Standing in the well of a jail on New York’s Rikers Island as profanities rain down on you from the cells above, you realize the absurdity of academia’s most celebrated book on incarceration. Discipline and Punish, by the late French historian Michel Foucault, criticized jails and prisons for subjecting inmates to constant, spirit-crushing surveillance. The truth is that surveillance goes both ways in correctional facilities. Inmates watch their keepers as intensely as they are watched—and usually much more malignly.

Illustration by Jock.

Jails are the ideal testing ground for romantic myths about incarceration. As policing has gotten more efficient at nabbing wrongdoers over the last decade and a half, it has pumped a growing volume of increasingly troubled individuals into the jail system. Governing that population is a management challenge more complex than that faced by any other criminal-justice institution. Yet jails, unlike prisons, remain largely out of sight and out of mind. This public ignorance is unfortunate, because jails have been evolving important principles for controlling criminal behavior of late, ideas that directly contradict the Foucauldian critique.

To understand the difficulties of running a large jail, imagine that your job is personally to shepherd each of the thousands of commuters streaming through New York’s massive Penn Station to their trains safely and on time . . . except that the commuters are all criminals who keep changing their travel plans, and their trains, to which they don’t want to go, have no fixed timetables. A cross-section of the entire universe of criminal offenders, from the most hardened murderer to the most deranged vagrant, cycles through the nation’s 3,365 jails. But the majority of jail inmates show up with no predictable release date, since they have as yet only been charged with a crime and are awaiting a trial that may or may not occur and whose duration is unknown. Even before their trials begin, they may make bail at any moment and be released. Planning for pretrial detainees is therefore no easy task. “The ones who stay less than 36 hours drive you out of your mind,” says Michael Jacobson, a former corrections commissioner in New York City. “You think: ‘Couldn’t you have made bail ten hours ago rather than coming into my facility?’ ” Prisons, by contrast, hold only post-conviction defendants who have been found guilty or pleaded guilty and have been sentenced to a known term of more than a year. (Prisons and jails differ as well in their government overseers: the former are run by states and the federal government, the latter by cities and counties.)

Jail administrators are obligated to get pretrial detainees back and forth to court on time and to keep them safe until their cases are completed. But pretrial detainees are just less than two-thirds of the nation’s approximately 780,000 jail inmates. The remainder consists of post-conviction defendants with a sentence of a year or less, who serve their time in jail; post-conviction defendants sentenced to more than a year and awaiting transfer to prison; parolees and probationers who have violated their conditions of release; illegal immigrants detained for U.S. Immigration and Customs Enforcement; and inmates in transit between prisons.

These populations show up at all hours, often with no background information on who they are. Their turnover rate is extremely high: jails process as many admissions and releases in two months as state and federal prisons (which hold about 1.5 million inmates) process in a year. Managing that “churning mass of humanity” is a nightmare, says Jacobson, who now directs the Vera Institute for Justice. “So many arrestees lead unbelievably disorganized lives”—but as soon as they enter a jail, the jail becomes responsible for their well-being.

Two men in street clothes amble down the wide, buffed corridor of the Otis Bantum Correctional Center, one of ten razor-wired fortresses on Rikers Island that together hold nearly 14,000 inmates. They’re as odd a couple as any pair of Shakespearean rustics: one short and white, with mismatched eyes that look as though they’ve been squished by a pickup truck; the other tall and black, with a gap-toothed smile spreading broadly across his craggy face. The very picture of bonhomie and goodwill, the two flag down a passing official to share the happy news: they have just made bail after a day inside and are going home. “Congratulations. Now please don’t come back to my hotel again,” Deputy Chief of Staff of Operations Mark Cranston responds cheerfully.

The chances that they will honor Cranston’s request are slight. At Rikers, 40 percent of all inmates return within a year of release. Such recidivism is typical; all jails have “frequent fliers” who cycle through repeatedly. “I see these guys leave and they’re back in two days,” says Gerald, a slender, already-sentenced drug offender in a green uniform who is swabbing the floors of the Otis Bantum intake area. Fifteen men are standing, sitting, or lying down in a holding pen inside the bustling intake office. Plexiglas covers one side of the cell to protect the staff from “splashing” incidents—drinks or bodily fluids thrown at them by cell occupants. One tall detainee in a Yankees jacket is talking on a phone that reaches into the cell, trying to line up bail. As Gerald cleans, he keeps his distance from the new admits. “This place can get you in trouble; there’s no need to get involved,” he says.

Over the next few days, Rikers officials will try to gather as much information as they can on the men in the pen, seeking to determine how securely to house them and whether they need medical or psychological care. They will analyze their criminal records, intake questionnaires, medical examinations, and current behavior. Such inmate classification is the cutting edge of jail management. Jails are only now starting to recognize the importance of rigorously analyzing information to maintain order, just as policing has in the last decade. Some jails still practice “open bed” classification, housing an inmate wherever there is an empty bed or, at best, separating felony and misdemeanor pretrial detainees. But careful inmate classification acknowledges that a Mike Tyson in on a drunk-driving charge, say, is likely to be more dangerous than many a felony auto thief—and should be housed accordingly.

At Rikers, new inmates spend up to 72 hours under observation in quasi-quarantine, until the results of their tuberculosis and other medical tests come back. Outside one such quarantine area, a square, windowed dormitory filled with rows of mussed beds, a sentenced inmate working as a suicide-prevention aide roams the hall every ten minutes, checking on new admits suspected of suicidal tendencies. Inside, a pallid transvestite with widely set eyes, several missing teeth, and a curtain of hennaed hair stares intently at a group of visitors. Though he claims that he has just been slapped for not providing a sexual favor to another inmate, the 28-year-old heroin addict otherwise isn’t worried about getting attacked. “At Rikers, unlike New Jersey, they cater to you,” Della says. “They won’t put me in protective custody here”—isolating an inmate for his own safety, that is—“or make me feel like an animal.” (Recent studies suggest that rape among inmates is far less prevalent than was previously claimed.)

The spread of quality-of-life policing, which targets low-level offenses like aggressive panhandling, public urination, and littering, has brought a more mentally unstable, troubled population into jails—one that mental hospitals would have treated before the deinstitutionalization movement of the 1960s and ’70s shuttered most state mental hospitals. In fact, jails have become society’s primary mental institutions, though few have the funding or expertise to carry out that role properly. Mental illness is much more common in jails than in prisons; at Rikers, 28 percent of the inmates require mental health services, a number that rises each year. “People are coming right off the streets with a whole range of street problems,” Jacobson reports. “You have to deal with them immediately and figure out: ‘Are you a dangerous lunatic, or just tough?’ ”

Jail administrators worry constantly about inmates’ killing themselves—in 2008, Rikers spent $5.3 million in overtime pay for officers to sit watch, 24 hours a day, outside the cells of potential suicides. Mentally ill jail inmates stay longer than other inmates because of pretrial competency hearings and other self-induced complications. One mentally ill Rikers inmate has been incarcerated for five years, during which time he has staged numerous diversionary courtroom dramas to defer his case. Another “has defied all efforts to safely get her to court by injuring herself every time a hearing is approaching,” sighs Assistant Deputy Warden Helena Smith, who oversees a Rikers mental health ward.

The Kent County Correctional Facility, a pair of squat, striped cylinders housing 1,300 inmates in Grand Rapids, Michigan, is a leader in correctional mental health care. Mental clinicians sit in on every intake interview, and they have access to mental health records on every inmate dating back to 2003—an unparalleled degree of medical record-keeping. Every morning, clinicians and corrections supervisors discuss each detainee who has been placed in a suicide cell—a camera-monitored cell lacking a table or any kind of fixture that the inmate could use to hang himself. The discussion sounds as though it is taking place in a hospital, not a jail.

The head clinician brings up a picture of a bedraggled female on a PowerPoint screen. “Miss Wilson is a frequent flier,” she begins. “I tried to direct her to the positive aspects of her life; she said she didn’t have any.” “Does she have a place to go?” asks a supervisor. “She wouldn’t engage with me,” the clinician answers. A series of men appears on the screen. “Derek wanted someone to talk to; he has a strained relation with his child’s mother. I allowed him some time to process those thoughts and feelings. Mr. Taylor continues to be evasive. He smiles and reported that he doesn’t want to hurt himself, but if he had to move, he would make no promises for his safety. He says he hears voices, but he doesn’t tell me what they say.”

The night before, a man just sentenced to life imprisonment for second-degree murder had been moved into a suicide cell under protest. He now lies on Cell 34’s green bunk, under a long horizontal slit of a window, tightly wrapped in a blanket like a shroud, his face invisible; a dictionary, slippers, manila envelopes, and rolls of toilet paper litter the floor. In his old cell, a dozen sheets of paper are attached with toothpaste to a wall, each displaying a softly shaded word: judas, guilt, death, sinner, empty, fear, pain, lust, hope? A blob of green-and-white soap on the wall holds a pencil. This convict has required more than 1,000 staff interventions during his 20-month stay, over ten times the number of incidents that a typical non–mentally ill inmate would generate over a comparable period.

Outside the convict’s cell, other mentally ill inmates wander around the cell block’s dayroom (the common area where inmates not on lockdown spend their time) following a group-therapy discussion; a few sit at a table with a jigsaw puzzle. John, an elfin Sudanese drunk driver who looks as though he has been dusted in flour, complains in barely comprehensible English about getting bossed around and says that he refuses to take showers. Spencer, a 45-year-old unmarried father in on an assault-and-battery charge that he won’t discuss, says with an affectless, unblinking gaze that he has tried to get a job but gets too stressed out to hold it.

A few mentally ill inmates defeat all efforts to get them to take their medication. They cannot be forced to do so without a court order, but they refuse to go to court, thus requiring a judicial order to extract them from their cells. Others hide the pills or spit them out. Once treated and sobered up, some mental-needs inmates become model prisoners, while others remain difficult to control. The question of whether mentally ill inmates are disproportionately violent is a fraught topic. The Wisconsin state auditor reported in March 2009 that mentally ill state prisoners commit nearly 80 percent of assaults on staff, though they are 30 percent of the state prison population. At Rikers, the rate of fights in the mental health observation units is 12 times greater than in the general population.

One day in February, a black-uniformed search team is mustering outside Rikers Island’s redbrick Robert M. Devoren Center after an assault by a mentally ill teen. A delusional 17-year-old robbery suspect who fancies himself “King the Punisher” has slashed a fellow adolescent’s ear with a razor fashioned from radiator metal. A perennial troublemaker, the assailant has been at Rikers for two years while his trial stretches on. Now the elite search team, drawn from security officers across the island, will search every crevice of the jail for more weapons, while the victim goes into protective custody. The assailant is at a court hearing as the search team gathers; when he returns, officials will try to bring a new charge of attempted murder against him.

Two of Rikers Island’s “punitive segregation” wings—areas where inmates who have broken the rules stay confined up to 23 hours a day in a single cell—are devoted to mentally ill detainees, like King the Punisher. improve the moment, exhorts a mural leading into the two wings, located in the George R. Vierno Center, a low tan building with blue corral-type rails on top. The most violent mentally ill offenders, almost all of whom have assaulted staff, are housed in the Mental Health Assessment Unit for Infracted Inmates (MHAUII), a long, double-tiered room of cells with heavy doors. The hatches in the middle of the doors where officers deliver food are bolted like safes, because they also serve as complicated mechanisms for handcuffing inmates.

Before a detainee can leave the MHAUII for a visit to the doctor or other appointment, he must be handcuffed and searched for weapons, all while isolated beyond striking range of another human being. He is first cuffed from behind through his food tray while in his cell, then escorted to a small cage containing a battered gray magnetometer. There he is uncuffed through an opening in the cage and asked to strip and walk through the magnetometer. After dressing again, he is recuffed through the cage. Even these precautions don’t always protect officers. A cuffed and scanned inmate on his way to the health clinic has just kicked an officer in the face, cutting him deeply. That inmate will now have leg irons as well as handcuffs when outside his cell.

The violence of MHAUII inmates exceeds the jail’s capacity to punish it. No Rikers inmate may spend more than 90 days at a time in punitive segregation, but the most recalcitrant accumulate far more punitive time (known at Rikers as “bing days”) than they can pay in a single stretch. The average MHAUII resident owes 303 bing days; one current inmate owes 4,000. When an inmate with a backlog of punitive segregation leaves the island, his debt stays on the books. If he returns, he’ll go right back to punitive segregation to start paying down his bing days again.

Even in punitive segregation, mentally ill Rikers inmates continue to receive treatment from a clinical psychologist and a psychiatrist, who try to persuade them to enter behavior-modification therapy. As a reward for good behavior, a group of MHAUII detainees in orange jumpsuits clusters around a TV set in a tall glassed-off room. Outside several cells on the block, single officers sit on stools, each staring at the cell’s door. Rikers, unlike Kent County, may not film inmates on suicide watch inside their cells; instead, it must post an officer outside each cell, 24 hours a day. One such detainee, a jailhouse bully awaiting trial on gun, assault, and drug charges, is curled up in a ball under his green mattress, while his monitor sits watching.

Illustration by Jock.

Such surveillance, said Foucault in Discipline and Punish, turns inmates into powerless subjects of the “disciplinary” state. Foucault called jails and prisons “Panopticons,” after a circular prison model by eighteenth-century philosopher Jeremy Bentham that let guards observe a large number of inmates from a central position.

But jails and prisons are also reverse Panopticons; to walk around one is to be under constant observation from the inmates. The moment that Deputy Warden Thomas Hall enters another Rikers punitive segregation unit, inmates watching from their cells unleash a torrent of obscenities: “Fucking Hall!” “Call Hall!” While the authorities’ surveillance of inmates is often protective—as in the ubiquitous suicide watches—the inmates’ surveillance of the authorities can be aimed at corrupting them. Like surveillance, power in jails flows between officers and inmates in multiple directions.

Rikers Island’s most dangerous inmate stands watching at his cell door in Twelve Main, a small, square space that contains the island’s maximum-security prisoners. Lee Woods, tall and well built, with a broad mouth, has just been sentenced to life in prison for the murder of New York police officer Russel Timoshenko. Woods was driving a stolen SUV with two other thugs on the night of July 9, 2007, when Officers Timoshenko and Herman Yan pulled it over. Gunfire burst from the SUV. Timoshenko died of face and neck wounds five days later, and Yan was seriously injured. Now the former Bloods leader and career criminal stares impassively at some visitors, a white T-shirt tied rakishly around his head, as a corrections officer sits outside his cell.

Woods is in Twelve Main—where the walls are made of reinforced steel so that inmates can’t rip out sinks and toilets for weapons—because of the eerie power he exercises over corrections officers. After his arrest for the Timoshenko murder, he arrived at Rikers owing 3,000 bing days, accumulated from 14 assaults on staff, three arson attempts, and several contraband violations during previous stays. In earlier state-prison stints, he had devised homemade razor blades and thrown feces at staff. Yet on his current visit to Rikers, this monster managed to seduce at least six staff members to break the rules on his behalf. Two rookie officers brought him marijuana, tobacco, and alcohol; another two officers supplied him with a cell-phone memory chip and a handcuff key (the key was discovered inside his bowels by a magnetometer); a female officer exchanged photos with him during her frequent stops outside his cell (a violation of fraternization rules); and a captain let him make a phone call without supervision. When news of Woods’s treatment broke in the summer of 2008, New York police officers were so angry that they allegedly went on a ticket-writing spree against vehicles belonging to the Department of Correction, a separate entity from the NYPD.

Woods could corrupt his keepers because of the two-way nature of surveillance and power in jail. “You’re in a mirror as a corrections officer,” says Rikers captain Sean Jones. “The inmates watch you continuously; they can get a heads-up on you before you get one on them. They know when you get a haircut, whether you’re wearing your wedding ring any longer. ‘Your hair looks great today,’ they’ll say, or ‘I see you took your ring off.’ ” If a supervisor berates an officer, they will appear to sympathize.

The goal is to gain control of the officer. An inmate’s most potent method of corruption is to persuade an officer to break the rules for him. “Inmates know that once they get a corrections officer to do something for them, even if it’s just bringing them a cheeseburger from McDonald’s, they own the officer,” says Frank Straub, the police commissioner of White Plains, New York. It is illegal to bring an inmate so much as a stick of gum, as corrections officers learn from their first day in the academy. But there will always be a few officers who are turned by a skilled con man. The manipulators test officers quickly—looking for a sign of weakness or perhaps kindness to exploit, or using intimidation to try to bring an officer to heel. “Inmates say it all the time: ‘I’ll see you when I get out,’ ” says Jones.

The threat can be realistic because at Rikers, as at other urban jails, inmates and officers often share community ties. The Rikers workforce is predominantly minority; there is a proud tradition in black families of correctional work, dating from the time when police and fire departments were less friendly to blacks. Given the black incarceration rate, the chance that an officer and an inmate know each other is not negligible. Most officers don’t let any prior acquaintance they have with an inmate—which they’re supposed to report to supervisors—affect them; most perform their vital law enforcement function with honor. But for a few, familiarity with an inmate or intimacy with the culture of crime raises the risk of malfeasance, and not just because of fear of retribution when the inmate is released. The president of the New York corrections officers’ union recently addressed the problem of fraternization on the union’s website. “Are you just willing to surrender all that you’ve worked for and all that your family is proud of just because of some individual who lives down the street from you and you want to be a part of some acceptable click [sic]?” wrote Norman Seabrook. The 1 percent of officers “who say it’s ok to bring in that cell phone or that contraband for an inmate, because they are ‘friends’ ” put the lives of their colleagues at risk, Seabrook warned. “Today it’s a cell phone; tomorrow it’s a 9 millimeter.”

The results of any favoritism, even if it’s less shocking than bringing contraband to a cop killer, are disastrous for a jail. “When a corrections officer builds a relation with an inmate, the system starts to collapse,” says Straub, who has studied corruption in New York prisons. “The whole process is undermined.”

On October 18, 2008, four months after the Woods contraband scandal came to light, 18-year-old Christopher Robinson was killed in his Rikers cell by three fellow adolescents. Investigation into the murder revealed a different locus of corruption.

Officers Michael McKie (“Mack”) and Khalid Nelson (“Nel”), with help from Officer Denise Albright (“Mama A”), had licensed 12 young thugs (the “Team”) to enforce discipline on their cell blocks by means of an extortion racket. New inmates had a choice: they could join “the Program,” which meant forking over a percentage of their commissary accounts and telephone privileges to the Team; or they could receive “spankin’s” (beatings) in their cells. Nelson and McKie taught the Team how to beat the holdouts so as to avoid bruises that supervisors might detect. Team members got to dictate who sat on which chairs in the dayroom, who went to the bathroom and when, and who could go into and out of his cell. Lowly lookouts policed the communal seating areas, “pop-off dummies” delivered the beatings, and the top dogs controlled the housing areas. In exchange for their extortion privileges, Team members were to keep order on the wing.

Throughout the history of corrections, officers seeking to maintain order have tried to exploit the brutal authority that inmates exercise over one another. Mississippi and Louisiana prison officials, among others, issued pistols to selected prisoners to keep other inmates in line in the 1940s and 1950s, writes John DiIulio in Governing Prisons, with predictably predatory results. A respected Texas prison commissioner, George Beto, tried to civilize this so-called con-boss system in the 1960s by more carefully choosing the inmates who would perform staff functions, but the same abuses developed, including quasi-official beatings of noncompliant inmates. Other institutions, like the small South Dakota jail where Kent County captain Randy Demory once worked, had more informal means of delegation. “A guard got a local college football player out of a cell and put him into another cell with instructions to ‘settle these boys down,’ ” Demory recalls. “The football player took care of business and was rewarded with extra food.”

McKie and Nelson’s Program had a particularly dire outcome. Rikers officials had transferred Christopher Robinson into McKie and Nelson’s housing unit in an effort to disrupt Robinson’s own predatory behavior toward fellow inmates in a different cell block. Accustomed to being the top dog himself, he refused to join the Program. Two days later, three inmates known as “Pop Brim,” “Ant Luv,” and “Fire” beat him to death.

Such delegation arrangements are no longer officially tolerated, yet they keep breaking out. Why? Perhaps because keeping a lid on inmate violence is so difficult. After all, the other benefits to corrections officers from such schemes appear modest. Police corruption usually entails considerable financial gain for the dirty cop, but in jail there are few deep pockets. At best, McKie and Nelson were freed from the necessity of patrolling the cell block every hour or so. They were undoubtedly lazy (and probably also part of the same criminal culture to which their charges belonged), but what we should learn from the Program scandal is just how great the challenges for corrections officers are. Left to its own devices, inmate society is not carnivalesque spontaneity, as the Foucauldians might have us believe, or grassroots democracy, as proponents of the misguided 1970s and ’80s “prisoner self-governance” movement suggested, but Lord of the Flies cruelty in which the strong try to control and exploit the weak.

The symbols of detainee power would be laughable if the means of attaining them weren’t so dangerous. Power-seeking inmates will try to control bathroom use and food consumption. Low-status inmates will wind up excluded from the dayroom. Commissary accounts, with which inmates purchase snacks and personal hygiene items from the jail store, are an obvious target for extortion. And chairs are a particularly useful tool of domination. A detainee sitting on a stack of four chairs is broadcasting: “I’m king.” A battle-gray adolescent dayroom at Rikers recently got rid of its freestanding chairs in favor of a green cement picnic table with attached benches. “The inmates were trying to control the seating: stacking chairs and not allowing certain people to sit,” says Edmund Duffy, the tall, energetic warden who now runs the jail where the Robinson murder occurred. Eliminating movable chairs solved another problem as well: they’re the first thing grabbed during a fight.

The predatory drive is particularly strong among younger inmates. Not only are young detainees in thrall to their hormones, but only the worst of the worst adolescent criminals go to jail, rather than to the far milder juvenile-justice system. Three-quarters of the adolescent detainees at Rikers are awaiting trial for the top felonies, such as homicides, shootings, and robberies, a far higher percentage than in the detainee population at large.

In such a world, ordinary objects must be viewed as potential weapons—such as a King of Hearts card containing a razor blade made from a battery shell that had just been discovered during one of my trips to Rikers—and attacks can come at any time. Moments before I entered Rikers’s Central Punitive Segregation Unit, which houses non–mentally ill infractors, an inmate had thrown an unknown liquid at a guard—it could have been water or urine—angry that he had to wait to use the showers. Red lights flashed throughout the facility, and a team in riot gear trotted by to make sure that the unit stayed calm, as the inmate looked coolly around while being escorted to a medical evaluation.

Simply directing the movement of bodies in a jail can be a monumental task. Officers work alone or in pairs, vastly outnumbered by the inmates. One evening at the intake area of Rikers’s Anna M. Kross Center, a 1960s facility whose anachronistic baby-blue geometric security gates appear to have been inspired by Mondrian, several dozen inmates press their faces against the barriers of six small pens, their features barely visible through the fine blue mesh. A table outside the pens is strewn with the recent admits’ sneakers, MP3 players, and jeans. A wiry inmate with dreadlocks in an orange jumpsuit shouts at a visitor, “You have food poisoning, too!” as he is escorted from a pen into the main jail. The detainees are all on different trajectories to and from court, the Bellevue mental hospital, and punitive segregation, and all need to be searched or are waiting to be fed. Getting 20 bodies from court and through the search procedures without a fight requires considerable command presence and finesse, especially if some inmates have previously been separated for security reasons.

Now, imagine that you are a corrections officer in such a situation. Your safety depends on inmates’ cooperation. But there’s a very fine line between saying, “Hey, guy, work with me here,” and letting an inmate call the shots. Time and again, officers have crossed that line, trying to solve the inmate-control problem by ceding authority for order maintenance to the inmates themselves. “It’s not easy managing this kind of humanity,” says former corrections commissioner Jacobson. “One way to do it is to make deals between yourself and the people.”

Observers are divided over how devastating a managerial failure the Robinson case represents. Most acknowledge that in an institution as large as Rikers, some bad things will happen, despite management’s best efforts. Bernard Kerik, whose accomplishments as New York corrections and police commissioner in the 1990s have been tarred by a federal indictment for corruption and tax fraud, is the most damning. “If the supervisors were doing their job, there is no way that you could have enforcers,” he says. “Where was the captain? If he was making his rounds, inmates should have told him.”

Current New York City corrections commissioner Martin Horn disagrees. The Robinson corruption scandal was essentially undetectable, he argues. The noise of opening and shutting heavy gates to cell blocks means that supervisors can’t secretly enter a housing area, and participants had a system of signals to alert one another about approaching visitors. “I have to rely on the integrity of my officers,” says Horn. “Without integrity, all the staff in the world won’t make a difference.” The ethic among inmates and officers against informing further complicates efforts to detect corruption, Horn adds: “There may be no secrets in a jail, but there’s also no snitching.” A month before Robinson’s death, the mother of an inmate on the McKie-Nelson wing told Rikers officials that her son’s eye socket had been fractured after he refused to join the Program—but when questioned by authorities, the inmate responded, “I ain’t talkin’.” Horn doesn’t know whether other officers were aware of the ring, though the likelihood that no other adults saw what was happening seems small. In a paramilitary organization, however, bad news does not flow easily up the chain of command; a jail leader constantly needs to figure out ways to extract the facts on the ground, such as showing up at 3 am to talk to inmates and staff.

Compared with New York police-corruption cases—such as the Dirty 30 in the NYPD’s 30th Precinct in the early 1990s, in which more than two dozen officers were stealing drugs and money from West Harlem drug dealers—the McKie-Nelson affair is peanuts and does not discredit Commissioner Horn’s achievements overseeing the most complex jail facility in the country. He has continued to drive violence down at Rikers through data-driven managerial accountability; the jail bears no resemblance to the inmate-dominated anarchy of the mid-1990s (see “Why the Jails Didn’t Explode,” Spring 1999). In 1995, there were more than 1,000 inmate slashings; in 2008, there were 21, with a total population down by just 2,000 or so. Horn has kept the facility clean and orderly through rigorous attention to prisoner classification and Broken Windows–type detail. New York City’s jail-homicide rate is a small fraction of Baltimore’s, Philadelphia’s, Chicago’s, or Los Angeles’s.

Nevertheless, recent revelations of officially sanctioned favoritism toward inmates further suggest that top management has lost contact with some operations on the ground. The most damning disclosure concerns the royal treatment that the politically influential Hasidic rabbi Leib Glanz secured for Jewish inmates, benefits that included an elaborate, catered bar mitzvah in a Manhattan jail for the son of an inmate; semiweekly feasts for Jewish inmates in the rabbi’s office; and inmate use of the rabbi’s private phone to call girlfriends and bookies, according to the New York Post. When such special treatment is conducted with the approval of jail officials, as Glanz’s apparently was, it is even more destructive of correctional fairness than surreptitious favors passed from a guard to an inmate—and when targeted at an already resented ethnic or religious group, it is the most inflammatory of all.

However outrageous the Jewish dispensation in the New York City jail system, the more common problem throughout corrections is inmate misbehavior and secret guard corruption. Such corruption is best prevented by ensuring order through discipline and as much information-gathering on inmates as possible. These principles remain abhorrent to a Foucauldian perspective on corrections, but they’re essential to sound management.

Foucault criticized visual surveillance as an oppressive encroachment on inmate autonomy. In fact, visual surveillance has only a limited effect on inmate behavior and leaves officers largely ignorant about what happens on the cell block. The D1A pod in the Kent County Correctional Facility is a semicircle of cells on the second floor of a maroon atrium, holding the jail’s most assaultive inmates. It’s a pie-slice-shaped version of Bentham’s Panopticon—guards stay behind a glass barrier observing the cells in front of them, emerging only for their prescribed tours of the cell block. Recently, the occupant of Cell 37 chiseled out the plate holding his window to the point where he could stick his hand out. Every time he heard the guard entering the cell area, he would replace the plate with a papier-mâché version made out of toilet paper. “The incident resolved for all time the question: ‘Do we know what’s going on inside the cells?’ ” observes Captain Demory.

More important than visual surveillance is information-gathering. Just as the NYPD started debriefing every suspect it arrested in the 1990s to collect knowledge about unsolved crimes, progressive corrections officials recognize the need for grassroots information in fighting jail crime. “We teach our officers that the most important thing they can do is to listen to the inmates,” says Rikers deputy warden Hall. “Jot things down: Who’s Big Daddy, or Red-O? Feed it up the chain so we can figure out who’s running things, who the predators and victims are. When someone gives you information, be receptive; don’t shut it off.” The Kent County jail reports the information it gathers to the Grand Rapids Police Department—in 2008, 70 reports about unsolved crimes, including 23 homicides and six robberies.

The imperative to be open to information from inmates is in tension with officers’ instinct to erect a wall between themselves and their charges, a useful check against corruption. Corrections officers develop a strong degree of skepticism toward everything they hear. “If it’s not verifiable, you have to take it as not true until you can prove otherwise,” says Rikers captain Jones. Indeed, officials can develop a hair-trigger authority reaction to the barrage of claims and requests that they face. As Ernest Lewis, assistant warden of New York’s Westchester County jail, walks though a high-security area, a wiry inmate with a history of making trouble shouts through a barred cell: “I’m about to lose my cool. Can I switch to a pen?” In a flash, the previously relaxed warden turns on his command presence: “You better chill out,” he warns.

Illustration by Jock.

Despite the complexity of the officer-inmate relationship, an emerging philosophy of correctional design, “direct supervision,” seeks to break down the physical and psychological barriers between officers and inmates. In traditional prisons and jails, the corrections officer stays behind a barrier at a workstation, emerging only to make his rounds on a predetermined schedule, while inmates spend their day either in their cells or in a group dayroom. “It’s difficult to manage behavior under such conditions,” says Demory, “because for 59 minutes of the hour, the inmates control the turf.” The traditional design’s passive management style is good at containment but not at shaping the culture, he adds. Demory wouldn’t let me into the remotely supervised high-security D1A pod because the inmates, beyond the influence of the guards, would likely start exposing themselves and throwing bodily fluids when I entered. “Right now, it’s settled down; the officers like it nice and quiet,” he explained.

In a direct-supervision facility, by contrast, when inmates congregate out of their cells, the corrections officer is in the same space, either at an accessible workstation or circulating among them, like a community-policing cop walking his beat. The objective is to break down any distinction between the territory of inmates and of officers; the officer is supposed to talk with inmates, set the tone, and intervene immediately in predation and misbehavior.

The direct-supervision model has had impressive results in managing one of Rikers Island’s most difficult populations. A few dormitories of the adolescent jail are devoted to a program in which officers engage constantly with detainees to try to encourage self-control and respect for authority. One day in February, about 50 teenagers enrolled in the Institute for Inner Development sit quietly on immaculately made cots that extend in three rows down a long, shady hall; a fellow inmate is mopping the floor. Francis X, a slender, goateed officer, speaks deliberately and with mellow self-assurance to the group: “I have to say that y’all are quite remarkable. You make me proud. You’ve only been here for a week or two, but there’s already a trust factor here.” Officer X invites me to point to a boy at random. I choose a hefty boy with soft features. “Rodriguez, stand up!” X commands. “Why do you want to be here, Rodriguez?” Awkwardly, but still audibly, Rodriguez, in on a drug charge, answers: “To try to learn discipline, to try to change. I came here saggy, I came here with a temper, but if you get jail time, this is the place to be.” Another boy is summoned to stand. “Sharif, tell me something about maturity, brother!” Sharif is awaiting sentence on an attempted murder and armed robbery charge. “When I came here, I was running around, but Brother Francis told me what I was doing was wrong and how to humble myself to others. Hopefully, I’ll take Brother Francis’s teaching when I go upstate. When I come out, I’ll be an adult.”

In the next wing, a laid-back female officer, Anisah Watkins, leads a rambling discussion with young inmates. A few lay their heads down on their folded arms—the classic ghetto gesture of classroom indifference—while another is lying on his bed. In short order, the discussion tacks from what it means to grow up too fast (it means “you grew up with no father,” a boy suggests) to sexual relations between boys and girls (“sometimes the mom had babies just like her teen daughter”) to the need to be prompt and organized in life (here the only insights come from the officer). The boys push back against the implacable Watkins, complaining that the regimen in the Institute for Inner Development is too strict. She calmly reminds them: “That’s how it’s gonna be down here; it’s not going to change.”

“We’re trying to take the urban street that’s inbred in the detainees and file it down, soften it a bit, in order to make them productive members of society,” says Officer Alfonzo Miller, a tall, composed man standing in a glass vestibule overlooking another Institute for Inner Development dorm. Still, Rikers officials stress that the main purpose of the Institute is maintaining order, not reducing recidivism when detainees get out. It has accomplished that goal, lowering the fight rate in what was once the jail’s most violent area. One boy ejected from Officer X’s wing for violence now wants back in. Patrick, a diminutive 17-year-old gun defendant, presses his face against the light-green bars of his cell in the Administrative Segregation Unit and calls Warden Duffy over. As a gesture of courtesy, Duffy opens the food-tray door and starts to bend down toward it (some inmates prefer to speak through the open slot, where there is no screen). Patrick remains standing, however, and asks Duffy in hushed tones if he can get back into the Institute. “I want to get my GED. It’s hard to focus here. I need structure,” he says. “With a high school diploma, you can get a good job.” Duffy says that he doesn’t know if there’s still room in the program but that he’ll look into it.

Direct-supervision theory is evolving further in the direction of community-policing concepts. “Place-based management,” for instance, teaches officers to think of themselves as owners of the housing units they supervise. They’re accountable for everything that happens on their watch, just as a community-policing officer should feel responsible for what happens on his beat. Too often, says Demory, who trains jail officials in the concept for the National Institute of Corrections, “if a fight breaks out, the officer thinks, ‘They’re just fools,’ rather than, ‘I’m the manager. How did that happen?’ ”

Very few jails offer rehabilitation-oriented programming for the pretrial population because the length of pretrial detainees’ stay is unknown and usually brief. Merely getting them safely to and from their legal appointments on time consumes most jails’ attention and budgets. However, the Westchester County jail, a redbrick hilltop facility north of New York City and surrounded by geese, recently began an antiviolence program for adult pretrial detainees that combines direct supervision with intense group therapy.

The 44 inmates in the Resolve to Stop the Violence Project (RSVP) are all awaiting trial on murder, rape, robbery, and other top felonies or have just been sentenced and await transfer to prison upstate. On Monday mornings, they sit in a large oblong circle in the light-filled atrium of their two-tiered cell block, arms crossed, orange-jumpsuit-clad legs splayed out in front of them, paying close attention to an unusual conversational ritual. Pat, a designated inmate leader, begins. “I heard you say, Fred, that I discounted you. You said you were angry. Can I ask you a clarifying question?” Fred’s response is less than clear, but Pat presses on: “Can I propose an agreement that we support each other and let us nurture each other?” Another man chimes in: “Can I validate that I heard you say that you propose an agreement that you support each other?” A third inmate picks up: “I support you validating the agreement that they support each other,” and everyone claps.

The men are performing their “one-on-ones,” public interactions between two inmates who bring up any tense moments they’ve recently had with each other, discuss what they learned the previous week about controlling and expressing their emotions, and announce what they intend to learn as the day progresses. The conversational conventions would undoubtedly be familiar to group-therapy habitués, but without such therapy experience, the self-referentiality of the discussion seems like a madly literalistic enactment of social-contract theory.

The participants turn to the future. “Now can I give you some feedback, Greg?” the leader starts. “My kernel of truth is that I’m excited because I’m expecting a visit on Saturday.” Greg responds: “Pat, can I give you some feedback? Would you like me to validate it? I heard you say that your kernel of truth is that you are happy because you are expecting a visit; my kernel of truth is that I’m happy, too.” Pat replies: “I heard you say that you heard me say that I’m asking for validation; however, you said . . .” Some members of the ring aren’t paying rapt attention to this exchange. A facilitator rebukes them: “Is that respectful that you are talking while these brothers speak?” The inattentive detainees soberly apologize.

That men with a criminal inability to control violent impulses can participate in this artificial ritual, known as “conscious languaging,” with seeming sincerity says a lot about human malleability. What will remain of these conversational mannerisms when they collide with the code of the streets is unknown. But program director Eddie Concepcion says that the inmates are already bringing their new language to court. “A detainee will say to the judge: ‘I hear you say that if I take this plea, I can go home. Can I ask you a clarifying question?’ The court personnel are blown away. Usually these guys are cussing out the judge.”

RSVP aims to train the inmates to think twice about their reactions before acting on them and to take responsibility for everything that they do. If a participant even hints at blaming drugs, alcohol, or anger for his past violence or a present conflict, program leaders will call him out. “Every day, we remind them that they are violent and are responsible for being violent,” Concepcion says. The Westchester RSVP is too new to have produced reliable post-release outcomes yet, but the rate of infractions among its members is a small fraction of those among the inmate population at large. Graduates who have been sent upstate to prison have requested the RSVP manuals so that they can start their own groups there. And the San Francisco prototype of RSVP does boast a significant impact on recidivism: the one-year rearrest rate for violent offenses dropped 82 percent for participants who spent 16 weeks in the program; participants who spent less time in the program had higher rearrest rates.

The challenges of running jails exceed anything that the academic world—and most of us—can begin to understand. In addition to the huge problems of logistics and safety that jails present on their own, commissioners also face a well-organized inmates’ rights lobby that fights commonsense antiviolence measures. Until recently, for example, New York City officials weren’t allowed to put pretrial detainees in uniform, which made detecting contraband more difficult. Only last year did Commissioner Horn win the right to monitor detainees’ phone calls. Adolescents arrive at Rikers with their criminal histories largely concealed from officials to protect their privacy, hindering the determination of their security risk.

But the order that the lobbyists, academic critics, and neo-Foucauldians see as oppressive is inmates’ only hope for safety and even, perhaps, rehabilitation. The recent insights of urban policing—that order matters, that small violations lead to greater crimes, and that information must be gathered and analyzed—are all equally pertinent to jails, where chaos and corruption always threaten.

Research for this article was supported by the Brunie Fund for New York Journalism.


City Journal is a publication of the Manhattan Institute for Policy Research (MI), a leading free-market think tank. Are you interested in supporting the magazine? As a 501(c)(3) nonprofit, donations in support of MI and City Journal are fully tax-deductible as provided by law (EIN #13-2912529).

Further Reading

Up Next