The Architecture of Not Knowing
Willful Ignorance and the Societies That Depend on It
The Third Investigation in the “Invisible Dead” Series
Bill Friend • February 14, 2026
I. The Pattern
This investigation began with a simple question about prison deaths. It did not stay simple.
In the first essay in this series, “The Invisible Dead,” we documented a global failure to count deaths in custody. Governments across the developed world—democracies with functioning statistical agencies and constitutional protections—systematically fail to track, classify, or publish data on people who die behind bars. In the second essay, “Who Are the Invisible Dead?,” we examined the demographics of incarceration and found that the people most likely to be among those uncounted dead are overwhelmingly unconvicted, non-violent, poor, and drawn from marginalized communities—people the system had the least justification for holding in the conditions most likely to kill them.
Those two investigations were supposed to be the project. They are not.
As the data accumulated, a pattern emerged that extended well beyond prisons. The architecture of the failure—the specific sequence by which a society produces suffering, avoids documenting it, relabels it into something manageable, and then spends decades solving the wrong problem—was not unique to incarceration. It was a template. And once I saw it, I could not stop seeing it.
The same architecture operates in the abandonment of the mentally ill to the streets. It operates wherever mass casualties become a debate about terminology rather than a demand for accountability. It operates wherever evidence of crimes by the powerful is released to the public and met with outrage that fades into exhaustion. In each case, the evidence is available. The suffering is real. And the society that could act instead arranges, with extraordinary sophistication, not to know.
This essay is an attempt to describe that architecture—not as a conspiracy, not as a moral failing of specific individuals, but as a structural feature of how societies manage information they cannot afford to confront. It is, in the language of economics, a market. There is a supply of ignorance and a demand for it. The people who die in the gap between what is known and what is acknowledged are the unpriced cost of that transaction.
II. The Mechanism
The architecture of willful ignorance follows a consistent sequence. Not every case includes every step, but the skeleton is remarkably stable across contexts, across countries, and across decades. It proceeds as follows.
Step One: A Crisis Exists
This is the precondition. Somewhere, people are suffering in ways that are preventable, and the prevention would require resources, political will, or structural change that the society is unwilling to provide. The suffering is not hidden at this stage. It is simply not yet organized into a form that demands response.
In the 1950s and 1960s, investigative journalists and civil rights advocates documented conditions in American psychiatric institutions that were, by any standard, barbaric. Patients were warehoused in overcrowded wards, subjected to forced treatments, and left in conditions that would have been illegal in a prison. The crisis was visible. The evidence was published. The response—deinstitutionalization—addressed the setting without replacing the care.
In prisons today, the crisis is equally visible to anyone who looks. Overcrowding, medical neglect, suicide, and preventable death are documented by the facilities’ own records—where records exist. The evidence is not concealed. It simply has no audience.
Step Two: The Crisis Gets Relabeled
This is the critical move. A problem that has a clear identity—abandoned mentally ill people, preventable deaths in custody, mass civilian casualties—is renamed in terms that obscure its nature and diffuse responsibility.
The mentally ill, turned out of institutions without community care infrastructure to receive them, became “the homeless.” That single word transformed a healthcare crisis into a housing problem. For more than forty years, American cities have been trying to solve homelessness with shelters, housing-first initiatives, and encampment sweeps—approaches that address the symptom while ignoring the fact that a significant percentage of the people living on the streets are suffering from serious mental illness. Recent federal data indicates that 20 to 30 percent of the homeless population has a severe mental disorder, and some research places lifetime prevalence of any mental health condition as high as 67 percent. Jails have quietly become the nation’s largest mental health providers. None of this is secret. But by calling it “homelessness,” society gave itself permission to treat the problem as one of real estate rather than medicine.
In the incarceration context, the relabeling is subtler. People awaiting trial become “inmates.” People convicted of drug possession become “offenders.” The language of the system erases the distinction between the person who committed murder and the person who could not afford bail on a misdemeanor. Once everyone behind bars is an “inmate,” the public can imagine a homogeneous population of dangerous people—and the deaths of those people carry what we described in our second essay as an “implicit asterisk of diminished concern.”
Step Three: Jurisdictional Fragmentation
Once the problem has been relabeled, the next step—often unintentional, sometimes deliberate—is to distribute responsibility across so many agencies, levels of government, and reporting systems that no single entity owns the problem and no unified picture can emerge.
In the United States, jails are run by counties, prisons by states, immigration detention by the federal government, and juvenile facilities by a patchwork of state and private agencies. Each system has its own definition of a reportable death, its own data standards, and its own oversight body. A person who dies in police custody during transfer to a county jail may not appear in anyone’s mortality statistics. The Death in Custody Reporting Act was supposed to fix this. A 2022 Senate investigation found that the Department of Justice had failed to implement it, with 70 percent of sampled death records incomplete. The data did not disappear because someone shredded it. It disappeared because the system was designed—or at minimum, allowed to evolve—in a way that made disappearance inevitable.
The same fragmentation protects other forms of not-knowing. Mental health care in the United States is divided among state agencies, county systems, private insurers, the VA, Medicaid, and a constellation of nonprofits, each with different eligibility criteria and different data systems. Tracking a single person’s journey from psychiatric hospital to street to jail to emergency room to street again is nearly impossible—not because the information doesn’t exist at each stop, but because no system connects the stops.
Step Four: Oversight Erodes
The final structural step is the weakening of independent watchdogs—the people and institutions whose job is to look where the public won’t.
In the United States, the Department of Justice shifted responsibility for collecting death-in-custody data from the Bureau of Justice Statistics, a research agency with decades of institutional expertise, to the Bureau of Justice Assistance, a grant-making body with no research infrastructure. The result was the total collapse of a data collection effort that had been functioning, however imperfectly, for twenty years. As one expert observed, the only explanation for this failure is “if the goal is to actually not know, and to do everything you can not to know.”
This erosion is not unique to prisons. Investigative journalism—the traditional last line of external accountability—has been gutted by the collapse of local news. The institutions that once assigned reporters to cover county jails, state hospitals, and juvenile facilities no longer exist in much of the country. The watchdogs have not been muzzled. They have been defunded.
III. Supply and Demand
The mechanism described above—crisis, relabeling, fragmentation, erosion of oversight—explains how the architecture of not-knowing is constructed. It does not explain why it persists. Structural failures, once identified, can be repaired. Data systems can be unified. Oversight can be restored. The Death in Custody Reporting Act could be implemented tomorrow. The question is not whether these problems are fixable. The question is why no one with the power to fix them is in a hurry to do so.
The answer, I have come to believe, is that the architecture of not-knowing is not a failure. It is a service. It provides something that the society needs—or, more precisely, something the society has come to depend on. In the language of economics, there is a demand for ignorance, and the system supplies it.
This is not a claim about conspiracy. Most of the people who participate in this system—the politicians, the corrections officers, the judges, the voters—are not consciously choosing to suppress information. Many of them believe, with some justification, that they are doing the best they can within the constraints they face. The county sheriff is understaffed. The state legislator is balancing a budget. The voter has a job and a family and limited bandwidth for policy questions. Each of these is a reasonable individual position. But the aggregate effect of all those reasonable positions is a system that reliably produces suffering and reliably fails to document it.
The demand side of this market operates at a level that is difficult to discuss honestly, because it implicates nearly everyone. When a voter supports a “tough on crime” candidate, they are purchasing a specific product: the removal of certain people from view. The transaction is not stated in those terms. It is framed as public safety, as justice, as accountability. But the underlying exchange is simpler than the language suggests. The voter wants certain people to go away. The politician makes them go away. What happens to them after they go away is, by unspoken agreement, not part of the transaction.
The supply side is the apparatus of plausible deniability that prevents the human cost of that transaction from becoming visible. Fragmented data systems. Inconsistent definitions. Underfunded oversight. Language that collapses the distinction between the dangerous and the merely poor. These are not accidental features of the system. They are the product. They are what the system provides. They allow the voter to maintain what I would call willful ignorance—not the targeted avoidance of a specific fact, but a comprehensive arrangement of one’s informational environment so that uncomfortable truths never arrive uninvited.
The Distinction That Matters
The law has a term for the targeted version of this behavior: willful blindness. The legal doctrine holds that if you deliberately avoid learning a fact—if you are the drug courier who never opens the package—you can be treated as though you knew what was inside. The doctrine presupposes a specific suspected fact and a conscious decision not to confirm it.
What I am describing is broader than willful blindness and, I believe, more dangerous. Willful ignorance, as I use the term, is not the avoidance of a particular fact. It is the construction of an entire way of living that structurally excludes categories of knowledge. The person who does not know that 41 percent of Australian prisoners are unconvicted is not actively avoiding that specific number. They have simply never been in a context where that number would reach them—and they have never sought one out. The not-knowing is maintained at a level that precedes any individual decision to look away. It is built into the media they consume, the conversations they have, the assumptions they carry about who is in prison and why.
This is why willful ignorance is harder to confront than willful blindness. You cannot lay a specific piece of evidence in front of someone and say “you knew.” They didn’t know. They arranged, without ever making a conscious decision to arrange, a life in which they would not know. And the system they live in—the news they watch, the politicians they elect, the language they use—is designed to make that arrangement as effortless as possible.
IV. The Architecture in Practice
The four-step mechanism—crisis, relabeling, fragmentation, erosion of oversight—is not a theory. It is a description of observable, recurring patterns across multiple domains. The following cases are not intended as comprehensive accounts. They are structural illustrations, selected because they demonstrate the same architecture operating in different contexts.
Incarceration and the Uncounted Dead
This is the case we know best, having spent two full investigations documenting it. The crisis: people die in custody at rates that no government reliably measures. The relabeling: everyone behind bars becomes an “inmate,” collapsing the distinction between the convicted murderer and the unconvicted person who could not post bail. The fragmentation: responsibility for counting deaths is distributed across thousands of jurisdictions with incompatible systems. The erosion: the one federal agency tasked with unifying the data was effectively dismantled. The demand: a public that wants certain people removed from view and does not want to know the cost. The supply: a system that removes them and ensures the cost remains invisible.
Deinstitutionalization and the Mentally Ill
The crisis: hundreds of thousands of seriously mentally ill people living without adequate care. The relabeling: “the mentally ill” became “the homeless,” transforming a medical emergency into a logistical problem. The fragmentation: responsibility for mental health care is scattered across state, county, federal, private, and nonprofit systems with no coordination. The erosion: community mental health funding was never adequately provided; the institutions closed, but the replacement infrastructure was never built. The demand: a public that wanted the institutions closed (for legitimate reasons) but was unwilling to fund the alternative. The supply: a system that closed the institutions and relabeled the consequences, allowing the public to believe the problem had been addressed when it had merely been moved outdoors.
This is the oldest and most complete case study available, because the cycle has had forty years to run. The outcome is visible on the streets of every major American city. It is also visible in the county jails that have, without anyone voting for it, become the largest mental health facilities in the country. The mentally ill were not abandoned by a single decision. They were abandoned by a sequence of individually defensible decisions made by people who could each point to someone else as the responsible party.
The Pattern Beyond These Walls
Incarceration and deinstitutionalization are the cases examined in detail here because the evidence is thoroughly documented and the architecture is fully visible. But the reader who has followed the mechanism this far will recognize it operating elsewhere—in the management of mass civilian casualties overseas, where the argument over what to call the killing consumes the space that should be occupied by the killing itself; in the handling of evidence that powerful people committed crimes against children, where millions of pages of documents produce the appearance of transparency without its substance; in the treatment of veterans, of environmental contamination in poor communities, of elder abuse in underfunded care facilities. The specifics vary. The architecture does not.
In some of these cases, the not-knowing operates through darkness—information that is suppressed, fragmented, or never collected. In others, it operates in plain sight, using an excess of competing narratives to produce the same effect as a deficit of information. When everything is contested, nothing is known. When evidence is released but no one acts on it, revelation and concealment become functionally identical. The reader who wishes to test the framework described here has no shortage of material. The architecture is not hiding. It is, like the suffering it manages, available to anyone willing to look.
V. The Uncomfortable Conclusion
I began this project believing that the problem was primarily one of information. If the public knew that most people in prison hadn’t been convicted, they would care. If they knew that no one was counting the deaths, they would demand accountability. If they knew that the “homeless” were largely the abandoned mentally ill, they would insist on treatment. The first two essays in this series were written from that premise: lay the evidence on the table, and the system will have to respond.
I am no longer sure that is true.
The evidence, in every case examined here, is already available. It is published by government agencies, compiled by international monitoring bodies, documented by journalists, and catalogued by advocacy organizations. The prison mortality data—incomplete as it is—is public. The demographic composition of the incarcerated population is public. The fact that the mentally ill were turned onto the streets without adequate care is not disputed by anyone. The information is not hidden. It is ignored. And the reader who looks beyond the cases documented here—at civilian body counts overseas, at document releases that produce no prosecutions, at any of the other places where the architecture is plainly visible—will find the same pattern: evidence available, action absent.
And it is ignored not because people are stupid or callous, but because acting on it would be expensive—financially, politically, and psychologically. To acknowledge that most people in prison do not need to be there is to acknowledge that the system you have been supporting, through your votes and your taxes and your silence, has been causing unnecessary suffering on an industrial scale. To acknowledge that the “homeless” are mentally ill is to accept a debt of care that no budget currently funds. To acknowledge what is happening in places where the argument over labels has replaced the argument over bodies is to confront the possibility that people and institutions you support have committed acts you cannot defend. Each of these acknowledgments carries a cost.
Willful ignorance is the decision—made not once but continuously, not consciously but structurally—that the cost of knowing is higher than the cost of not knowing. It is a rational calculation, if you exclude from the equation the people who pay the price of your ignorance. The pre-trial detainee who dies of an untreated infection. The schizophrenic man who freezes to death under a bridge. The unnamed dead in places the reader already knows but the architecture has taught them not to name.
A society that arranges its informational architecture to avoid these costs is not a society that has failed. It is a society that is functioning as designed—designed not by any single architect, but by the accumulated preferences of millions of people who have decided, independently and repeatedly, that they would rather not know. The architecture of not-knowing is not a bug. It is a feature. And the people who die inside it are the operating cost.
Breaking the Contract
There is a bilateral contract at the heart of this architecture. The politician promises to keep the problem out of sight. The citizen promises not to look too closely at how that is accomplished. Both parties benefit. The politician stays in power. The citizen keeps a clean conscience. The only people who lose are the ones inside the system—and they, by definition, are not party to the contract.
Breaking that contract requires something more than information. It requires the removal of the possibility of not knowing. That is what the detective does when laying evidence before a suspect: not merely presenting facts, but eliminating the suspect’s ability to claim ignorance. After the evidence is on the table, continued denial requires active effort. It becomes visible—to the suspect, to the room, to the record.
That is the function these essays attempt to serve. After reading that 41 percent of Australian prisoners have not been convicted, you cannot claim you didn’t know. After reading that no government reliably counts its prison deaths, you cannot claim the system is working. After reading that the mentally ill were relabeled “homeless” and left to die on the streets, you cannot claim the problem was housing. The evidence is on the table. What was willful ignorance now requires willful denial—and denial, unlike ignorance, is difficult to sustain in the presence of a clear record.
This will not, by itself, change anything. Information does not automatically produce action. But it changes the moral calculus. The person who did not know has no obligation. The person who knows and does nothing has chosen. And the record of that choice—the evidence on the table, the data in the public domain, the names and numbers and preventable causes of death—is permanent. The architecture of not-knowing depends on the absence of a record. These essays are an attempt, however small, to ensure that the record exists.
Problems cannot be solved if they are not properly identified. And they will not be properly identified as long as the society that produces them has a structural incentive to mislabel them, fragment responsibility for them, and erode the institutions that might hold someone accountable for them. The first step is not a solution. The first step is to see the architecture for what it is—and to say so plainly, in language that does not soften what should not be softened.
The information is not unknowable. The dead are not invisible because they cannot be seen. They are invisible because we have built, collectively and meticulously, a world in which we do not have to look.