On September 16, 1992, 10,000 protesters descended on City Hall. They blocked traffic for the better part of an hour, climbing over cars, buses, and police barricades. Some were violent and inebriated, and a few physically assaulted members of the press, as others hurled racist epithets at New York’s first African-American mayor, David Dinkins. They eventually burst through barricades into the City Hall parking lot, much to the indifference of the 300 uniformed police officers there to oversee the demonstration. The protesters were off-duty cops.

Bused in by police unions and egged on by would-be mayor Rudolph W. Giuliani, they were indignant over that day’s heated hearing on a bill supported by Mayor Dinkins. The bill was designed to establish an independent civilian agency providing oversight of police, at a time not too different from today, when unrelenting police brutality was the subject of both weekly headlines and unyielding protests. The agency was pushed for by a “rainbow coalition” of community groups, civil-liberties agencies, and City Council representatives.

In the wake of the September 16 police riot, public opinion turned more vehemently against the police. The bill, which had been six votes shy of passing the day before the police stormed City Hall, was passed by a substantial majority the day after. It established an independent Civilian Complaint Review Board, which began its work a few months later. It was the first significant independent entity in the city to oversee civilian complaints against police. Supporters were euphoric.

But euphoria would turn to disappointment and the CCRB pass from a dream to a nightmare. For all the hope the board’s creation generated, supporters soon realized that “passing legislation [had been] relatively easy,” as Norman Siegel, the director of the NYCLU for 15 years and one of the key instigators behind the CCRB told The Nation: “The real challenge was implementation.”

Hope of civilian oversight soon withered with the 1994 election of Rudolph Giuliani, who had been so opposed to the board’s implementation. The board was firmly placed under mayoral control: All the CCRB’s decision-making powers lie with 13 persons, three members of which are appointed by the police commissioner, five by the mayor, and five by the City Council, putting the majority vote in the hands of the same city administration that is in charge of policing. Giuliani packed the board with former prosecutors, commonly believed to be sympathetic to the police. During Giuliani’s tenure, several CCRB employees and some board members resigned in protest of board decisions not to pursue incriminating investigations.

Four years into the Giuliani administration, the mayor’s staunch opposition to “any external check on police conduct,” as an NYCLU report emphasized in 1998, left the agency so understaffed and underfunded that it “virtually ensured it would not provide the oversight called for in the city charter.” While the city reeled from news of Abner Louima’s torture and sodomy at the hands of NYPD officers, the CCRB revealed a month after Giuliani’s reelection that it had previously undercounted the number of complaints against police filed in the aftermath of Louima’s case. Police accountability seemed far away. As recently as 2007, according to the latest NYCLU report on the CCRB, the CCRB had declined to investigate 55 percent of complaints filed in the preceding four years, and the investigations that did occur were so delayed and cursory that evidence disappeared and statutes of limitations expired.

But it wasn’t just the mayor’s grip on the CCRB that stymied oversight. The structure of the board severely limits its ability to punish police for infractions. The board can only recommend sanctions to the NYPD’s police commissioner, which he imposed at his discretion. In the majority of cases, he simply declined. The NYPD dismissed 63 percent of the CCRB’s findings of police officer misconduct between 1998 and 2004, or instead endorsed discipline that was, according to the NYCLU, “little more than a slap on the wrist.” As a result of the distorted proceedings, New York was found by the NYCLU to have one of the lowest rates of substantiated complaints of police misconduct nationally, at a rate of 5 percent.

At times when the police have challenged the CCRB’s authority, the board has remained suspiciously silent. This was the case in the late 1990s when the NYPD announced it would reinvestigate CCRB complaints to review their findings, insinuating that the CCRB investigations had been badly conducted, in a blatant violation of the purview of the civilian review board. The CCRB as an institution did not protest even when public advocate Mark Green reported in 2000 that this “reinvestigation” had uniformly meant rejecting CCRB findings with no investigation whatsoever. The board almost never called out publicly the uncooperative nature of the NYPD, let alone its crimes, barely commented on its policies, nor did it have its own internal processes reviewed, becoming complicit in absolving most cops from wrongdoing. As the NYCLU report concludes, this “extraordinary deference to the NYPD” could only be seen as a “capitulation.” Recent documents and interviews reveal that much of this paralysis remains today.

* * *

On July 17, 2014, 22 years after the CCRB’s creation, Eric Garner died after being placed in a chokehold by NYPD officer Daniel Pantaleo. Pantaleo had been the subject of at least one CCRB complaint—one complainant has come forward, though files were never officially released—and never terminated in response (though the NYPD itself had disciplined him for other violations). On the same day, in an attempt to address the national policing crisis, Mayor Bill de Blasio appointed Richard Emery to clean up the compromised CCRB. Emery, a civil-rights lawyer famous for winning millions of dollars for plaintiffs in a court settlement against illegal strip searches conducted at Rikers, promised to turn the board around and “start from scratch.”

Upon his arrival, Emery appeared to raise investigative standards: After a six-month effort to clear dockets overcrowded by lingering cases, the time that it took to process cases decreased 80 percent in a year. The board also created a field investigative unit that visited the scene of alleged excessive force incidents within 24 hours and collected evidence before security footage could be erased. The substantiation rate (cases where misconduct is found to have occurred) and the number of actually investigated complaints rose. Administrative prosecutions were, starting in 2012, brought forth by the CCRB instead of NYPD, although the trial was presided over by the NYPD and the police commissioner had the power to determine the level of discipline, if any. These were touted as successes by the CCRB’s new leadership. Still, CCRB recommendations remain subject to the NYPD’s discretion.

Community members and activists across the city were skeptical of Emery’s progress and of his appointment: His work with police commissioner Bill Bratton in the past, and the fact his son was working as Bratton’s policy adviser, suggested a relationship too close for comfort. Emery also fundraised for de Blasio, Bratton’s boss.

A closer look at the CCRB’s reported data shows that the CCRB’s increased efficiency under Emery came with a compromise: leniency. Under Emery, the CCRB saw a 24 percent yearly increase in pushing complainants to accept mediation, an in-person meeting between police officer and complainant. The option is first offered to complainants in 41 percent of cases and tends to shield police officers from retaliatory action for misconduct, since according to the CCRB “mediation lets you resolve a complaint by talking with the officer face to face, rather than seeking an investigation. It is a process that does not lead to discipline for the officer, but can often result in mutual understanding between the civilian and the officer.” During mediation, the complainant and police officer meet to discuss the incident, and pledge that what is said is confidential so can’t later be used in court against the police officer. As Josmar Trujillo, head of New Yorkers Against Bratton reports, they are “essentially handshake agreements with cops [which] allow the agency to check off investigations from their backlog of complaints, claiming efficiency.” It is also unpopular with complainants: Fifty-seven percent reject the proposition.

The CCRB reported in 2015 that the NYPD has accepted its penalty recommendation more than ever before, disciplining 92 percent of CCRB recommended cases compared to 70 percent in 2014 and 57 percent in 2013. But the CCRB’s disciplinary recommendations stand at an extraordinary level of lenience, as they recommend that a lower percentage of cases be charged compared to preceding years. This indulgence could explain the NYPD’s apparent surge in cooperation. The NYPD itself tends to mete out only lenient punishments: In 2015, only one police officer received one of the top three sanctions (which apart from being terminated, which rarely happens, involve dismissal or probation or losing 21 to 31 vacation days), as opposed to six in 2014 and 11 the year before. It has also been three years since an officer was terminated, though charges routinely include serious, sometimes deadly force: In 2015, even the CCRB established that 19 police officers had committed chokeholds (a massive increase compared to recent years), and that 168 cases of force had occurred, while the NYPD administrative trials found 58 police officers guilty after trial, not one officer was terminated, a chilling fact because, as for Daniel Pantaleo, the abuser of today can be the killer of tomorrow.

Ultimately, the number of police officers disciplined amounts to a drop in the bucket: Of the 5,234 yearly complaints against police officers processed by the board, a paltry 246 are eventually disciplined. The discipline itself was mostly symbolic: In 2015, more than half of the police officers were sentenced to some form of additional training. All but one of the others, were subjected by their commanding officer to sanctions ranging from “an oral warning and admonishment to a forfeiture of up to 10 days of vacation or accrued time”; though there are two levels of seriousness of command discipline, the sanction received is not clearly reported and is at the discretion of a police officer’s command officer.

Numbers from their 2015 annual report also reveal CCRB’s apparent refusal to recommend discipline in cases that involve physical force. The substantiation rate for cases involving use of force against civilians has remained at an extreme low, of 3.2 percent from the years 2011 to 2015. (Force cases amount to 49 percent of the complaints made to the CCRB.) In contrast, the total substantiation numbers for complaints not involving force stands at 12 percent. The more serious the crime, it seems, the less likely it is to be punished.

But this is not accidental. As a former CCRB investigator who would prefer to remain anonymous told me, “Institutionally, [there is] definitely a bias in favor of the police department: no question, very clearly.” As another former investigator who left before 2015, and who preferred to remain anonymous summed up, “Let’s figure out how to invalidate this whole thing: That’s what the training was.”

Reached for comment, a CCRB spokesperson stated that CCRB investigators go through an extensive training curriculum to ensure that they are ready to perform their duties. In 2015, he said, the Investigations Division implemented a comprehensive CCRB Training Academy with competency-based graduation requirements for the first time ever in the agency’s history.

Three recent investigators I spoke to described laziness that led some investigators to dismiss cases rather than investigating them. They also all point to pressure to unsubstantiate cases originating from a couple of the CCRB’s managers. One of the aforementioned former CCRB investigators tells me, “If something was leaning pretty strongly towards a substantiation, and if your supervisor was pro-cop, you would get pressure to unsubstantiate it” by finding “discrepancies” inherent to every case and using them to discount a complainant’s testimony.

But much of this institutional bias seem to lie within the CCRB’s very investigative patterns, which sheds a new light on the fact that 55 percent of cases at the CCRB were “unsubstantiated” in the past five years—meaning that the CCRB deemed there was insufficient evidence to determine whether or not the misconduct occurred.

This becomes less surprising when seen in light of the fact that the CCRB relies predominantly on the police to conduct its investigations, conducting seven in 10 of their investigatory interviews with police officers in 2015, compared to only one out of four with complainants themselves. Though the CCRB has non-binding subpoena power over security footage and medical records, a substantial portion of the investigation depends on police records. In virtually every other justice venue, relying solely on the accused for an investigation would be seen as a laughable breach of justice. This is even more problematic given the possibility of police officers’ false statements—the board noted 60 false official statements last year, caught thanks to an obvious contradiction between a police officer’s testimony and video footage or other witnesses, but more likely exist. Cases where officers lie under oath are classified as miscellaneous, making it hard to actually determine whether the false statements resulted in any sanction.

In contrast, the general skepticism towards complainants’ testimony might in part be responsible for the high number of cases on which the board doesn’t follow through: As the founder of police watchdog group El Grito de Sunset Park, Dennis Flores, reports, some investigators seem to second-guess those who report police misconduct and sometimes try to get complainants to give statements without their lawyers, which “adds another layer of pain to the whole experience.” He opened his own 2002 CCRB case after police officers pepper-sprayed him, “struck [him] in the head with a radio…causing a laceration to the back of his head that was later closed with four stitches” (according to Flores’s CCRB report), and, according to Flores, the NYPD then tampered with 911 recordings of the event and lied under oath in the administrative trial. The trial ended in a recommendation of police officer termination by the CCRB. This recommendation was overturned by then-Police Commissioner Ray Kelly.

Further accusations of shielding the NYPD have emerged as well, in a 2014 lawsuit brought by former executive director of the CCRB Tracy Catapano-Fox against what she alleges was Emery’s project to systematically discard stop-and-frisk complaints and the CCRB’s practice of knowingly publishing inaccurate data. The lawsuit was eventually settled out of court, mostly based on another claim. Emery eventually stepped down from the agency on April 13, 2016, a day after he was accused of sexual harassment by another member of the board, Mina Malik. He was replaced by Maya Wiley, the City’s Minority/Women Owned Business Enterprise Director and a de Blasio advisor, this July.

* * *

Police misconduct has been especially difficult to correct because police discipline records are protected by law. In the state of New York, Civil Rights Law 50-a prevents individual police officers’ records from being released to the public, a “unique protection” that “no other state provides” according to the Committee for Open Government. The NYPD did lobbying to influence the law at its inception. But a 1999 precedent, Daily Gazette Co. V. City of Schenectady, clarified that records could be undisclosed only if the documents demonstrated a “substantial and realistic potential…for the abusive use against the officers.” Still, the NYPD has often abused 50-a when faced with requests to disclose documents. Just this September, it announced its own interpretation of 50-a, declaring that, in fact, releasing any police discipline records was actually illegal. The department says that it will now cease to release disciplinary records altogether. The de Blasio administration has said it will advocate amending 50-a for greater transparency, but continues to fight in court to keep Pantaleo’s files sealed.

The CCRB’s files technically do not fall under the category covered by 50-a because the CCRB is an independent agency and should be released with the appropriate Freedom of Information requests—and for a brief time between October 2013–14, the CCRB released bare-bones information about cases to the public. But using 50-a from October 2014 onward and with the public support of Emery, the CCRB has prevented the release of records of substantiated charges against police officers from being released to the public. A lawsuit struggling for the release of CCRB files is currently ongoing, led by Eric Garner’s mother against the city in an effort to access Daniel Pantaleo’s past complaints with the CCRB.

Making CCRB files available to the public is crucial for true police oversight: In Chicago, where police personnel files are actually public, the Citizens Police Data project built a virtual database of police misconduct cases in the city with 56,000 misconduct complaint records for more than 8,500 Chicago police officers. It laid bare that the civilian review board discriminated against complainants of color, enabling much overdue oversight. These files could give people leverage over individual police officers in various other contexts as well. As Dennis Flores told me, one of the police officers who had physically assaulted him in 2002 and lied on oath arrested a young black man last year: Flores’s own CCRB files (CCRB complainants have access to their individual files) helped the young man’s public defender shed light on the police officer’s past history of violence and false testimony. The jury cleared him of wrongdoing.

This August, Bill de Blasio passed the NYPD Use of Force Encounter Reports law mandating that each precinct report the percentage and number of police officers with two or more substantiated investigations from the CCRB in the last three years. But this won’t mean anything if no one is individually or institutionally held accountable.

There are many ways in which to make civilian oversight stronger. Campaign Zero, a “data-informed platform [that] presents comprehensive solutions to end police violence in America,” lists some recommendations on its website, including a requirement that police chiefs justify each refusal to discipline. An effective civilian oversight board would require the power to issue sanctions to police officers directly; could issue meaningful sanctions like termination in certain cases of abuse; would enable special prosecution in the justice system at large and give the public access to misconduct files, making the process and outcome of complaints more transparent to civilians. More fundamentally, as Norman Siegel points out, true civilian oversight would require the board of the CCRB to include former victims of police misconduct or community representatives rather than solely mayoral, City Council, or NYPD appointees—or at the very least, like in Newark, representatives nominated by civil-liberties agencies like the NAACP and ACLU.

As of now, civilian oversight of police in New York is a good idea gone wrong. New Yorkers are instead taking police oversight into their own hands, aggregating videos and cases of police misconduct through the ACLU, the attorney database Cop Accountability Project, or through watchdog groups like Copwatch.

Many New Yorkers don’t deposit complaints with the CCRB anymore, a fact that is likely behind the recent drop in CCRB complaints—though the CCRB has touted it as a sign that police brutality is decreasing. As Norman Siegel told The Nation and Dennis Flores confirmed, “A lot of civil-rights lawyers go directly to court now, and no longer to the CCRB. And when people call me, I don’t recommend that they go to the CCRB anymore.”