David Grossman / Alamy Stock Photo

Pick a Number: The Facts and Politics Behind New York City Limiting the Jail Population to 3,300 People

Elizabeth Glazer

March 12, 2024

Reducing crime and incarceration requires information, discipline and follow-up

Reducing crime and incarceration requires information, discipline and follow-up

In early 2014, the newly elected mayor of New York City, Bill de Blasio, hired me to run the Mayor’s Office of Criminal Justice. It is a strange office. It does not have executive authority to move the pieces on the board, in the way that Sanitation is in charge of picking up trash or the housing agencies building shelter. In fact, despite the lofty name of “Mayor’s Office,” it wields no authority over, well, anyone, and relies largely on the power of convening and cajoling. 

Perhaps that limit is inherent in the question of whether the criminal justice “system” can be organized at all, since it is less a system than a collection of agencies that have no single boss. Some are City agencies, some are state agencies, some have bosses who are independently elected, and some have a separate duty entirely, not to reduce crime or incarceration, but to defend their clients zealously. In the past, the City had intermittently had a deputy mayor overseeing criminal justice, but that had fallen away in favor of a “coordinator’s” office, reflecting a fundamental ambivalence about who is in charge of keeping New Yorkers safe. Is it entirely the province of the police — the police commissioner, one of only three agency heads who traditionally report directly to the mayor? Or is the enterprise of safety something bigger? 

Reducing the number of people in jail? Impossible!

With no small lack of humility, I tried to figure out how the office could perhaps be, as my colleagues and I joked, the “hidden hand,” helping to organize that bigger effort across many agencies. Reducing the jail population was at the very top of the list, first, because it is a fundamental issue of justice that the fewest number of people, consistent with safety, be deprived of their liberty, especially when the overwhelming majority of those in jail are awaiting disposition of their cases and presumed innocent. But also because how many people are being held is a bellwether of how the rest of the system is working, or not. Who goes in is a measure of how the discretion of police and prosecutors is wielded, and how long people stay in jail is an indicator of whether the works of justice are gummed up or moving with efficiency and fairness. 

My office’s effort to map out how to provide the most safety with the least amount of incarceration was made up of a million little pieces. Here is the story of two, one a success and one a failure.

Each week I met with the first deputy mayor, Tony Shorris, who was in charge of the operations of the city, among many other tasks. An urbane man, deeply experienced in the running of government agencies of which he had had command of a few, he was both inventive and practical. He lasered his way through the material I brought. When he got to the jail charts, he tapped his finger on the trend line that was showing modest declines in population and said softly: “Here, why can’t this come down, say, 500 or 1,000 people? After all, crime is down.” At that point in 2014, there were about 11,500 people in jail. A 10% reduction in a year? Impossible!

I had many reasons for thinking this. Almost nothing was in our control. How many people entered jail depended on the strength of the evidence the cops arrested on, on the inclination of prosecutors to charge and judges to release. Who entered, why and in what number reflected how higgledy-piggledy the “system” was. There were a high number of admissions for low-level offenses. Many of these cases had commensurately low levels of bail, signaling that judges didn’t think that they should be in jail at all. The bail system, widely used as a proxy for detention, was swiss cheese: 75% of those who made bail made it within a week, and it didn’t matter whether the bail was high or low. 

At the same time, many cases languished in court for unconscionably long periods, with all sides pointing fingers at one another. Prosecutors blamed the judges for letting deadlines slide and for working partial days or weeks. Defense attorneys blamed the prosecutors for using delaying tactics to squeeze pleas out of their jailed clients. Court administrators blamed defenders for dragging their feet, hoping that witnesses’ memories would fade. It was a quagmire of dysfunction.

Puzzling over how to break through these dueling accusations, I retreated to my office to huddle with Michele Sviridoff, the brilliant head of research, who by some miracle had agreed to stay on in the new administration. The first time she modeled what would need to happen to cut the jail population in half we just laughed because it seemed so impossible. Still, bit by bit, the team of gifted and dedicated researchers that we had started to hire began to take apart the “system,” turning each piece in the light and sketching out what was possible. That became the discipline — to try to model what the “path to 10,000” might be, then the path to 7,500, to 5,000. And finally 3,300. Initially, the “reduce by half” modeling was simply exploratory since the jail population at that point stood at half what it had been at its peak. But as various diversion and other programs began to produce results, the forecasts integrated those results as well.

As we parsed the data, we learned what in fact affects the size of the jail population and what does not. Crime, yes, to a degree, but not every crime leads to a jail admission. In fact, only about 15% did. Marijuana arrests and criminal summonses? Not a whit. The tens of thousands of people churning through for a few days at a time on minor offenses? A lot of churn, a lot of disruption, a lot of injustice, but not such a big effect on the number of people in jail at any given time. 

open in new tab
Download: Data, Chart Image

On the other hand, permitting people who had been sentenced to jail (where the maximum sentence is under a year) to serve their sentence outside of jail made quite a bit of difference. Having state agencies, such as parole, who use the jails to hold people but have their own process to adjudicate them, take responsibility for these individuals made a very big difference. The holy grail of shortening the time in custody by making the system operate competently would make an enormous difference, but how would we unlock it?

The first step was to build on the successes of the Bloomberg administration. Judges often put people in jail because there were few options between jail and release. Recognizing this, Bloomberg had started a modest pilot in two boroughs called supervised release. It permitted defendants to go home but to have some tether and some support that could help ensure their return to court. Judges welcomed this in-between option and the program seemed to be having some success.

Safely reducing the number of people in jail? Possible! Step by step

Tragedy opened the door to opportunity. At the beginning of 2014, Jerome Murdough, a homeless veteran who struggled with mental health issues, had ended up at Rikers after police found him sleeping in a stairwell in a public housing development. Unable to pay the modest bail, he died in an overheated cell. The able and talented deputy mayor for human services, Lilliam Barrios-Paoli, and I assembled a task force with representatives from every stage in the system. Over a 100-day period, 400 people participated in mapping out who goes in and why, how long they stay, what happens when they leave. The result of the 100 days was a set of investments aimed at improving the complicated interplay between mental illness and criminal justice.

One of the key recommendations was to expand supervised release citywide, an expansion that was largely funded by then-Manhattan District Attorney Cyrus Vance. Vance had recently obtained an enormous judgment in asset forfeiture fees which he was investing in key city improvements. Between 2013-2018, supervised release, along with other forms of pretrial diversion, contributed to 34% of the jail population reduction, even as crime continued to fall.

I felt we were making real progress but I got a reality check when I met with the police commissioner, Jimmy O’Neill, to lay out the plan. An outside observer might scoff that in New York City, the differences of opinion in criminal justice seem to run the gamut merely from A to B compared to the national scene or some other cities where battlelines are sharply drawn. But it turns out that even that modest distance can be enormous, leading to ferocious fights. Why? Because that small distance is everything. 

Even people who share the same overall goals can disagree about how much risk should be taken in deciding whether to hold or release people who have been arrested. If every single person was locked up, risk would be reduced to zero. But that would be extreme, unfair and ultimately damaging. Where should we draw the line between “everyone should be out” and “everyone should be in”? 

Figuring out how people could be released safely while their cases went through the system, helped reduce one big driver of the jail population: how many people go in. My colleagues and I were less successful with the other big driver: how long people stay.

The Police Commissioner didn’t like where the line was being drawn with supervised release. I had thought the plan was reasonably conservative: No people charged with violent felonies were included, among other “carve-outs,” even though a strong case could and would be made for this, since how much risk a person poses is not always the same thing as the offense with which they are charged. (In all these discussions, there is a hall of mirrors feel because while danger is top of mind for judges and other New Yorkers in thinking about how jail is used, New York State, uniquely in the nation, limits the bases on which judges make release decisions. Judges must consider the likelihood the defendant will return to court but not the danger posed.)

O’Neill’s was not a momentary objection. Supervised release has continued to be a burr under the saddle for the Police Department under several different commissioners. They see any additional crime — and there were some terrible crimes — as a failure. How rare the crimes were does not vitiate how terrible. But the vast majority of participants were released without incident. There will always be risk with release, no matter where the line is drawn. The question is how much, how avoidable and how many people should we (the public, judges and others) be willing to incarcerate when those released and the community at large could be better off if they were at liberty.

Supervised release has become the backbone of pretrial release in New York City and has grown its enrollment from several hundred people in 2014 to approximately 10,000 today, and from covering only a few crimes to covering all crimes. That has tested the line drawing. Elsewhere in this issue, two people involved in both the day-to-day operations and planning of pretrial release programs note that much work still lies ahead to understand the limits of the program. But for now, as the demand grows among judges, it appears this option is here to stay. 

Reducing case delay? The holy grail

Figuring out how people could be released safely while their cases went through the system, helped reduce one big driver of the jail population: how many people go in. My colleagues and I were less successful with the other big driver: how long people stay. It had been front-page news in 2013 when the New York Times ran a searing account of the chaos in the system and the price it exacted on people’s lives. This was not a new problem. In the 1990s, the problem was also urgent, and for the same reasons. The court system periodically brought in a “swat team” of judges to clear backlogs. But sure enough, the backlogs, like kudzu, always grew back. 

I asked the then-chief judge of the New York State court system, Jonathan Lippman, if he would be willing to join forces and try to crack this problem. He was gracious and amenable but warned that judges are hard to corral because they are independent. Of course, judges must be independent in rendering their decisions. But, I wondered, what about accountability when it comes to the number of cases backlogged? One has to do with making judgements on the merits of a case without fear or favor, unpressured by outside forces. The other has to do with the nuts and bolts of operations, addressing administrative issues that contribute to the excruciating delays.

Was it possible to create a version of OpenTable for court calendaring that would help ensure that “reservations” for court appearances were not just made but kept?

Together we set up a system that put representatives of every agency that might affect delay in one room — judges, prosecutors, defenders and jail administrators, among others. Each borough’s effort was led by the administrative judge of that borough. And we brought data. Tons of it. Information many had never seen before, such as how the boroughs compared to one another and where the bottlenecks were. The process revealed a number of solvable problems, often laughably solvable but tragically unaddressed. For example, defendants weren’t getting to court in time, in part, because court officers didn’t have phone numbers for the corrections officers responsible for transporting the incarcerated people. As a result, often the court didn’t know that defendants had arrived and were waiting in the court building in holding cells. So the court conference would be adjourned and another month would be added to a defendant’s time in jail until a new date could be mustered.

Other things were much harder to fix. For example, the way the city courts are set up, everyone — defendants, their lawyers, prosecutors and witnesses — is required to be present at a “cattle call” assembly at 9:30 a.m. Then they wait. And wait. And wait for their case to be heard. Often, the case is not heard and the defendants are “not seen.” In fact, one study (on file with Vital City) showed this happened 30% of the time. 

Was it possible to create a version of OpenTable for court calendaring that would help ensure that “reservations” for court appearances were not just made but kept? In this day and age, where there is an app for everything and most everyone keeps their calendars online, why can’t technology help our courts become more efficient? Despite enormous effort by everyone involved, we failed to make much progress toward this vision.

Science and politics

Between 2013 and 2019, the Mayor’s Office of Criminal Justice made a deep investment in data to support jail reduction. Among the tools that the office pioneered was Jail Population Reduction-stat or JPR-stat. (At this point, CompStat has sired almost as many progeny as Watergate.) It was led by the blazingly smart and effective first deputy head of the office, a former prosecutor and someone who understood minutely the importance of both listening and incorporating ideas to reach consensus. JPR-stat brought the entire office together on a weekly basis to scrutinize who was in and for what reason. This information was then used to confer with the DAs, the defenders, the Department of Correction and the courts to move the cases.

The effort to reduce the jail population took on more visible importance as 2019 drew to a close. A movement to “Close Rikers” and build new, smaller and decent facilities in four of the boroughs had become law. How many people would these new facilities hold? The location of the jails had been a political choice, one that epitomized the art of second-best choices. Four sites were ultimately chosen in no small measure because the local council members in whose districts the jails would be built supported the plan. This was crucial because as a matter of comity, council members could veto the siting of a project in their home district. But the sites were, largely, in dense neighborhoods. The size of the building and thus neighborhood life would be affected by how the buildings were designed and importantly how many people they were intended to house.

My office had been making steady and careful progress in reducing the size of the jail population while crime dropped. We were active in developing, funding and implementing programs, and then meticulously following the results that we plugged into our increasingly accurate population projections. In 2017, when the jail population stood at a little over 9,000 and the plans to close Rikers were first in hot discussions, we calculated that we could reach a population of 5,000 by 2027, the then-proposed closing date, by continuing with the various initiatives including supervised release.

Then in April 2019, the state Legislature passed a sweeping law that essentially eliminated bail (and thus detention) for all misdemeanors and for most non-violent felonies. In one stroke, lawmakers reshaped who would be coming into jail with a much blunter instrument than our approach that relied on results and persuasion. Our researchers revised their modeling, now taking into account the impact the law would have once it took effect in 2020. Under our new calculations, 3,300 was possible, and depending on the scenario, was not even the lowest we might expect. In October 2019, the City Council approved a “Close Rikers” plan, planning for buildings that would hold not more than 3,300 people. 

To make progress that sticks depends upon the unsexy but critical work that unfolds day by day, with a relentless focus on improvement on operations over the long haul.

The City was making gradual progress toward 3,300 when the COVID-19 pandemic hit. For some years, all the parts of the criminal justice system, in one way and another, grumbling or glad, had been working together — on supervised release, case delay, gun violence and many other issues. Suddenly, in March 2020, the question of incarceration became quite literally one of life and death. It was clear that congregate settings were the most dangerous place to be as the fearsome and contagious virus raged. So it was critical to figure out how to release safely those most vulnerable, and at the same time,  create room inside the jails for social distancing.

Once again, district attorneys, defenders, courts and city agencies came together, looking case by case at every single person incarcerated and whether they could be released. At the same time, defenders were going directly to court to get “writs” from the courts for particularly sick clients. Every available tool was mobilized: a long-languishing provision of law that permitted the corrections commissioner to release some people to serve their short city sentence at home; accelerated focus on operations — in particular, the backlog in parole violation hearings — to release those waiting for an adjudication on charges of rule violations. Within six weeks, from mid-March to the end of April 2020, the number of people in jail dropped from approximately 5,400 to 3,809. All of a sudden, 3,300 was in sight. 

And then it wasn’t. In the first year of the pandemic, shootings doubled and have only recently started to decline. COVID slowed many pieces of the justice system to a crawl, and the jails’ population grew again. It now stands at about 6,000.

How can the work of improving justice — in this instance, the fewest people in jail with the most safety in the city — move forward at a steady and unyielding pace, even in the face of cataclysmic events? There is nothing like a crisis to mobilize government but an initiative or a task force will do as well.  Except what happens when the crisis is over, the initiative runs out of steam or the task force disbands? To make progress that sticks depends upon the unsexy but critical work that unfolds day by day, relentlessly focused on the improvement of operations. 

If there is any lesson to take from this history, it is this one: Someone needs to be in charge of bringing all the pieces together. It doesn’t matter who — it can be a mayor, a judge, a district attorney or someone entirely different. But in a fractured system with multiple moving parts and conflicting agendas, it is essential to have a person who can lead with respect and knowledge both to set a common goal and to help guide all of the various participants towards it. Otherwise it’s kudzu all the way down, constantly growing back.