When the Case Was Real and the Investigation Wasn’t: Five True Crime Cases Where Police Misconduct Surfaced After the Fact

The framing question that matters least in a misconduct case is whether the crime happened. The crime, in most of these cases, happened. That is not the question. The question is whether what the investigation produced — the confession, the lab result, the forensic conclusion, the charge — was real. Whether it was what it was represented to be. Whether the record holds.

In the five cases examined here, it does not. The misconduct was not invented retroactively by defense attorneys or advocacy organizations. It was documented — by judges who reviewed the record, by chemists who retested the samples, by an officer who eventually described what his unit had been doing for years, by a federal court that found no written documentation that a blood test was ever performed. The record is the source. And across five cases, spanning four decades and four different misconduct mechanisms, the record says the same thing.

Case 01: The Exonerated Five — New York, 1989

What the Record Shows

On the night of April 19, 1989, Trisha Meili was attacked while jogging in Central Park. She survived. The NYPD arrested five teenagers from Harlem within days. What followed was not an investigation. It was a performance of investigation, conducted in a city under enormous pressure to produce suspects, against five Black and Latino teenagers who had no adult legal representation during the hours that mattered most.

Four of the five boys were interrogated for more than 24 hours. Per the documented record, they were deprived of food, water, and sleep during those sessions. The interrogations that preceded the videotaped confessions were not recorded. There were no adults present — in most cases, not even parents — during the pre-tape sessions in which, the boys later said, the coercion actually occurred. By the time the cameras turned on, the boys were exhausted and had already been told, by investigators, what the evidence purportedly showed against them. The DNA evidence, in fact, showed nothing against them. No physical evidence linked any of the five to the crime. The DNA recovered from the victim matched none of them.

The prosecution relied on the confessions. The confessions were inconsistent with each other, inconsistent with the physical evidence, and inconsistent with the actual timeline of the attack — a fact established by the medical evidence about when Meili was struck. None of the five confessed to being the primary attacker. None described the attack in ways that matched each other or matched the documented injuries. The jury convicted them anyway. The city, the press, and the prosecution had told a story, and the confessions — coerced or not, accurate or not — were what made the story possible to tell.

In 2002, Matias Reyes, then serving a life sentence for a series of violent crimes including a rape committed near Central Park in the same period, confessed to the attack on Meili. DNA testing confirmed his account. He had acted alone. The five men had been in prison for crimes one of them did not commit, in whole or in part, for years before this came to light — not because of any internal review, but because Reyes chose to come forward.

The structural lesson here is specific: the coercive interrogation did not happen in secret. It happened in a facility controlled by the NYPD, documented — imperfectly, and only partially — by the department’s own records. What the record does not contain is any documentation of what happened during the unrecorded hours. The absence is the document. The record shows interrogation sessions. It does not show what occurred during those sessions before the cameras ran. That gap is where the misconduct lives, and it was designed to be unchallengeable.

Case 02: The LAPD Rampart Scandal — Los Angeles, 1995–2000

What the Record Shows

The LAPD’s CRASH anti-gang unit in the Rampart Division was not corrupt in the way that corruption is usually framed — a few bad officers cutting corners. It was corrupt structurally, across the unit, over years, in ways that its supervisors permitted and that its internal affairs function failed to detect until a drug theft arrest forced one officer’s cooperation.

Rafael Perez was arrested in August 1998 for stealing approximately eight pounds of cocaine from an LAPD evidence room. In exchange for a reduced sentence, he agreed to cooperate with investigators and provided testimony on more than 70 officers across the Rampart CRASH unit. What he described was not a collection of individual rogue incidents. It was a unit culture in which planting evidence was routine, in which officers bet on whether they could make a fourteen-year-old cry by applying physical pressure, in which supervisors were present or informed and did not intervene, and in which Perez himself said he knew that 90 percent of CRASH officers had framed innocent people and committed perjury.

The most documented individual case to emerge from Perez’s cooperation was that of Javier Ovando. In October 1996, Perez and his partner Nino Durden shot Ovando — who was unarmed — in an apartment building. They then handcuffed him to his wheelchair and planted a semiautomatic pistol on him to justify the shooting. Ovando was paralyzed from the shooting. He was convicted of assault on a police officer and sentenced to 23 years. He served three years before Perez’s cooperation led to the vacation of his conviction. The City of Los Angeles paid Ovando $15 million, the largest police misconduct settlement in Los Angeles history at the time.

The LAPD was placed under a federal consent decree following the Rampart scandal, requiring external oversight of department practices for years. All CRASH anti-gang units were disbanded. The consent decree was lifted in 2013. The structural conditions that allowed a unit to plant evidence, commit perjury, and shoot unarmed civilians over a period of years without internal detection were addressed in policy language. Whether they were addressed in practice is a question the consent decree’s lifting does not resolve.

Case 03: Annie Dookhan and the Massachusetts Hinton Lab — Boston, 2003–2012

What the Record Shows

The Dookhan case is not a story about one dishonest chemist. It is a story about what happens when an institution mistakes high throughput for quality, when supervisory oversight is absent, and when a forensic function that is supposed to sit between investigators and courtrooms is instead permitted to operate as an extension of prosecutorial goals.

Dookhan worked at the Hinton Lab for nine years. During that time, she processed more samples than any other chemist at the lab — substantially more, at a pace that her supervisors appear to have viewed as exceptional rather than suspicious. She admitted to dry labbing: recording results for samples she had not actually tested, or testing only a fraction of a batch and recording the same result for the rest. She admitted to tampering with samples to make negative results positive. She took calls from police officers and prosecutors informing her what drug they expected a sample to contain. She signed off on reports not only for her own work but for other staff members she was supposed to be supervised by, not supervising. She falsely claimed academic credentials to inflate her professional standing.

When a colleague raised concerns about Dookhan’s work in 2011, state police investigated and brought her conduct to the state Attorney General. She admitted to the misconduct during an interview with the AG’s office in August 2012. The lab was shut down by Governor Deval Patrick on August 30, 2012. Dookhan was arrested in September.

A separate but concurrent scandal involved Sonja Farak, another Massachusetts state chemist, this time at the Amherst lab. Farak was convicted in 2014 of using drugs she was supposed to be testing — consuming drug evidence from cases rather than reporting it. A 2017 Superior Court judge found that her drug abuse occurred at the Hinton Lab as well. The ACLU filed multiple lawsuits seeking dismissal of cases tied to both Dookhan and Farak. The cases are distinct in mechanism but identical in consequence: criminal convictions built on forensic certifications that did not reflect actual science.

Dookhan’s stated motivation, per her defense attorney, was to be the hardest-working, most productive chemist in the lab. The institution rewarded her for exactly that, right up until the moment it became impossible to ignore. That sequence of events is not incidental to the story. It is the story.

Case 04: Kalief Browder — New York, 2010–2013

What the Record Shows

Kalief Browder’s case does not fit neatly into the category of fabricated evidence or coerced confession. The misconduct here is procedural, and it is documented in the court record itself. The prosecution repeatedly told courts it was ready to proceed. It was not ready to proceed. More than 30 times, the Bronx District Attorney’s office requested continuances — postponements of the trial date — while Browder sat at Rikers Island, unable to post the $3,000 bail that would have allowed him to fight his case from home. He was sixteen when he was arrested. He was nineteen when he was finally released.

The Bronx’s court system in this period was, by documented accounts and court records, severely backlogged. That is a structural condition, not a secret. But the practical effect of a backlogged system on a defendant who cannot post bail is not the same as the effect on a defendant who can. Browder was offered plea deals repeatedly — 3.5 years, then 2.5 years — that would have given him time served and release. He declined, maintaining his innocence. His refusal to accept a plea deal for something he said he did not do was the factor that kept him incarcerated longer than the system expected.

The prosecution eventually revealed it had a single witness — a Mexican national named Bautista who had identified Browder as the person who stole his backpack. By May 2013, Bautista had returned to Mexico. The prosecution could not locate him. Without him, they had no case. The charges were dismissed on May 29, 2013. Browder had been in prison for 1,110 days for an accusation that the prosecution could never have supported at trial.

Browder died by suicide on June 6, 2015, two years after his release, at the age of 22. He had spoken publicly and at length about the psychological damage from his time in solitary confinement. His case became the primary driver of reforms to New York City’s pretrial detention system, including an end to punitive segregation for young people on Rikers Island. His death also became the basis for a $3.3 million settlement with his family, reached in 2019. The reforms carry his name. The institution that held him accountable for a crime it couldn’t prove, in conditions it would later prohibit, is the same institution that settled the case rather than defend the record.

The Browder case is included here not because the police fabricated evidence — the record does not show that — but because the misconduct was institutional and procedural, documented in the court docket itself, and because the consequence was as severe as any fabricated confession. Three years of incarceration, the majority in solitary, for a case that was dismissed because the sole witness could not be found. The record showed the system knew what it was doing and did it anyway.

Case 05: Dr. Henry Lee and the Connecticut Murder Convictions — 1985–2023

What the Record Shows

Henry Lee was, at the time of the Birch and Henning trial, the head of the Connecticut State Police Forensic Science Laboratory and one of the most credentialed forensic scientists in the country. He would later be retained as a consultant in the O.J. Simpson case, the JonBenet Ramsey investigation, and the Scott Peterson trial. His name carried institutional authority. That authority is the mechanism through which the misconduct in this case worked.

Ralph Birch and Shawn Henning were convicted of murdering Everett Carr in 1985. The crime scene included more than 40 fingerprints and hair samples — none matched either defendant. No blood was found on their clothing or in their vehicle. Lee testified at trial that a towel found in Carr’s home showed stains consistent with blood, and that this established the killers had used the towel to clean up after the attack. His testimony provided the forensic anchor for a conviction that had no other physical evidence linking the defendants to the crime.

Decades later, the towel was retested using modern techniques. It was free of blood. There were no bloodstains on it. Post-conviction testing established what Lee’s testimony had asserted as fact was not fact at all.

Birch and Henning’s convictions were vacated in 2020. They filed a federal lawsuit against Lee, eight police investigators, and the town of New Milford. In July 2023, U.S. District Judge Victor Bolden ruled on Lee’s portion of the case on summary judgment. His ruling is worth quoting in its essential finding: the court found no evidence that Lee had ever performed the blood test he testified about. His own experts, in the litigation, concluded that there was no written documentation or photographic evidence that the TMB blood test was conducted. There was also evidence in the record that the tests actually performed did not indicate the presence of blood.

The Lee case is significant to the broader analytical frame here for a specific reason. In the Rampart scandal, the misconduct came from within the investigative function itself. In the Dookhan case, it came from within the forensic support function. In the Exonerated Five case, it came from the interrogation room. In the Lee case, it came from a credentialed outside expert whose institutional authority was specifically the reason his testimony carried weight with the jury. The trial presented Lee to the jury as a man whose forensic conclusions could be trusted. The federal court, thirty-eight years later, found that a central forensic conclusion was not supported by any documented evidence that a test was ever performed.

Expert witness testimony in criminal cases is structurally difficult to challenge at trial, particularly when the expert’s credentials are presented as a guarantee of reliability. What the Lee case documents is that credentials are not documentation. The question is not whether a test was claimed. The question is whether it was performed, and whether the results were what the testimony represented them to be. In this case, the record could not confirm either.

What These Five Cases Have in Common

Each of these cases involves a different institution, a different jurisdiction, a different mechanism of misconduct, and a different set of actors. What they share is the thing that makes them useful to analyze together: in each case, the misconduct was not detected by the institution responsible for detecting it. It surfaced through external pressure — a plea deal in Rampart, a post-conviction DNA match in the Exonerated Five case, retest results in Dookhan and Lee, a public records environment that eventually documented Browder’s detention timeline clearly enough that it could not be defended.

Internal oversight did not surface any of these. The systems responsible for self-regulation — internal affairs, supervisory oversight, lab quality control, prosecutorial ethics review — were either absent, insufficient, or captured by the same institutional culture that produced the misconduct. This is the pattern the record shows. Not that bad actors appear in otherwise functional systems, but that the systems failed to detect the bad actors until the cost of concealment exceeded the cost of disclosure.

The other shared element is accountability asymmetry. The people who were harmed — convicted on false forensic conclusions, jailed for years without trial, framed by planted evidence — bore the full consequence of the misconduct. The people who produced the misconduct, in most cases, did not. Dookhan served approximately three years for nine years of fraud touching 24,000 cases. The NYPD detectives who interrogated children for more than 24 hours without consistent adult supervision were never charged. Lee settled civilly and continues to hold an academic position. The LAPD officers who planted evidence on a man they shot and left paralyzed — one of them served five years, for cocaine theft, not for the framing.

There is a version of this observation that becomes an argument for systemic despair. That is not what the record supports. What the record supports is more specific: that institutions which protect their own records from external scrutiny, that treat internal discipline as an administrative matter rather than a public accountability function, and that structure plea deals, confidentiality agreements, and civil settlements in ways that resolve liability without requiring admission of misconduct — those institutions are producing conditions under which what happened in these five cases continues to happen. The record is already saying so. The question is whether anyone is reading it.