantial risk to health.
In 2014, rather than proceed to trial, the parties settled. The settlement included 103 specific performance standards — a detailed blueprint for what constitutionally adequate care looked like in practice. ADCRR agreed to meet those standards. ADCRR did not meet those standards.
The court's response to noncompliance moved through the available remedies in sequence. First, monitoring. The court appointed independent monitors to assess compliance. The monitors reported ongoing failures. Then, findings of contempt. Then, fines — $2.5 million accumulated, an extraordinary sum by the standards of institutional civil rights enforcement. Then, a 15-day bench trial in 2022 — a rare evidentiary hearing that produced a 200-page findings-of-fact order. The court's 2022 conclusion: ADCRR's healthcare system was "pervasively and systemically unconstitutional." Not inadequate. Not failing. Pervasively and systemically unconstitutional.
In 2023, the court issued a permanent injunction with 154 specific quality indicators. The monitoring continued. The failures continued. Communication between ADCRR and the court's monitors — the people tasked with measuring compliance — broke down. ADCRR, the court found, had rejected and ignored monitor recommendations. The 2022 order had already used the phrase "merry-go-round of for-profit correctional health care vendors" to describe ADCRR's history with private healthcare contractors — a sequence of companies, contracts, and failures spanning the litigation period.
In February 2026, Judge Silver determined that no standard remedy remained unexhausted. The receivership was the last tool available that did not simply accept the unconstitutional status quo as permanent.
ADCRR Director Ryan Thornell responded with a statement that described the ruling as likely to be "exorbitantly expensive" and disruptive of "the significant progress we have made." Governor Katie Hobbs expressed being "deeply disappointed," citing "the immense strides" ADCRR had made in complying with prior orders. ADCRR announced its intent to appeal "aggressively."
These responses deserve examination alongside the documented record. The court did not find significant progress. The court found no semblance of compliance. The $2.5 million in fines, the 15-day trial, the 200-page findings of fact, the 14 years of monitoring, and the 154-indicator injunction were not a court that lost patience prematurely. They represent the full sequence of intermediate steps, applied in order, with outcomes that the court documented in detail.
The parallel most frequently cited in coverage of the Arizona ruling is California. In 2005, a federal court placed California's prison medical system under receivership after finding that one prisoner per week was dying of medical neglect or malpractice. As of 2026, that receivership remains in place — approximately twenty years later, with no specified end date. This is the precedent Arizona's lawyers cited to characterize the measure as extraordinarily expensive. It is also, inadvertently, a characterization of what intractable unconstitutional conditions look like across time.
The structural questions the Arizona case raises are not limited to Arizona. Across the country, state prison healthcare systems are delivered primarily through private contractors — NaphCare, Centurion, YesCare (formerly Corizon), Wellpath — operating under state contracts where the state retains Eighth Amendment liability but has outsourced operational control. The incentive structure of this arrangement deserves examination: the private contractor has financial incentives to minimize costs; the state has political incentives to minimize visible spending; the incarcerated patient has constitutional rights but no market power. The court is the only check on this arrangement, and as the Arizona case documents, the court's checks are slow, expensive, and — after fourteen years — culminated in the conclusion that the only remedy was to remove the agency from the equation entirely.
CHAPTER EIGHT: HEALTHCARE COLLAPSE AND THE RECEIVERSHIP PRINCIPLE
On the Eighth Amendment, the private healthcare contract, and the discovery that some things cannot be delegated.
The legal foundation of prison healthcare rights in the United States is a 1976 Supreme Court case, Estelle v. Gamble. The holding was simple: the government, having deprived a person of liberty and thereby of the ability to obtain their own medical care, assumes a constitutional obligation to provide that care. Deliberate indifference to serious medical needs constitutes cruel and unusual punishment under the Eighth Amendment. The case was decided unanimously. The principle has not been challenged.
What has been contested, continuously and expensively, is implementation.
The gap between the constitutional principle ("the government must provide medical care") and institutional practice has proven, in many states, to be vast. The mechanism that most states have adopted to bridge this gap is the private healthcare contract. Rather than building internal medical capacity — hiring physicians, nurses, and specialists as state employees — state corrections departments contract with private companies to deliver care. This is a logical delegation: running a healthcare system is technically complex, and corrections departments were not traditionally structured to do it. The private companies bring existing infrastructure.
The problem is structural. Private healthcare contracts create a vendor whose financial interest is in minimizing the cost of services delivered, operating within a contract that the state negotiated under conditions that are not ideal for cost transparency. The state, as the contracting authority, must monitor compliance — but monitoring clinical quality requires either independent medical expertise or trust in the contractor's self-reporting. The contractor has incentives to report favorably on its own performance. The state has political incentives to declare the situation managed. The incarcerated patient has constitutional rights but no ability to switch providers, no ability to file a complaint with a regulatory body in any meaningful timeframe, and no market power whatsoever.
When the gap between contracted obligation and delivered care becomes visible — through a lawsuit, a court monitor's report, a series of preventable deaths — the remedy process begins. And the remedy process, as the Arizona case documents, is slow in a way that is structurally predictable. Courts are not healthcare administrators. They cannot directly supervise clinical operations. They can issue orders, assess compliance, impose fines, and — as a last resort — appoint a receiver. But each of these steps requires time and generates litigation. The constitutional violation can persist for years while the remedial machinery works through its sequence.
The receivership model attempts to solve this by replacing the responsible agency with a court-supervised administrator who has operational authority. The California experience — approximately twenty years of receivership in the prison medical system — suggests that this is not a quick fix. It is an acknowledgment that the institutional configuration that produced the violation cannot, by itself, produce the remedy.
The Arizona receivership covers approximately 34,000 people in state-run prisons. It explicitly excludes approximately 10,000 people incarcerated in private prisons under state jurisdiction — a gap worth noting. The constitutional obligation does not change based on the contractor operating the facility. But the procedural posture of the litigation was organized around ADCRR's own facilities, and the receivership authority follows that structure.
The $300 million NaphCare contract — which the receiver will have authority to renegotiate or terminate — illustrates the scale of what states spend on private correctional healthcare, and the power that spending confers. A $300 million contract is a significant commercial relationship. It creates vendor dependency: switching contractors requires knowledge transfer, staff transition, system migration. The "merry-go-round of for-profit correctional health care vendors" that Judge Silver's 2022 order described is a consequence of this dependency meeting repeated failure — each new vendor offering the promise of a fresh start, each new start eventually cycling back through the same structural dynamics.
None of the private healthcare vendors operating in this space are charities. They are companies with shareholders or private equity backers, operating in a market where their customers — state corrections departments — face political pressure to control costs and limited ability to monitor care quality. The vendors know this. The state knows this. What neither party has fully solved for is the constitutional obligation that sits at the center of the arrangement, indifferent to contract terms.
The Arizona case is a particularly well-documented instance of a nationwide structural challenge. It is not unique. The frequency with which state prison healthcare systems appear in federal civil rights litigation — as plaintiffs, defendants, subjects of consent decrees, parties to contempt proceedings — suggests that the private contract model, as currently structured, has systematic compliance problems. Understanding why requires no assumption of malice. It requires only attention to what the incentive structure actually rewards.
CHAPTER NINE: IF WE DESIGNED IT THIS WAY ON PURPOSE
On the gap between declared purpose and observable architecture.
Let us conduct a thought experiment. Suppose you were an institutional designer, and someone commissioned you to build a system with the following operational characteristics:
It should generate revenue for local governments from the people it processes. It should create concentrated economic benefits in rural areas, sufficient to generate political constituencies that resist the system's reduction. It should process the vast majority of cases through negotiated outcomes rather than adversarial review, minimizing the friction of independent fact-finding. It should carry the minimum constitutionally required healthcare obligation while delegating delivery to parties with financial incentives to minimize costs. It should expand in periods of public fear, contract slowly in periods of reform, and maintain its core employment base and revenue streams through both cycles. It should generate legal exposure for those who pass through it that outlasts the initial sentence, creating ongoing fiscal claims and ongoing legal vulnerability. And it should be organized in a way that makes any individual actor within it identifiable as responding reasonably to the incentives they face, so that systemic responsibility is difficult to locate.
You would have designed, more or less, what exists.
This is not a conspiracy claim. No evidence suggests that anyone sat in a room in 1975 and sketched the outline above as an intentional plan. The system that emerged was built by thousands of individual decisions — legislative votes, budget line items, contract negotiations, court settlements, zoning decisions, union agreements, prosecutorial guidelines — each responding to local conditions, political pressures, and available incentives. The overall architecture emerged from these decisions the way a river valley is shaped by water: the water is not planning; it is responding. But the valley is real, and its shape has consequences.
What is worth examining is not intent but function. What does the existing system demonstrably do?
It incarcerates approximately 1.9 million people at any given time, at a total cost of approximately $445 billion per year.
Of the $87 billion spent on state and local corrections in 2021, roughly half — approximately $43.5 billion — went to staff compensation. This money circulates in the local economies of rural counties that host correctional facilities. It is stable, recession-resistant, government-sourced income.
The system generates approximately $13 billion per year in fine and fee revenue for state and local governments. It has generated at least $68.8 billion in civil asset forfeitures since 2000.
It resolves 97 to 98 percent of federal criminal cases through guilty pleas, producing a high-throughput conviction process that rarely requires the evidentiary standard of a jury trial.
It maintains probation and parole supervision over 3.77 million people, generating supervision fee revenue from those people and maintaining a population with ongoing legal exposure — a violation of probation conditions can result in reincarceration without a new criminal conviction.
It creates a legal record that affects employment, housing, and civic participation for tens of millions of Americans, generating demand for background check services, expungement attorneys, and a growing industry of collateral consequence navigation.
It exports public dollars from high-incarceration urban neighborhoods to rural counties, representing a substantial and largely invisible geographic transfer of public investment.
These are outputs. They are documented. They emerge from an institutional structure, and they sustain constituencies that have interests in the continuation of that structure. Whether these outputs constitute a "design" depends on your definition of design. They constitute, at minimum, a stable equilibrium — a configuration of incentives and interests that tends to reproduce itself.
The reform question, under this analysis, is not primarily moral. It is institutional. What would need to change — which budgets, which contracts, which sentencing laws, which revenue streams, which employment bases — to shift the equilibrium? And which constituencies would oppose each change, and why? The structural answer to those questions is considerably more sobering than the moral answer, because the moral answer can change relatively quickly when public opinion shifts. The structural answer requires reorganizing interests that have had decades to solidify.
CHAPTER TEN: THE INTERNATIONAL CONTRAST
On why "other countries do it differently" is more complicated than it sounds, and less complicated than its critics claim.
The standard response to international incarceration comparisons is: it's not comparable. The United States is different — different culture, different history, different crime rates, different demographics, different legal traditions. The comparison isn't useful.
This response contains some truth and a great deal of evasion. Let us try to be precise about each.
The numbers first. The U.S. incarceration rate, at approximately 541 per 100,000, is roughly seven times the rates of Germany (76), Norway (75), and the Netherlands (69). Finland, which actively reformed its incarceration system in the second half of the twentieth century, reduced its rate from approximately 200 per 100,000 in the 1950s to approximately 50 to 60 per 100,000 by the 2000s — a reduction achieved through deliberate policy choice, not passive social drift.
Norway's recidivism rate — the share of released people who return to prison — is approximately 20 percent within five years, down from approximately 70 percent in the 1990s. The U.S. rate, depending on how recidivism is defined (re-arrest, reconviction, or reincarceration) and over what follow-up period, ranges from approximately 40 to 70 percent. These figures are not comparable on an apples-to-apples basis: the countries use different definitions, different follow-up periods, and different base populations. But the directional difference — Norway 20 percent, United States 40 to 70 percent — is large enough that methodological adjustments are unlikely to eliminate it.
Norway spends approximately $90,000 to $128,000 per prisoner per year. The U.S. national average is approximately $35,000 to $45,000, with significant state variation (Massachusetts at roughly $285,000; Mississippi at roughly $20,000). Norway spends more per prisoner and produces dramatically lower recidivism. This is a consequential data point. It does not, by itself, establish causation — Norway differs from the U.S. in many ways, and attributing its low recidivism entirely to prison philosophy involves methodological leaps. But the outcome difference is real, the cost-per-prisoner difference is real, and the policy philosophy difference is real and documented.
Norway's stated penal philosophy is the "principle of normality": with the exception of freedom of movement, prisoners retain all other rights, and life in prison should approximate life outside to the greatest extent possible. The goal of the Norwegian system, stated explicitly by its Ministry of Justice, is successful reintegration. Facilities provide education, vocational training, therapy, and family contact. Halden Prison — frequently cited as an example — looks nothing like a typical U.S. correctional facility. It looks, to American eyes, like a peculiar college campus. Critics find this absurd or offensive. Defenders find it efficient: if the goal is to produce people who do not reoffend, building environments that support functioning is more cost-effective than building environments that reinforce dysfunction, even at higher per-day cost.
The Vera Institute delegation that visited Germany and the Netherlands in 2013 documented that both countries rely heavily on fines, community penalties, and short sentences rather than incarceration for most offenses. German prison sentences are typically shorter than equivalent U.S. sentences; German correctional staff receive what amounts to university-level training; the system is organized around rehabilitation as a primary goal.
The "it's not comparable" objection has several legitimate components. First, the United States has significantly higher rates of violent crime than most Western European countries — higher murder rates, more firearms, different patterns of interpersonal violence. A direct comparison of incarceration rates without accounting for crime rates can be misleading. Second, the welfare state context differs substantially: Norway and Germany provide more extensive social safety nets that reduce the economic desperation that drives some crime. Third, historical and cultural context differs in ways that affect how communities respond to crime, how political accountability works, and what voters demand from criminal justice systems.
These are real differences. They partially — not entirely — explain the gap. The problem with using them as a full explanation is that they don't explain the rate of change. The U.S. incarceration rate in 1970 was much lower than it is today, even though violent crime rates were higher in the early 1990s than they are now. The crime rate cannot explain a sevenfold difference in incarceration rates between the U.S. and Europe, because the crime rate difference is not sevenfold. Something else explains the gap, and the most plausible candidates are the structural factors examined in this work: sentencing philosophy, plea bargaining dynamics, political economy of prosecution and correction, and the fiscal incentives created by the fine-and-fee system.
Finland's deliberate reduction from 200 to 50–60 per 100,000 is the most analytically useful international comparison precisely because it involves the same country at two different time points. Finland did not change its demographics, geography, or culture. It changed its policies. If a country can reduce its own incarceration rate by 70 to 75 percent through deliberate policy reform over several decades, the claim that structural factors make such reductions impossible becomes harder to sustain.
The comparison is not a policy prescription. It is evidence that the current U.S. configuration is not inevitable.
CHAPTER ELEVEN: COUNTERARGUMENTS
On the strongest cases for what exists.
Any honest account of the U.S. incarceration system must grapple with the strongest arguments in its defense. These arguments are not trivial. They deserve presentation on their own terms, not as straw men to be knocked down.
The public safety argument. The dramatic increase in incarceration from the mid-1970s through the early 2000s coincided with what is now the largest sustained decline in violent crime in American history. From its peak in the early 1990s, the U.S. murder rate fell by more than half. Property crime rates fell even more. These declines occurred while incarceration was high. When incarceration declined after 2008, some cities experienced subsequent increases in violent crime — a correlation that many observers found concerning.
Econometric research has attempted to isolate the incapacitation effect of incarceration — the reduction in crime achieved simply by keeping offenders in facilities where they cannot commit crimes against the public. Studies by Steven Levitt and others in the 1990s and 2000s estimated that incarceration explains somewhere between 5 and 35 percent of the crime decline from the early 1990s peak. This is a contested estimate range — the methodology is difficult and the studies have been critiqued — but the core claim has not been refuted: keeping people who would otherwise commit crimes incarcerated does, in fact, reduce crime. The question is how much, at what cost, and whether alternatives could achieve similar outcomes more efficiently.
The steelman version of the public safety argument: During the crime wave of the 1970s through early 1990s, the communities most affected by violent crime demanded effective law enforcement responses. Incarceration — whatever its costs and distortions — provided an answer. The question of what the alternative would have been is not rhetorical. If incarceration had not been expanded, some additional number of violent crimes that did not occur because offenders were incapacitated would have occurred. The victims of those crimes are real people whose experiences are not captured in structural analyses of fiscal incentives.
The violent crime objection. The popular description of mass incarceration as primarily a product of non-violent drug offenses is inaccurate for state prisons, where approximately 55 to 57 percent of the population is incarcerated for violent offenses. The narrative that the U.S. prison system is full of people who smoked marijuana is simply wrong. Most people in state prison are there because a court found that they committed a violent crime. This does not resolve questions about sentence length, conditions of confinement, or recidivism outcomes — but it does require a more accurate accounting of who is actually incarcerated.
This matters for policy: a reform agenda focused primarily on drug offenders, though politically easier, would not substantially reduce the state prison population. The harder and less politically tractable question involves violent offenders — their sentences, their prospects for rehabilitation, and the consequences for public safety if their sentences are reduced.
The democratic accountability argument. In many U.S. jurisdictions, prosecutors and judges are elected. Legislators who pass mandatory minimum sentencing laws are accountable to voters. The expansion of incarceration was not imposed on an unwilling public — it was, to a substantial degree, demanded by a public experiencing high crime rates and terrified by the homicide statistics of the early 1990s. Voters who supported "tough on crime" candidates were responding to genuine experiences of fear, harm, and neighborhood deterioration. Dismissing those preferences as manipulation or false consciousness requires a level of confidence about public psychology that the evidence does not support.
The political accountability argument also runs the other way: as crime rates fell and as information about incarceration costs and conditions became more widely available, public preferences have shifted. Bipartisan criminal justice reform coalitions — the "Right on Crime" movement on the conservative side, traditional reform advocates on the liberal side — have generated real policy changes in some states. Texas reduced its prison population substantially through conservative fiscal arguments: mass incarceration was expensive and produced high recidivism, meaning it was failing on its own terms. The democratic system produced reform in those states. It can do so elsewhere.
The budget constraint argument. Alternatives to incarceration are not free. Drug courts, mental health courts, expanded probation, community supervision, housing support, vocational training, reentry programs — all of these cost money. The infrastructure required to deliver these alternatives at scale, across diverse jurisdictions with varying levels of government capacity, is substantial. Reform advocates sometimes underestimate these costs; critics of reform sometimes overestimate them. The honest accounting suggests that alternatives to incarceration can be cost-effective on a per-person basis, while acknowledging that the transition costs — closing facilities, retraining staff, building new infrastructure — are real and politically difficult.
The reform failure cases. Several jurisdictions that implemented significant reforms subsequently experienced crime increases — though causation is contested. California's 2011 criminal realignment, which shifted low-level offenders from state prisons to county jails, produced mixed outcomes including crowding in some county facilities. Bail reform in New York, implemented in 2020, generated significant political backlash and subsequent legislative modification. The "defund" framing that emerged in 2020, whatever its policy content, appears to have generated public reaction that set back reform efforts in several cities. These are not invented problems. They are evidence that public preferences about public safety are real, that crime rates are politically consequential, and that reform efforts that do not adequately address violent crime may face democratic rejection.
The strongest version of the counterargument: the existing system, with all its structural distortions, is the outcome of a democratic process responding to real crime conditions. Changing it requires persuading a public that still, in the most recent polling, prioritizes public safety, that alternative approaches will deliver acceptable safety outcomes. The record of alternative approaches is mixed, the evidence for their effectiveness is context-dependent, and the political risks of getting it wrong are borne by whoever runs on a platform of change. In this environment, what might look like systemic inertia is partly the outcome of genuine democratic uncertainty about what works.
CHAPTER TWELVE: THE CIVIL AFTERLIFE
On what happens after the sentence ends.
The sentence ends. The door opens. The person walks out.
This is where the standard account usually stops. It is also where the machinery continues.
The American Bar Association has catalogued approximately 44,000 legal consequences that can attach to a criminal conviction in the United States. These are not punishments imposed by a sentencing court. They are statutory collateral consequences — automatic legal disabilities that follow the conviction regardless of what the judge said, regardless of whether the sentence was served, and in many cases regardless of how many years have passed.
They include:
Employment restrictions. Many professional licenses are restricted to people without criminal records — including licenses for nursing, teaching, law, accounting, cosmetology, plumbing, and real estate. The specific restrictions vary by state and by profession, but the aggregate effect is substantial. People released from prison into an economy where legitimate employment is a primary route to stability find that their options are legally constrained in ways that are often not explained at the time of conviction.
Housing restrictions. Public housing authorities in many jurisdictions exclude applicants with certain criminal histories. Private landlords routinely conduct background checks and decline applicants with criminal records. The Fair Chance Housing movement has achieved some legislative successes in restricting when and how landlords can use criminal history, but these policies are not universal.
Public benefits. Federal law restricts access to certain public benefits for people with drug felony convictions. The specific restrictions have been modified by subsequent legislation, but in some states, a drug conviction can affect eligibility for Supplemental Nutrition Assistance Program benefits, public housing, and student loans.
Voting rights. As of 2025, approximately 4.6 million Americans were disenfranchised due to a felony conviction, according to the Sentencing Project. [NOTE: This specific figure was not in the research dossier and requires independent verification.] State laws on this vary enormously — from states that restore voting rights immediately upon release, to states that require completion of all supervision, to Maine and Vermont, which allow people to vote even while incarcerated. The patchwork creates significant geographic variation in civic participation rights based on where a conviction occurred.
Debt. As documented in Chapter Five, people leaving incarceration frequently carry fine and fee obligations accumulated during their criminal case and incarceration. This debt is not forgiven upon release. It accrues interest in some jurisdictions. It can affect credit, licensing eligibility, and in some states, voting rights. Failure to pay can constitute a probation violation, which can result in reincarceration — a loop that is documented in the literature and costly to all parties.
The aggregate of these consequences creates what some legal scholars call "the civil death" — a legal status that is not imprisonment but that severely restricts the economic and civic participation of people with criminal records. The function of this legal architecture is not entirely clear. Some restrictions have plausible public safety rationales — restricting someone convicted of financial fraud from working as a licensed financial advisor, for instance. Many do not have obvious public safety rationales and appear instead to reflect a policy preference for permanent punishment rather than bounded punishment.
The practical consequence is a permanent second-class legal status for roughly 73.5 million Americans with some form of criminal record — a number that, as Chapter Two established, represents approximately one in three adults. The functional meaning of "having a criminal record" varies enormously across this population: the person arrested once in their twenties and the person serving their third felony sentence have very different experiences of what their record means. But the structural point is that criminal record status, in the United States, functions as an ongoing legal category affecting employment, housing, civic participation, and economic opportunity — in many cases for life.
This creates a feedback dynamic that matters for understanding incarceration levels. If post-conviction legal disabilities make stable employment and housing more difficult, they increase the probability of reoffending — which increases the probability of reincarceration — which deepens the criminal record — which further restricts employment and housing options. The system does not merely punish crime; it creates conditions that increase the probability of more crime, which sustains the demand for more punishment.
Whether this feedback is an intentional feature or an emergent consequence of thousands of separately enacted laws is a question the evidence cannot definitively answer. That it operates is documented.
CHAPTER THIRTEEN: INCENTIVES VERSUS INTENTIONS
On the difference between bad people and bad systems, and why the distinction matters more than it sounds.
One of the most common misunderstandings about structural analysis is that it requires the assumption that someone, somewhere, planned it this way with malicious intent. This misunderstanding cuts both ways. Critics of reform argue that structural accounts are thinly veiled accusations of conspiracy. Some reform advocates, to be fair, do drift toward conspiratorial framing. But the most analytically powerful structural accounts require neither conspiracy nor villainy.
They require only incentives.
Public choice theory — associated most prominently with economists James Buchanan and Gordon Tullock — offers a framework for understanding how government agencies, like private actors, respond to the incentives they face. The core insight is simple: government employees are people, and people respond to incentives. When those incentives push toward a particular behavior, that behavior tends to occur, even when the nominal purpose of the institution points elsewhere.
Apply this to American corrections:
A state corrections department is funded by appropriations. Its budget is largely determined by headcount — the number of people it houses, which drives staffing requirements, which drives payroll, which drives the budget request. A department that reduces its population reduces its budget. The institutional interest of the agency — not of any individual malicious administrator, but of the agency as an organizational entity — is to maintain population. This does not require anyone to want more crime. It simply creates a budget structure in which population is the relevant metric.
A county that has organized its rural economy around prison employment has a structural interest in maintaining that employment. The county commissioner does not need to be indifferent to justice to oppose prison closures. They need only be responsive to the economic circumstances of their constituents.
A prosecutor who is evaluated — formally or informally — on conviction rates has a structural interest in avoiding losses. Plea bargaining produces convictions reliably. Jury trials do not. The structural response is a high plea rate, independent of whether any individual prosecutor is cutting corners or acting in bad faith.
A law enforcement agency that retains civil asset forfeiture proceeds has a budget that partially depends on forfeiture activity. Research finds that forfeiture rates increase when local economies contract. No individual officer needs to consciously decide to shake down motorists for budget reasons. The institutional incentive exists, and behavior at the margin — which stops to prioritize, which vehicles to search — responds to it.
This is the distinction between incentives and intentions. Bad outcomes do not require bad people. They require institutional structures that align individual rational behavior with outcomes that are collectively harmful or constitutionally troubling. The policy implication is significant: if the problem were primarily bad people, the solution would be to find better people. But institutions tend to shape the people within them rather than the reverse — a finding that is robust across organizational psychology, sociology, and economics. If the problem is bad incentives, the solution is to change the incentive structure. That is harder than finding better people, but it is more durable.
This framework also helps explain why reform efforts that focus primarily on changing personnel — electing different prosecutors, hiring different wardens, appointing different commissioners — often produce less change than their proponents expect. The individuals change; the institutional incentives remain. The new prosecutors face the same caseload pressures, the same political accountability metrics, the same plea bargaining dynamics. They respond to those incentives in ways that resemble their predecessors, because the incentives are the stable element.
The Arizona case illustrates this at scale. Fourteen years of litigation produced multiple changes in ADCRR leadership, multiple changes in healthcare contractors, multiple compliance plans, and multiple findings of failure. The court's conclusion — that only removing the agency from operational control could produce a remedy — reflects a judgment that the institutional configuration, not the individuals within it, was the durable problem.
Understanding this requires neither cynicism about public servants nor naivety about institutional dynamics. Most people who work in corrections, prosecution, and law enforcement are doing a job they believe in, following procedures their institutions have established, responding to the incentives their institutional context creates. The structural critique is not of them. It is of the configuration.
CHAPTER FOURTEEN: STRUCTURAL PROBABILITY
On why geography and economics move incarceration risk.
The question of why some people are incarcerated and others are not is, at the individual level, partially a matter of choices and events specific to that person's life. At the population level, it is substantially a matter of structural conditions.
Researchers at Harvard's Opportunity Atlas project, led by Raj Chetty and Nathaniel Hendren, found that the zip code where a child grows up is predictive of adult incarceration probability — independent of individual characteristics. This is not a finding about personal morality or family structure. It is a finding about geography: the same person, born into the same family circumstances but in a different zip code, has measurably different incarceration risk.
The structural risk factors the research literature identifies — working from the research dossier — include poverty, deindustrialization, educational inequality, and the social mechanisms described by sociologists Robert Sampson and W. Byron Groves as "social disorganization."
Poverty correlates with incarceration probability through multiple mechanisms: it restricts access to private legal counsel; it makes bail unaffordable; it concentrates people in neighborhoods with higher policing density; and — through the cognitive bandwidth mechanism described by Mullainathan and Shafir — it may reduce the cognitive resources available for the deliberate, forward-looking decision-making that reduces risk-taking. The Mullainathan-Shafir scarcity theory has faced replication challenges in its specific mechanistic claims, but the behavioral associations between poverty and high-risk decision-making are more robustly supported.
Deindustrialization removed stable employment from communities that had organized economic and social life around manufacturing work. The consequences — concentrated unemployment, reduced family stability, weakened community institutions — are documented in the research literature as correlating with increased crime and, subsequently, increased incarceration. This is not an abstract historical observation. The communities most affected by deindustrialization in the 1960s through 1980s included the same communities that experienced the crime wave that drove political demand for incarceration expansion. The opioid epidemic, from the 1990s to the present, maps similarly onto communities that lost manufacturing employment — generating crime, incarceration, and human suffering in places where economic disruption preceded the crisis.
Educational inequality affects incarceration probability through the labor market — people without educational credentials have fewer legitimate economic options — and through direct interaction with institutions. Schools in lower-income districts have, in many documented cases, more frequent contact between students and the criminal justice system, through resource officers and disciplinary referrals. The "school-to-prison pipeline" — a term that has been both used analytically and overused as a slogan — describes a documented pattern in which school disciplinary policies, particularly in resource-constrained districts, channel students toward criminal justice involvement.
Social disorganization theory, as developed by Sampson and Groves in their landmark 1989 study and extended through Sampson's Project on Human Development in Chicago Neighborhoods, identifies neighborhood-level structural conditions — poverty, residential instability, family disruption — that weaken informal social control mechanisms. Communities with strong networks of mutual surveillance, reciprocal trust, and willingness to intervene in disorder — what Sampson calls "collective efficacy" — have lower crime rates even when socioeconomic conditions are similar. Communities where these networks have been disrupted — by concentrated poverty, by high turnover, by incarceration itself (which removes adults from communities and disrupts households) — have higher crime rates and higher subsequent incarceration rates.
The feedback loop here is documented: high incarceration in a neighborhood reduces collective efficacy by removing adults, disrupting families, and depleting economic resources, which increases crime risk, which increases incarceration, which further reduces collective efficacy. Incarceration does not just respond to neighborhood conditions — it alters them.
Does environment shift incarceration probability independent of race? The dossier summarizes the evidence as follows: yes, structural and environmental factors are genuine causal variables that operate across racial groups and that shift incarceration probability substantially. White communities that experienced deindustrialization — in Appalachian coal country, in the Rust Belt, in rural manufacturing areas — have experienced significant increases in criminal justice involvement, particularly around drug enforcement. The opioid crisis in predominantly white communities generated incarceration patterns that mirror the earlier crack cocaine crisis in predominantly Black urban communities: economic disruption, substance abuse, criminal enforcement response, incarceration, civil afterlife.
The evidence for environmental and structural factors is strong and replicable. The question of whether these factors fully explain racial disparities in criminal justice outcomes — once structural conditions are controlled for — is more contested. Researchers consistently find that racial disparities do not entirely disappear when poverty, neighborhood, and individual characteristics are controlled. What accounts for the residual is debated: unmeasured structural factors, differential policing density, prosecutorial and judicial decision-making, and the historical processes that produced the structural conditions in the first place.
What the structural analysis establishes is that incarceration risk is not randomly distributed across the population; it is systematically associated with economic and geographic conditions. Two people born into the same society face substantially different incarceration probabilities based on conditions they did not choose. This is a distributional fact. What conclusions to draw from it — moral, policy, or otherwise — is a question the data cannot answer. The data can only establish that the variation is real and that structural conditions contribute to it.
CHAPTER FIFTEEN: RE-EVALUATION
On what the evidence requires us to consider.
We have arrived at several documented facts that, taken together, require a coherent explanation.
The United States incarcerates at a rate roughly seven times higher than comparable wealthy democracies. It spends approximately $445 billion annually on its criminal legal system. Its incarceration rate grew 346 percent in real corrections spending terms between 1977 and 2021 — a period during which crime rates first rose and then fell dramatically, meaning that corrections expenditure growth was not simply a function of crime trends. Its system resolves 97 to 98 percent of federal criminal cases through negotiated pleas, producing a high-throughput conviction mechanism that largely bypasses the adversarial fact-finding the constitutional trial system was designed to provide. It generates $13 billion annually in fine and fee revenue and has forfeited at least $68.8 billion in civil assets since 2000 — with forfeiture rates that demonstrably increase when agency budgets are under pressure.
It has created concentrated rural employment constituencies that resist facility closure independent of facility performance. It has delegated constitutionally mandated healthcare obligations to private contractors whose financial interests are structurally misaligned with the delivery of adequate care. It has produced, in Arizona, a fourteen-year litigation that exhausted every remedy available to a federal court before resorting to the extraordinary measure of stripping state administrators of operational control.
It maintains approximately 3.77 million people under community supervision — on probation and parole — generating supervision fee revenue from those people and maintaining a population with ongoing legal exposure whose members can be returned to incarceration for administrative violations rather than new crimes. It attaches approximately 44,000 collateral legal consequences to criminal convictions, creating a permanent legal category that affects employment, housing, and civic participation for tens of millions of people who have, technically, served their sentences. It generates a criminal record that the FBI attributes to approximately 73.5 million Americans — roughly one in three adults.
The explanations for this configuration are multiple and non-exclusive.
The crime wave explanation: Real crime increases in the 1970s through early 1990s generated genuine public demand for incapacitation and deterrence. The incarceration expansion was, in part, a response to documented public safety failures. This explanation has force. It does not explain why incarceration continued to expand after crime declined, or why it expanded to a level seven times greater than comparable countries facing their own crime challenges.
The political economy explanation: Legislators responded to constituent demand for punitive policies during high-crime periods, then found themselves structurally constrained from reversing those policies by the constituencies that incarceration expansion created — rural employment, union contracts, fine revenue streams, private prison lobbying. The political economy of expansion proved more durable than the political economy of the crime wave that motivated it.
The fiscal incentive explanation: State and local governments discovered, through the expansion of fines, fees, and civil asset forfeiture, that the criminal justice system could generate substantial revenue. Institutions that generate revenue for government tend to persist and grow, because they reduce the fiscal cost of government on net — or appear to, before accounting for all indirect costs.
The structural failure explanation: The institutional configurations created during the expansion period — private healthcare contracts, plea-dominant adjudication, mandatory minimum sentences, rural economic dependency — created path dependencies that are difficult to reverse even when the political will exists. The system is no longer primarily responsive to crime conditions. It is primarily responsive to its own institutional interests.
None of these explanations is complete on its own. The most defensible account draws on all four, weighted by the evidence.
What the evidence does not support is the explanation that the U.S. system exists at this scale because it is uniquely effective at producing public safety. The United States has higher violent crime rates than most comparable countries, including those that incarcerate at one-seventh the rate. Norway's recidivism rate of approximately 20 percent compares to the U.S. rate of 40 to 70 percent. The correlation between incarceration rate and public safety outcomes, internationally, does not run in the direction that would justify the U.S. scale on purely functional grounds.
This is the re-evaluation the evidence requires. Not a moral verdict on mass incarceration — that is for the reader to render. Not a policy prescription — reasonable people can disagree about what alternatives are feasible and at what scale. But an honest assessment of what the documented facts require us to think about the relationship between stated purpose and observable function.
The stated purpose of incarceration in American law is punishment, deterrence, incapacitation, and rehabilitation. The observable function of the system as currently configured includes all of these, plus: rural employment anchor, fine and fee revenue mechanism, private contract revenue stream, rural fiscal transfer recipient, and generator of a supervised population with ongoing legal exposure. The stated purposes did not require the system to grow to its current scale. The unstated functions did.
Understanding this distinction — between what an institution says it is for and what the incentive structure actually rewards — is the beginning of understanding why the system is what it is. It is also the prerequisite for any serious discussion of whether it should be different.
The data on that question is available. The question itself is not complicated. The path from understanding to change runs through the territory mapped in this work: through rural counties that depend on prison payrolls, through fine and fee budgets that governments now depend on, through plea bargaining dynamics that have generated institutional dependencies, through private contracts and public constituencies, through the civil afterlife that sustains incarceration's economic consequences long after the cell door opens.
These are not immovable objects. Finland moved them. Germany never built them to this scale. Norway dismantled them over decades. The United States built something different, for reasons this work has tried to document honestly. Whether it builds something different again is a question that will be answered not by evidence alone — evidence rarely answers questions like that — but by the political choices of people who have read the evidence and decided what to do with it.
This work tried to provide the evidence. The decision is yours.
APPENDIX: DATA NOTES AND CONFIDENCE SUMMARY
On numbers in this work: Every data point in this narrative is drawn from the research dossier assembled from primary sources (BJS, FBI, Urban Institute, BEA/FRED, Institute for Justice, DOJ, Treasury Department, ABA, Vera Institute, and peer-reviewed academic publications). Where the dossier notes contested claims [C] or moderate confidence [VM], this narrative presents them as documented evidence with appropriate qualification rather than established fact.
Key verified anchors used in this work:
- 1.9 million total confined (PPI/BJS, 2023)
- 3.77 million on probation/parole (BJS yearend 2023)
- 73.5 million with FBI-defined criminal records (FBI data)
- ~19.8 million with felony convictions (Couture et al., 2016, using 2010 data)
- $445 billion total criminal legal system spending (PPI 2026)
- $87 billion state/local corrections, 2021 in 2021 dollars (Urban Institute)
- $19 billion state/local corrections, 1977 in 2021 dollars (Urban Institute)
- $13 billion fine/fee/forfeiture revenue (Urban Institute Tax Policy Center, 2021)
- $68.8 billion civil forfeiture since 2000 (Institute for Justice, 3rd edition, 2022)
- 97–98% federal plea rate (ABA Task Force 2023; Supreme Court, Missouri v. Frye)
- U.S. rate 541/100,000; Norway 75; Germany 76; Netherlands 69 (World Prison Brief, 2022)
- Norway recidivism ~20% / U.S. 40–70% (definition-dependent)
- Arizona receivership: February 20, 2026; Judge Silver; 128-page order; 14 years litigation; $2.5M fines; 34,000 covered; NaphCare $300M contract (multiple primary sources)
- Mississippi: 847/100,000; Massachusetts: 118/100,000 (USAFacts/BJS, 2023)
International comparison caveat (noted in dossier [VM]): Cross-national recidivism comparisons use different definitions, follow-up periods, and base populations. The U.S.-Norway gap is large enough that methodological adjustments are unlikely to eliminate it, but direct numerical comparison should be understood as directional rather than precise.
This work was produced as part of an archival research and analytical writing project on the political economy of incarceration in the United States. It does not represent the legal, policy, or editorial position of any institution.
Top comments (0)