The disinformation and falsehoods that can accompany breaking news online — involving terror attacks or national elections — have become a familiar plague in recent years. Big weather stories, it now seems clear, are not immune.
Meanwhile, Conservative News Today posted a Facebook Live video of a bus being toppled by Irma and a ship in enormous seas said to be carrying “hurricane chasers” heading into Irma. Neither was true. With help from social media, ProPublica tracked the original video of the endangered ship back to January 2013; it was shot in a terrible storm 60 miles or so off the coast of Portugal.
Over the last two weeks, there have also been a host of simplistic proclamations online about the role of human-driven climate change in Hurricane Irma and Hurricane Harvey, both overstating and discounting it. As Irma approaches Florida, the issues with sifting fake and real news become more consequential.
Here’s a condensed and edited version of our brief chat, followed by some reliable sources of information on Irma and other extreme storms.
Did you move to BuzzFeed specifically to cover mis/disinformation?
I did. We look at online mis/disinformation and how it spreads.
Does the flow of misinformation around weather events or climate change seem different in some ways to that around other breaking news?
In some ways, misinformation around weather is a little different. We see a lot of people share old photos or images seemingly by accident, without malice or a political agenda. During Harvey, many of the hyper-partisan and fake news sites that post false information were quiet during the first couple of days. That’s not the case during terror attacks or political events, for example. In those cases, the spin and misinformation come very quickly and each side tends to settle on a few key messages.
At the same time, what unites misinformation around weather and politics is emotion. If you scroll through fake news headlines, you’ll see that most of them want to inflame a sense of anger or injustice. Sometimes that means trash-talking a politician, other times that means playing on racial tensions. During Harvey, for example, we saw stories about Black Lives Matter blocking emergency crews from entering the city. BLM was actually helping hurricane victims, but the fake news went viral because it inflamed people’s emotions and biases.
Do you feel like you’re whistling in a Category 6 windstorm? In other words, do you feel it’s useful to attempt this?
Sometimes. With individual debunk posts, it’s unlikely that the person spreading misinformation will see me refuting it. At the same time, it’s an opportunity to teach the audience what to look for and how to identify misleading information. People appreciate having one go-to place in times of crisis. It’s good to see people use our posts to actively call out misinformation. That means some people still care about having a grip on reality.
Have you been fooled? I have, twice that I know of. The first time came in 2015, when I was writing a post hurriedly on the Nepal earthquake and included an embedded YouTube video of a sloshing hotel swimming pool that turned out to be from a previous Mexican quake. The second came August 27, when I joined those retweeting an image of planes submerged on a flooded runway — which of course turned out to be a bootlegged photo visualization of the impacts of sea-level rise from a 2013 Climate Central article. I was quickly rebuked.
I haven’t gotten fooled recently, but that’s because my trust levels are so low. If I read an article on an unfamiliar website, I immediately start checking for clues of it being fake — was the image stolen, was it registered recently, is it only masquerading as a legitimate source? My friends make fun of me for it, but it means I keep the fakes out of my feed.
What drives you to pursue this as a beat?
Misinformation can be dangerous. During a hurricane, misinformation can clog official messages from getting through. Sometimes it sways attitudes and public opinion. If people don’t know who to trust, they’ll put more faith in their own biases and that can be dangerous.
Who do you see, if anyone, as the competition on this beat?
This is a tough question because the field of misinformation is so wide. We’ve seen other mainstream publications start debunk roundups like ours, which is really great. But when it comes to the big projects (like this study we recently published) we’re unique because we dedicate reporters to uncovering online misinformation trends. Some of us do nothing else.
For those eager to sift reality from belief-affirming fantasy, there’s plenty of accurate online guidance on hurricanes and other meteorological hazards. On Wednesday, CrowdTangle, a company making platforms for news organizations or others tracking social media, created a dashboard with the social media flow from trustworthy sources on hurricanes.
To put current tropical storms in historical context, Phil Klotzbach at Colorado State University is one of the top experts. On Wednesday, he posted a summary of Irma superlatives including the remarkable fact that this Atlantic Ocean giant had, as of Wednesday night, already become the extreme-wind endurance champion for such storms, maintaining winds of 185 miles per hour for 33 hours.
This roundly beat 2013’s Typhoon Haiyan in the Pacific, the previous record holder, which held such destructive velocities for 24 hours.
In the 1960s, New York began to clear out its scandal-ridden psychiatric hospitals. In their place, a new system emerged. Thousands of mentally ill New Yorkers moved into “adult homes,” large apartment complexes concentrated mostly in New York City and its surrounding suburbs. The homes were meant to provide a safer, more humane alternative to the hospitals; they were closer to where many of the patients lived, and promised modest psychiatric care and other services.
But decades later, that grand vision had devolved into something that looked more like a nightmare.
In 2001, New York Times metro reporter Cliff Levy spent a year investigating conditions of the homes. He found that more than 1,000 people died in a six-year period. Some threw themselves off of rooftops. Others succumbed to extreme heat, only to be found days later, decomposing in fetid rooms. He found that the homes were often staffed by unqualified workers paid a pittance to look after a population in desperate need.
Today, Cliff is a deputy managing editor at the Times. He has joined us on this episode of The Breakthrough to discuss the 2001 series, “Broken Homes.”
He describes how he developed his own novel way of obtaining records of deaths in the facilities, and how he tracked down former workers who detailed schemes invented by the home’s operators to maximize profits. He tells us how he made cold call after cold call to reach the relatives of dead residents.
“It’s exhausting, and it’s really depressing,” Levy said in describing the effort. “And you ask yourself, like, ‘Maybe I’m just wasting my time.’ But then, at some point, you reach someone.”
The stories helped prompt a class-action lawsuit, which led to a federal court order requiring New York state’s Department of Health to move as many as 4,000 mentally ill residents into their own apartments, where they can live more independently with individualized services.
ProPublica is now examining that transition and the effort to improve conditions at the homes. Thus far, the state’s progress has been slow and controversial:
Earlier this summer, we reported that the Department of Health is behind in its deadlines to move the residents. We learned that a federal judge has accused the state of trying to evade the regulations at the heart of his order by colluding with industry. We spent parts of several weeks at a home called Oceanview Manor in Coney Island, where residents wander around outside the facility drinking malt liquor, begging for change and eating from garbage cans, looking ill and unkempt. Workers seemed outmatched, and the home’s owners declined to be interviewed.
We are looking to continue our reporting on this subject.
During the 2016 presidential campaign, President Donald Trump’s operatives bragged to the press that they tried to dissuade African Americans from voting by targeting them with Facebook posts titled “Hillary Thinks African Americans are Super Predators.”
If similar ads had appeared on TV, radio or in newspapers, journalists and advocacy groups would have fact-checked them. Negative ads in those media are closely monitored because historically they have influenced elections — most notably in 1988, when a television ad accused presidential candidate Michael Dukakis of “weak-on-crime” policies that enabled a furloughed prisoner named Willie Horton to commit rape.
The Trump ads may have been effective as well. But since they supposedly appeared on Facebook, nobody can say for sure if they ran, what they said or whom they targeted. Even though it’s the world’s largest social network, what happens on Facebook stays on Facebook.
The nature of online advertising is such that ads appear on people’s screens for just a few hours, and are limited to the audience that the advertiser has chosen. So, for example, if an advertiser micro-targets a group such as 40-year-old female motorcyclists in Nashville, Tennessee, (Facebook audience estimate: 1,300 people) with a misleading ad, it’s unlikely anyone other than the bikers will ever see those ads. Yesterday, 10 months after Trump was elected, Facebook officials acknowledged discovering that a Russian “troll farm” paid $100,000 during the campaign to place political ads on issues such as gun rights and immigration, The Washington Post reported.
With online ads, “you can go as narrow as you want, as false as you want and there is no accountability,” said Craig Aaron, president and CEO of Free Press, a public interest media and technology advocacy group.
ProPublica wants to change that. Today we are launching a crowdsourcing tool that will gather political ads from Facebook, the biggest online platform for political discourse. We’re calling it the Political Ad Collector — or PAC, in a nod to the Political Action Committees that fund many of today’s political ads.
We will begin using the PAC this month to track ads during the run-up to the German parliamentary election, which will be held on Sept. 24. The election has drawn international attention as a referendum on Chancellor Angela Merkel’s refugee policies, and a test of the strength of an anti-immigration party, Alternative for Germany (AfD).
We plan to monitor other elections, including the midterm elections in the U.S. In the U.S., information about politicians’ use of online ads is especially sparse because of loopholes in the campaign finance laws that allow candidates to report fewer details about their online advertising than about other types of advertising.
A trove of internal documents sheds light on the algorithms that Facebook’s censors use to differentiate between hate speech and legitimate political expression.
We are working with three news outlets in Germany — Spiegel Online, Süddeutsche Zeitung and Tagesschau. They will ask their readers to install our tool, and will use it themselves to monitor ads during the election.
The tool is a small piece of software that users can add to their web browser (Chrome). When users log into Facebook, the tool will collect the ads displayed on the user’s news feed and guess which ones are political based on an algorithm built by ProPublica.
One benefit for interested users is that the tool will show them Facebook political ads that weren’t aimed at their demographic group, and that they wouldn’t ordinarily see.
The tool does not collect any personally identifiable information, and we will not know which ads are shown to which user. The political ads that we collect will be contributed to a public database that will allow the public to see them all.
Facebook gives users more information about why a particular ad is targeted to them than other online platforms provide to their customers. Our tool will also collect that targeting information provided by Facebook, which may help illuminate what viewership the ads are trying to reach.
After the U.S. presidential election, Facebook launched its own transparency efforts. Facebook CEO Mark Zuckerberg has rolled out a series of initiatives to tackle fake news on its site. And although it doesn’t fact-check ads, Facebook does require advertisers to comply with the law, which includes prohibitions against deceptive advertising. This week Facebook said it had shut down the “inauthentic accounts” affiliated with Russia that had placed ads during the 2016 election cycle and is taking steps to prevent similar accounts from popping up in the future.
Still, more can be done to hold politicians, PACs and others accountable for the messages they spread online. We hope that by monitoring political advertising on Facebook, we can increase the transparency and accountability of elections around the world.
On Oct. 15, 2008, James Owens shuffled, head high despite his shackles, into a Baltimore courtroom, eager for his new trial to begin. Two decades into a life sentence, he would finally have his chance to prove what he’d been saying all along: The state had the wrong man.
Owens had been convicted of murdering a 24-year-old college student, who was found raped and stabbed in her home. Then he’d been shunted off to state prison until DNA testing — the scientific marvel that he’d watched for years free other men — finally caught up with his case in 2006. The semen that had been found inside the victim wasn’t his. A Maryland court tossed his conviction and granted Owens a rare do-over trial.
State prosecutors balked, insisting they still had enough evidence to keep Owens locked away and vowed to retry him. But they had also offered him an unusual deal. He could guarantee his immediate release from prison with no retrial and no danger of a new conviction — if he’d agree to plead guilty. The deal, known as an Alford plea, came with what seemed like an additional carrot: Despite pleading guilty, the Alford plea would allow Owens to say on the record that he was innocent. The Alford plea was an enticing chance for Owens, by then 43, to move on as a free man. But he’d give up a chance at exoneration. To the world, and legally, he’d still be a killer.
Owens refused the deal. He told his lawyer he wanted to clear his name, and he was willing to take his chances in court and wait in prison however long it took for a new trial to begin. It was a startling choice for an incarcerated defendant — even those with persuasive stories of innocence typically don’t trust the system enough to roll the dice again with 12 jurors or an appellate court. Most defendants, lawyers say, instinctively and rationally, grab any deal they can to win their freedom back.
The decision cost Owens 16 more months behind bars. Then, on that fall day in 2008, when the trial was set to begin, the prosecutor stood and, without a glance at Owens, told the judge, “The state declines to prosecute.”
In a legal gamble in which the prosecution typically holds the winning cards, Owens had called the state’s bluff. He walked out that day exonerated — and with the right to sue the state for the 21 years he spent wrongly imprisoned.
It seemed the ultimate victory in a city like Baltimore, with its deeply rooted and often justified mistrust of police and prosecutors. But Owens wasn’t the only man convicted of murdering that 24-year-old college student. Another white Baltimore man, James Thompson, had also been put away for life. Tests showed that his DNA didn’t match the semen either, but the state’s attorney’s office refused to drop the charges. Instead, as it had with Owens, it offered Thompson an Alford plea. Thompson grabbed the deal and walked out of prison a convicted murderer.
Same crime. Same evidence. Very different endings.
Ever since DNA ushered in a new era in criminal justice, even the toughest law-and-order advocates have come to acknowledge a hard truth: Sometimes innocent people are locked away for crimes they didn’t commit. Less widely understood is just how reluctant the system is to righting those wrongs.
Courts only assess guilt or innocence before a conviction. After that, appellate courts focus solely on fairness. Did everyone follow the rules and live up to their duties? Getting a re-hearing of the facts is a monumental, often decades-long quest through a legal thicket. Most defendants never get to start the process, let alone win. Even newly discovered evidence is not enough in many cases to prompt a review. And, for the tiny percentage of defendants who get one, the prosecutors still have the advantage: They have final discretion about whether to press charges and how severe they’ll be. Powerful influence over the pace of a case, the sentence and bail. And, compared with an incarcerated defendant, vast resources.
No one tracks how often the wrongly convicted are pressured to accept plea deals in lieu of exonerations. But in Baltimore City and County alone — two separate jurisdictions with their own state’s attorneys — ProPublica identified at least 10 cases in the last 19 years in which defendants with viable innocence claims ended up signing Alford pleas or time-served deals. In each case, exculpatory evidence was uncovered, persuasive enough to garner new trials, evidentiary hearings or writs of actual innocence. Prosecutors defend the original convictions, arguing, then and now, that the deals were made for valid reasons — such as the death of a key witness or a victim’s unwillingness to weather a retrial. The current state’s attorney in Baltimore County, Scott Schellenberger, said that “prosecutors take their oath to get it right very seriously” and wouldn’t stand in the way of exoneration if the facts called for it.
The menace of such deals, though, is clear: At worst, innocent people are stigmatized and unable to sue the state for false imprisonment, prosecutors keep unearned wins on their case records and those of the department, and no one re-investigates the crime — the real suspect is never brought to justice.
The plea deals ProPublica examined in Baltimore City involved two prior state’s attorneys. A spokeswoman for Marilyn Mosby, the current chief, didn’t respond to numerous requests for comment or for interviews with prosecutors in those cases.
The pleas in two of these Baltimore cases were later overturned after misconduct was uncovered in the original convictions, and the men won full exonerations. One, Walter Lomax, a black man convicted by an all-white jury shortly after the 1968 race riots in the city, served 38 years of a life sentence before taking a time-served deal in 2006. The state didn’t concede he was innocent until 2014.
Wrongful convictions are bad enough, Lomax said, but they’re even more “horrible when it becomes obvious the person is innocent and the state won’t at the very least acknowledge that.”
Some legal and cognitive science experts suggest that once detectives and prosecutors commit to a suspect and a theory of the crime, it changes how they evaluate evidence, and then the system itself exacerbates that focus at every step. Prosecutors are rewarded for proving and defending their theories, leaving little incentive to acknowledge weaknesses in cases, particularly in high-stakes crimes such as rape and murder. This mind-set is bolstered by one of the great positives of the system, one which legal experts, even those dedicated to exposing wrongful convictions, acknowledge: Prosecutors generally get it right.
Psychologists have a myriad of terms for the powerful, largely subconscious biases at play, but most people would call the collective phenomenon “tunnel vision.”
Wrongful convictions involving violent crimes typically involve poor, often minority defendants, sometimes with limited education or IQs, who are convicted on scant evidence or flawed forensics. The cases are fueled by an early theory of the crime that relentlessly drives the investigation and prosecution — even, in some cases, to official misconduct.
“At some point psychologically, you go from figuring out what happened to figuring out how to prove it happened the way you said it did,” Barbara O’Brien, a law professor involved with the National Registry of Exonerations at the University of Michigan, said. “It’s very difficult to take a step back from that.”
Marty Stroud, a former Louisiana prosecutor, made national headlines in 2015 when he penned a rare public apology for putting an innocent man on death row for 31 years. He told me recently that the system comes down hardest on those without the means to defend themselves. “It’s easy to prosecute those people and put them away and not think twice about it because no one is speaking for them,” he said.
The certitude of detectives and prosecutors hardens when their theory is validated by a judge or jury, and later, by an appellate court. Time, instead of allowing for fresh eyes, often makes biases worse. When a defendant like Owens gets a new hearing, the district or state’s attorney’s office — long committed to his guilt — has to re-justify that decision.
If they admit they got it wrong, prosecutors have to accept that a person was robbed of years of his life, the real perpetrator was never found, the victim’s family was let down, and, to top it off, they now have a cold case that’s unlikely to be solved. With the Alford plea, not only is the real perpetrator not caught but the case is officially closed on the books. It also dings their won-loss record on typically high-profile cases. The idea of a wrongful conviction, Stroud said, assaults a prosecutor’s sense of identity that “we’re the good guys. We have the white hats and are putting the bad guys in jail.”
Exonerations are also like a Pandora’s box in two important and unsettling ways. First, looking closely at why wrongful convictions happen — even in cases when everyone worked in good faith — could force a reckoning about deeply held beliefs on what is required to solve and punish crimes. False confessions, for example, often are a result of time-honored, and perfectly legal, tactics to soften up a suspect, such as lying or conducting questioning in the dead of night, said Steven Drizin, the former director of Northwestern University’s Center on Wrongful Convictions. When wrongful convictions are a result of misconduct, there could be a string of other bad convictions connected to that prosecutor or detective.
It’s no coincidence, many defense lawyers across the country say, that when misconduct comes up, prosecutors are quicker to propose an Alford plea or similar deal, effectively quashing any further inquiry into the behavior. One ACLU attorney told me about a galling Alabama case in which prosecutors insisted they would re-seek the death penalty, and it was “only because we were continuing to expose prosecutorial misconduct that they finally agreed to settle the case.”
On a muggy August evening in 1987, police officers swarmed a block of squat brick rowhouses in a mostly white, working-class neighborhood in southeast Baltimore. A young woman had been raped, strangled with a sock and stabbed to death in her second-floor bedroom. Detective Thomas Pellegrini, who’d been assigned to homicide only the year before and, who, by his own admission, was green enough not to sweat the details, caught the case as lead detective. He was assisted by Detective Gary Dunnigan and the squad’s boss, Sgt. Jay Landsman. The trio would become famous a few years later when David Simon heralded them in his book “Homicide: A Year on the Killing Streets” and on the subsequent prime-time TV show it inspired.
The next morning, the neighborhood reverberated with the choppy drone of police helicopters circling overhead. Thompson, a gas station attendant who’d suffered a brain injury in childhood, lived down the street with his wife and their two young boys. He’d heard detectives were looking for a knife and offering a $1,000 reward. It seemed a prime opportunity for a quick buck. The short, stocky 27-year-old wandered over to the yellow police tape and handed Pellegrini a large switchblade. Thompson said he’d found the bloody weapon in the grass the night before, pocketed it, and cleaned it at home — somehow unaware of the massive overnight police presence. At Pellegrini’s urging, he fetched a pair of cut-off jeans he said he’d been wearing at the time, which had a small bloodstain on the back right pocket.
Forensics showed a possible presence of blood or other unknown substance on a small area of the knife and no evidence to suggest it was used in a violent struggle, such as a broken tip from hitting bone. The detectives moved forward on the assumption it was the murder weapon.
Two days later, rather than being thanked and handed the reward money, Thompson found himself under suspicion. In a panic, he fingered Owens. The two had been casual friends, but they’d had a falling out over accusations of theft when they’d briefly worked together at the gas station. In a thoughtless burst of vengeance, Thompson gave an official statement at the police station; he said the knife was actually his but claimed Owens had stolen it and then told him where to find it the day after the murder. Thompson noticed the detectives ate up everything and realized they had nothing else to go on. At the time, there seemed to be no risk in just making it up as he went along. After he retrieved the knife, Thompson told detectives, Owens washed it in the kitchen sink. Thompson didn’t give the police any details about the murder, but he said Owens had told him he’d had sex with the victim.
Owens, 22 at the time, was arrested and charged with burglary, rape and first-degree murder. In just 72 hours, the detectives had closed the case. There was no forensic evidence, motive or eyewitnesses linking Owens to the crime. Landsman and Pellegrini would later say they had believed at the time that without Thompson, Owens would walk. Even the prosecutor, Marvin “Sam” Brave, said he viewed Thompson’s story as “implausible” and didn’t think he had the truth, but he nevertheless pressed charges.
Brave recently told me that “if you think you’ve got the right guy, but not that you can necessarily prove it beyond reasonable doubt, it doesn’t mean you don’t go forward.”
When Owens’ trial began in February 1988, Thompson was the star witness. He’d considered coming clean several times but was afraid he’d be sent to jail. He’d lied to the cops during a previous encounter and had been arrested for making a false police report. Despite that history, the detectives in this case had made him feel like a hero. Pellegrini didn’t think Thompson was “the sharpest pencil in the box,” but at that point in his career, he said in a recent deposition, he thought only suspects would lie to him. Brave also was unconcerned. “If the part that you think he is telling the truth [about] contributes to your case, you use it,” he said. “He doesn’t have to be telling the truth about everything.” The rest of the case relied mainly on minor scratches Owens, a factory worker, had on his arm and a spot of possible blood that had been swabbed from his hand. Two jailhouse snitches who’d been Owens’ cellmates while he awaited trial claimed he had separately confessed to them, though the story Owens purportedly told them contradicted the version Thompson had given police.
In his opening statement, Brave told the jury that any notion that police had “bungled the investigation” and the defendant was innocent was from the fantastical realm of television. But Brave was concerned enough about Thompson’s story that he took him aside the morning of his testimony and warned he was going to “look silly” and it was time he “told us the truth about how that knife really got back into his possession,” according to testimony Brave later gave about the conversation. He even assured Thompson he wouldn’t be prosecuted for making a false statement.
When Thompson took the stand, he told the jury he’d had a “heart to heart” with the prosecutor and was “ready to tell the truth.” In this new version of events — which Brave described later as “sellable” to a jury — Thompson said that around 8 a.m. the morning after the murder, Owens had come by his house and given him the bloody knife. Except this story, too, was a lie. As one of the detectives noted to Brave afterward, Owens’ boss had told police he’d been at work by that point in the morning. “The more I tried to fix things to go in my favor, the bigger hole I dug for myself,” Thompson told me recently.
That Friday Brave went home “really worried about the case,” and stewed over the weekend that he was on “a sinking ship.” Late Sunday evening, he met with Pellegrini and told him to take blood and hair samples from Thompson for testing to exclude him as a suspect and bolster his credibility as a witness. Brave already knew the pubic hairs found on the victim didn’t match Owens. Neither did saliva on a cigarette found at the scene.
During a lunch break at trial the next day, Brave and the three detectives met with the city’s forensics expert who, they said, told them the hair was a match to Thompson. Detectives brought Thompson in, read him his rights, and told him “he was in a lot of trouble” and might be charged. His hair, Landsman told him, had been found in the victim’s house. Thompson later contended he knew this couldn’t possibly be true — he hadn’t been there at all. But at the time, he said, he was scared and thought if he just said what pleased the detectives and got Owens convicted, he’d be alright.
Like an actor doing take after take to accommodate the wishes of a director, Thompson went through several more versions about what supposedly happened, adjusting his story to reflect additional pieces of evidence the detectives told him about. Thompson first said he broke into the house but didn’t go upstairs. After the detectives told him his hair had been found on the second floor, Thompson then said he did go upstairs but hid in the bathroom while Owens attacked the victim after she unexpectedly came home. Detectives then told him his pubic hair had been found on the victim’s buttocks, suggesting his pants must have been down. After several hours of this back and forth, Landsman went to the courtroom and handed Brave a note, saying Thompson had admitted to burglarizing the house with Owens.
Thompson was taken directly from the interrogation room to the witness stand to testify a second time. Now, speaking so softly at first that the judge twice had to tell him to raise his voice, Thompson said he and Owens had broken into the apartment to steal jewelry, and Owens attacked the victim when she came home unexpectedly. Then, while Owens raped her, Thompson testified that he masturbated over her back — his newly concocted explanation for how the pubic hair the state claimed was his had ended up on the victim. Owens, Thompson said, then stabbed her and threw the knife on the ground, which Thompson picked up on the way out.
This was, unbeknownst to Owens or his lawyer, Thompson’s eighth version of events — the one that satisfied the officers that they had enough “to get James Owens,” as one detective later put it.
Even on the stand implicating himself in the crime, with both Brave and Owens’ lawyer stressing charges he might face, Thompson said the full ramifications of his lies didn’t dawn on him. He thought he’d be fine once the trial was over.
“I never hurt anyone. I never touched that young lady,” Thompson said again and again on the stand, adding at one point that he’d take a polygraph to “prove my innocence on that particular behalf.”
Owens was convicted of the burglary and the murder but found not guilty of the rape. Thompson’s changing stories had cast enough doubt that Brave acknowledged in his closing argument that either man could have committed the rape. Thompson, who had been arrested right after testifying and immediately recanted his confession, was later convicted of burglary, rape and murder. Thompson’s multiple different stories of the crime had been accepted as truth, but his multiple attempts to protest his innocence were taken as lies.
Both men were sentenced to life without parole. Owens was the first in Maryland to receive such a punishment.
Owens never resigned himself to his fate. A few years into his sentence, he read about DNA in a magazine and implored everyone he could think of to test the evidence in his case. He eagerly conferred over coffee with Kirk Bloodsworth, the inmate across the hall, then cheered Bloodsworth’s exoneration by DNA in 1993, the first of its kind in the nation involving a death sentence. Shaking Bloodsworth’s hand when he left prison, Owens thought, “Man, one day I’ll be out there.” Then the O.J. Simpson trial introduced him to Barry Scheck, the founder of the Innocence Project, and Owens sent his office a letter. Shunned by his family and cut off from the way most convicts got cash, he traded chicken sandwiches from his kitchen job for stamps to mail it. Still, no one took up the cause. The semen found in the victim and the blood on Thompson’s shorts sat undisturbed in the Baltimore medical examiner’s office for 19 years.
Finally, after a special division within the Maryland public defender’s office became interested, he got a new lawyer and a hearing. A judge ordered DNA testing in 2006 — over the objections of prosecutors — and the results dismantled the state’s theory of the crime. At both trials, the state had argued that the break-in, the rape and the murder were inextricably linked. At Owens’s trial, the prosecutor told the jury Owens had leered at the victim as she sunbathed and “decided that he wanted her.” He broke into her house, laid in wait for her to return, raped her, strangled her and “for good measure … mutilate[d] her with multiple stab wounds.” The prosecution doubled down on this narrative at Thompson’s trial, telling the jury he and Owens “had to humiliate [the victim] by taking turns raping her.” And the blood on the back pocket of Thompson’s shorts, the prosecutor said, was definitively the victim’s.
DNA proved most of those arguments false. The semen found in the victim didn’t come from Owens or Thompson, and the blood on the shorts wasn’t even from a woman. It was Thompson’s own. When Owens heard the news at Jessup Correctional Institution, just southwest of Baltimore, he sat on the floor of his cell and cried.
The Baltimore City State’s Attorney’s Office was unmoved. Prosecutors fought both Thompson and Owens as the two separately sought to have their convictions overturned.
Owens’ case moved faster through the courts. His new attorney was Stephen Mercer, a Maryland defense attorney with an earnestness that had survived more than 20 years in the trenches. Mercer knew the state, with its evidence decimated, was going to push for a deal. He fumed that prosecutors were using psychological warfare to do it — opposing bail and slowing the case, so Owens would spend more time on the inside thinking about being on the outside. Owens’ evidentiary hearing was moved from January to March to May. Only then, nine months after the DNA showed Owens wasn’t the rapist, did the state agree to a new trial while insisting that Owens was still guilty of murder.
The state’s attorney’s office, run at the time by Patricia Jessamy, argued that the rape was immaterial to the murder, and, a spokeswoman said, the DNA evidence was “trivial.” Mark Cohen, the new prosecutor, told Mercer that other evidence in the case, including Thompson’s confession and the testimony of jailhouse informants, was still persuasive. (Jessamy didn’t respond to several phone messages requesting comment and Cohen has since died.)
Mercer said the prosecutor’s stance was “very cynical. It really seemed that the desire to keep the conviction was for reasons that had nothing to do with the evidence.” The state’s guiding star, Mercer knew, was a rigid belief that what was long ago decided by a jury, and upheld by an appellate court, shouldn’t be continually second-guessed.
In Owens’ case, it wasn’t just the semen and the blood that didn’t hold up 20 years later. The type of hair analysis done on the pubic hair had subsequently been dismissed as junk science. The hair, along with the knife, had been destroyed. But the state’s own expert, who’d inspected the hair at the time of the original trials, said at a hearing that the scientific community no longer does a visual hair comparison to “draw the conclusions we drew back in 1988 with a microscope.” Now analysts use DNA analysis.
Not long after Owens was granted a new trial in May 2007, Cohen proposed a deal. It wasn’t surprising. The plea bargain is the lifeblood of the overburdened criminal-justice system. About 95 percent of cases never go before a jury. Instead, most defendants agree to plead guilty in exchange for lesser sentences. In cases like Owens’, in which new evidence undermines old, legal advocates question whether incarcerated defendants should even be offered a plea. In every case, prosecutors “need to really inspect their own motivations,” Thiru Vignarajah, a former federal and Baltimore City prosecutor who later served as deputy attorney general of Maryland, said. “Are they offering a plea or time served because that’s in the best interest of the case, or are they allowing some institutional interest of preserving the conviction to trump a prosecutor’s duty to seek justice?”
A year before Owens’ retrial, Jessamy’s office had convinced another defendant to take an Alford plea. Locked up for 20 years, that defendant had at first refused a deal after he, too, was granted a new trial because of DNA evidence. As the trial was set to begin, the prosecution requested a postponement. When the state again delayed the subsequent trial date, the defendant broke down. He accepted the plea.
Afterward, Jessamy’s spokeswoman scoffed at the defendant in a news story, saying it was “inconceivable” that after 20 years the defendant couldn’t wait a little longer, and “if he truly believes he is innocent, he should have gone to trial to see that justice is served.”
As Owens’ trial got closer, Cohen kept sweetening the deal, knocking down the charge and requiring less probation. Finally, they offered Owens an Alford plea for second-degree murder, time served and no probation. Mercer lost sleep over whether Owens should take it. A trial was risky and a chance at guaranteed freedom was rare for any defendant. Owens repeatedly asked himself: “Why are they doing this to me? Why should I have to plead guilty to something I didn’t do?” Now mostly bald and with a moustache, he’d grown up in the foster care system. He’d been viciously attacked while in prison. He didn’t have much to hold onto except his resolute insistence from day one that he was innocent. He wasn’t about to “admit there was sufficient evidence to convict him while playing this wink-and-nod game that he was claiming his innocence,” Mercer said. So the Alford plea, like all the others Mercer had passed to Owens through the Plexiglass, was flatly rejected: “Mr. Mercer, there is no way. I am going to trial.”
Cohen, suspicious that the deal hadn’t been properly relayed, had Owens and Mercer join him for a bench conference, so that the Alford plea could be offered in front of the judge. “I’m not taking nothing, dude,” Owens recalled saying. “I will die in the penitentiary if I have to.”
In October 2008, Owens was vindicated. Cohen was forced to tell the court he didn’t have the goods for a retrial. Owens stepped out of prison free for the first time in 21 years, telling gathered reporters, “You can’t give me that time back.”
Thompson, meanwhile, was fighting the same battles while incarcerated about 75 miles away at Roxbury Correctional Institution in Hagerstown, Maryland. But in his case, prosecutors were employing a perplexing logic. They’d agreed that the DNA evidence from the semen warranted a new trial for Owens, who had not been convicted of rape, but they refused a new trial for Thompson, who had been.
Thompson, by now gray-haired and hard of hearing, was dismayed. He’d saved the newspaper clipping about the DNA findings, and when he read that Owens had gone free, he was certain he’d be next. He couldn’t understand why the DNA could clear Owens of all charges while it did nothing for him, even though the DNA excluded him as well. But Mercer, who’d picked up Thompson’s case after freeing Owens, did. Thompson had confessed, and that was prosecutorial gold. In Simon’s book about the Baltimore detectives who’d secured Thompson’s confession, he detailed the interrogation tactics they had commonly employed. To get confessions, he wrote, the detective became a “huckster … thieving and silver-tongued,” and without the “chance for a detective to manipulate a suspect’s mind, a lot of bad people would simply go free.”
Poorly understood at the time is that such manipulation can also compel innocent people to agree to whatever the police want. As the U.S. Supreme Court noted in 2009, “a frighteningly high percentage of people … confess to crimes they never committed.” According to the Innocence Project, 28 percent of defendants later exonerated by DNA had falsely confessed.
During the initial trials in 1988, prosecutors had argued that the pubic hair and the blood on the jeans proved Thompson was telling the truth, but in 2009 the Maryland Court of Appeals wrote that the DNA finding “usurps the State’s arguments all together.” In essence this meant none of Thompson’s statements to police or prosecutors throughout the case were corroborated by evidence.
Despite the statistics, convincing a jury that someone would falsely confess to a crime — particularly to something as heinous as a murder or a rape — is incredibly hard. Juries want to believe that people are rational actors, like themselves, with an almost primal instinct toward self-protection. It wouldn’t matter that the state no longer had the evidence to prove it, Mercer knew, a jury would most likely myopically focus on the confession.
Thompson told me he’d been happy for Owens when he was released — he’d always wished he could apologize to him for what he did — but that feeling had faded into self-pity as the calendar went from 2008 to 2009 to 2010 and his case stalled in the courts. Now he was mostly anxious. He just wanted relief, whatever it might be, so when Sharon Holback, the new prosecutor on the case, eventually offered him an Alford plea — 23 years after he’d first fatefully approached police — his excitement overwhelmed his sense of injustice.
Mercer worked to make it the best deal he could. If Thompson took the plea, it meant the state would let him go, but the deal had some risky strings attached. Any charge that carried a life sentence had to come off the table, because in Maryland, a probation violation — even something as relatively minor as a DUI — sends the defendant back to prison to serve the remainder of his sentence. The two sides agreed to second-degree murder, which carries a maximum of 30 years. That way if Thompson violated probation, he’d only have seven and a half years over his head, since he had served more than 22.
Gregg Bernstein, Baltimore City state’s attorney from 2011 to 2015, oversaw at least two similar deals. He couldn’t remember the details but said he’d thought a lot about whether it was okay for an innocent man to take an Alford plea. In the end, he said, most cases lack black-and-white certainty, regardless of evidence suggesting innocence. “It’s not that simple to say yay or nay,” he said. “Pleas are a way to resolve them.”
Former prosecutor Vignarajah, though, told me he wonders if that kind of resolution only looks like a win for everyone on paper. “In reality everyone lost,” he said. “The victim sees no justice. The defendant is walking away with a conviction. And the prosecution didn’t get anyone to take responsibility [for the crime].”
On July 29, 2010, when Thompson left prison under the Alford plea, Holback got the last word: Thompson “is in no way exonerated.”
Since their releases, Thompson and Owens have led dramatically different lives.
Thompson thought he could go back to the person he was almost 23 years earlier, before the murder rap, but society didn’t look at him that way. When he applied for a job, he put a question mark where the form asked if he’d been convicted of a felony.
“I tried to explain I was wrongfully convicted, but people don’t want to hear that,” Thompson said. “There’s no reasoning with somebody. ‘Innocent people do not go to prison’ is just the motto.”
Thompson held onto his freedom for only a little over a year. In October 2011 he was arrested after his ex-girlfriend claimed that he had molested her young daughter. Thompson, who’d recently kicked the girlfriend out of his apartment, denied the charge, saying he’d spanked the girl’s bare butt to discipline her. The state reduced the charges to a misdemeanor for touching the girl’s buttocks and gave him time served for the five months he’d been in jail.
It didn’t end there, though. Because the misdemeanor violated his probation attached to his Alford plea, Thompson went from a local jail to a state prison to serve the remaining seven and a half years.
Mercer said he believes the Alford plea made it very difficult for Thompson to defend himself. “It was a question of credibility,” Mercer said. “Who’s going to believe him? He was stuck having to do damage control.”
Owens has fared better. He has been embraced by what little family he had. He has moved into a cousin’s house and has begun working with him cleaning gutters and doing landscaping. And he has grown close to his nieces and nephews, a bittersweet feeling for someone who’d had no chance to build a family of his own. Owens told me he has tried not to let the anger sink him, but he struggles. His exoneration came without compensation or even an apology. “What’s striking in these cases is a total lack of accountability,” said Michele Nethercott, of the Innocence Project in Baltimore. “Nothing ever really happens” to the police and prosecutors whose actions led to wrongful convictions.
Owens wonders today if his prosecution became all about keeping the win. “Instead of focusing on me and getting me to take a deal for something I didn’t do, they need to focus on the victim. Her murder has never been solved,” he said. “I think they should go back and look and do something for this girl.”
In 2011, Owens found a lawyer, Charles Curlett, to sue Baltimore. Curlett determined that there were several issues of misconduct involved in Owens’ conviction. First, his lawyer had been told nothing of the changing stories Thompson gave the detectives. The information could have been used to undermine Thompson’s credibility and failing to share it was likely a violation of Owens’ due-process rights. Such failures are known as Brady violations, after a 1963 Supreme Court case in which the justices determined that withholding favorable information from the defense is unconstitutional. Also, one of the jailhouse snitches who testified that Owens had confessed had been a police informant for years and said he recruited the other snitch. This, too, wasn’t revealed to the defense, nor were the informant’s letters asking for favors in exchange for his testimony.
Brady violations had become so prevalent in Baltimore’s courts that the Fourth Circuit Court of Appeals recently admonished the city’s prosecutors to remember their legal obligations: “Only this practice ensures the fair trial that our justice system aspires to provide” and makes it so “no one has to worry after the fact whether the jury convicted the wrong person.”
The city furiously fought Owens. Dodging such suits, many defense lawyers contend, is part of what drives these plea offers. “If not expressly that, it’s implicit in a lot of decisions made in this setting,” said Michael Imbroscio, an attorney who had a client in Baltimore City take a time-served deal. The city won dismissal of Owens’ suit against the state’s attorney’s office and Brave, who the court ruled had immunity, and the Baltimore Police Department. But the case is going to trial in federal court, likely early next year, against detectives Pellegrini, Landsman and Dunnigan as individuals. There’s millions in compensation at stake for Owens and a public airing of misdeeds for the city.
Civil litigation is “so important,” Mercer said. “Often, that’s the only time there’s scrutiny into what wrongs were done.”
The type of misconduct alleged in Owens’ case is echoed in nine more of the 14 exonerations out of Baltimore City and County since 2002, according to the National Registry of Exonerations. The 2014 exoneration of Sabein Burgess, for example, came after it emerged that Baltimore detectives never revealed a key detail to the defense: that a young witness had told them he saw the murder suspect and it wasn’t Burgess. The detectives even submitted a report falsely stating that the witness had been asleep during the crime. Like Owens, Burgess is suing, claiming that detectives “cut corners and rushed to judgment.” His trial is set for this fall and names a different group of detectives.
Misconduct can also be found in the cases of some of the remaining exonerated defendants who, like Thompson, aren’t officially considered exonerated at all but who were released under Alford pleas or time-served deals after questions were raised about their initial convictions. Curlett is representing one such man, Wendell Griffin, who was convicted of murder in Baltimore in 1982. Decades later, it came to light that three detectives — two also featured in Simon’s book and a third who is Landsman’s brother — had buried photo lineups and witness statements pointing to Griffin’s innocence. He was let out on a time-served deal in 2012.
The detectives named in the Owens and Burgess lawsuits have denied allegations of misconduct. Michael Marshall, who represents the detectives in Owens’ and Griffin’s suits, declined to comment, referring questions to the chief of legal affairs for the Baltimore City Police Department, who didn’t return several calls.
Thompson, whose parents died while he was in prison, has been abandoned by the rest of his family. He was released early for good behavior in February after serving a little more than five of his remaining seven and a half years, and as much as he blames himself for his mistakes, he now thinks his plea was a “bum deal.” He wishes there was a way to prove to his loved ones that “although I served 30 years … I didn’t commit the crime.”
The strain of the Alford plea proved too much for one of Baltimore’s wrongly convicted. Chris Conover left prison under the plea in 2003 after DNA called into question his murder conviction in Baltimore County. On the outside, he suffered from severe panic attacks and depression, but his wife told the local newspaper that he couldn’t face in-patient treatment, which meant being back behind locked doors. His petition for a pardon from Maryland’s governor was turned down in 2012. Three years later, Conover killed himself.
“Having been convicted really defines who you are — it becomes itself a prison,” Mercer said. “Once out, with a conviction still on your shoulders, having maintained your innocence in a Alford plea is of little comfort and of very little practical benefit.”
Despite new evidence undermining the convictions of at least eight men for violent crimes in both Baltimore City and County over the last two decades, none were exonerated. Instead, they left prison only after agreeing to plea deals with state prosecutors. In each case, the men took either Alford pleas, in which defendants can maintain their innocence for the record, or were given time-served arrangements. With these deals, the defendants were granted their freedom, but gave up the right to clear their names. (Two additional men took similar deals but years later were fully exonerated after more exculpatory evidence was found in the police files.)
ProPublica’s examination of these cases reveals a troubling pattern — one that legal experts say plays out across the country. Persuasive innocence claims were met with refusals by the state’s attorney’s office to reexamine the cases, sometimes despite — or perhaps because of — discoveries of official misconduct. Prosecutors often fought for years to prevent the consideration of any new evidence or the testing of old evidence for DNA. Or they accommodated contrary new facts by stretching their theories of crimes. If the DNA in a rape case, for example, didn’t match the defendant, prosecutors would assert that another unknown assailant was involved, too. When judges ordered new trials or granted writs of innocence, prosecutors started bargaining for plea deals that would maintain the convictions.
Over time, prosecutors have defended their decision to seek deals, claiming in each case that they still believed in the defendants’ guilt. They also argued that given the amount of time passed, the cases would be difficult to retry.
But Michele Nethercott, the head of the Innocence Project Clinic at the University of Baltimore School of Law, said with these cases, “often, the truth doesn’t seem to matter much.”
Type of Deal: Alford plea
The Original Case
The white victim identified him 10 weeks after the crime. The victim’s ID of Seward, an 18-year-old black man who had a moustache and goatee at the time of the murder, conflicted with her contemporaneous description after the attack of a clean-shaven assailant. Neither the fingerprints nor biological evidence from the crime matched Seward.
New Evidence Later Discovered
Seward’s employment records as a part-time dog washer, which were discovered 12 years after the trial, showed he’d been at work the day of the shooting. His boss also testified she kept the shop locked and it would have been “impossible” for him to have left.
Fought for the next 19 years, arguing, in turn, that the records weren’t admissible as new evidence and shouldn’t be given any consideration; that they didn’t provide an alibi because no hours were specified; and that they bolstered the case against Seward because the shop was near the victim’s house. One of the prosecutors on the case, John Cox, also told ProPublica that the records’ discovery so long after the trial meant they couldn’t be trusted.
Baltimore County State’s Attorney Scott Shellenberger said recently that because the victim saw her attacker up close, he wasn’t concerned that the case rested on a cross-racial identification. (That type of ID has been shown to be less reliable because people are generally bad at distinguishing facial features of people who aren’t their own race. Of the 351 people exonerated by DNA evidence since 1989, the national Innocence Project found that 41 percent had been convicted on mistaken cross-racial identification.)
How the Deal Happened
Judge said the employment records “thoroughly exculpate[d]” Seward and granted a writ of innocence. The state appealed and eventually lost. “The state’s immediate reaction was to offer a plea,” said Shawn Armbrust, of the Mid-Atlantic Innocence Project and one of Seward’s lawyers.
Shellenberger said that he’d been confident about the case and wanted to go to trial, but the victim didn’t want to testify again. “Keeping something on the record was extremely important to us.”
Why Defendant Agreed to Deal
Seward first turned prosecutors down, but then, as he awaited a new trial, a close friend was stabbed in prison. Seward had nine months before the trial began, so he reconsidered.
Type of Deal: Time served
The Original Case
Based solely on Barnes’ confession made after 31 hours in custody. A largely illiterate 17-year-old with a low IQ and no prior record, Barnes’ police-typed statement conflicted with the evidence in the case in major ways, such as how and where the 15-year-old female victim was killed. And he had an alibi for the time of the murder. (Barnes’ confession also incriminated two others, but no one else was charged.)
New Evidence Later Discovered
In 2009, 37 years after Barnes’ conviction, DNA evidence collected from the victim’s body was tested and excluded him from any sexual assault, further undermining his confession, which had described a violent gang rape that included Barnes and another man ejaculating. The DNA, which only came from one male, also excluded one of the other teenagers implicated in Barnes’ statement.
Prosecutor Sharon Holback said at the time that the state “vehemently and firmly believes that [Barnes] was fairly and properly convicted.” She argued that his confession was sound and that the third person implicated in it must have been the source for the DNA. That man couldn’t be found for comparison testing. (Holback was also the prosecutor who handled the post-conviction hearings in the case of James Thompson, whose rape and murder conviction was undermined by DNA testing, but was offered an Alford plea.)
How the Deal Happened
Judge Yvette Bryant went many months without issuing a ruling on the case, so Barnes’ lawyer took the innocence claims directly to Gregg Bernstein, who recently had been elected as Baltimore City state’s attorney on a reform agenda and had started a conviction integrity unit. The fighting over Barnes’ post-conviction motions had happened under Bernstein’s predecessor, so he had not publicly committed to any position. He was also free of one common concern prosecutors face when dealing with potentially wrong convictions: angry relatives of the victim who don’t want the case to unravel. With Barnes, the victim’s family so believed in his innocence that they had hired a lawyer to defend him.
Bernstein, who said recently that he didn’t recall the case, would concede only that Barnes didn’t deserve to be in prison anymore, seizing on a mistake in sentencing. The judge who had sentenced Barnes had thought wrongly that his only option was life.
Why Defendant Agreed to Deal
Barnes was 57 years old, had been in prison for more than 40 years and was in failing health. “I had to say to him ‘I’m confident in the end we will vindicate you, but it might be 1, 2 years or even 4 to 5 years, and there’s no guarantee,’” said Barnes’ pro bono lawyer, Michael Imbroscio, noting it was “the most difficult conversation I’ve ever had in my 22-year legal career.”
Type of Deal: Time served
The Original Case
A neighbor testified that she saw Griffin before and after the murder with a gun, and a second neighbor, who was 150 feet away, said she heard Griffin make threatening remarks the night of the murder. A set of keys found about 90 feet from the crime scene was connected to Griffin, who lived in the neighborhood.
New Evidence Later Discovered
In 2011, significant evidence was found in the police’s files that had never been given to the defense: three photo lineups in which eyewitnesses failed to identify Griffin and eight witness statements that either incriminated another suspect or contradicted the testimony used to prosecute Griffin.
One eyewitness pointed to Griffin’s picture in the lineup and said that he looked like the suspect, “but it’s not him.” Griffin’s picture was nine years old, so detectives went back to that witness and showed her another array with a current picture. She still did not identify him. Nonetheless, detectives used her description of the suspect to get a search warrant for Griffin’s home — never mentioning that she’d twice failed to pick him out of a photo array. The warrant also cited a neighbor who saw a man with a gun, but left out that he said the man wasn’t Griffin.
“There was pretty powerful evidence of innocence that was buried by the state,” Steve Mercer, Griffin’s attorney, said.
Baltimore City prosecutor Michael Leedy denied that the evidence represented a Constitutional violation. (In 1963, the U.S. Supreme Court declared that the state must turn over all favorable information to the defense in order for a trial to be fair, which has come to be known as the “Brady” requirement.) Leedy wouldn’t agree to a new trial.
How the Deal Happened
When a judge, who called the evidence “earth shattering,” indicated she’d be ordering a new trial, Leedy shifted, saying that although he didn’t believe “there were, in fact, any Brady violations” the allegations were “plausible enough” that he’d “concede to a resentencing on this matter.” This was the “best course,” Leedy said, to “ensure that Mr. Griffin will for the rest of his life remain convicted for the murder of James Wise.”
Leedy also wanted it on record that by accepting the deal Griffin gave up the right to an actual innocence ruling.
Why Defendant Agreed to Deal
Griffin was 61, knew his best years were gone and he might “die in here.” Having spent nearly 31 years in prison, he didn’t have it in him, he said recently, to wait another year-and-a-half for a new trial. But he is now trying to withdraw his deal, so he can clear his name and sue over the Brady violations. Marilyn Mosby, the current state’s attorney who ran in part on a platform of police accountability, is fighting his motion. (Her spokeswoman didn’t respond to multiple requests for comment.) A hearing is set for November.
Type of Deal: Alford plea (exonerated in 2000)
Pettiford, 23 and with a record, was identified as one of two shooters by two eyewitnesses and was tied to the murder weapon by a suspect in a related crime. But at trial, the witnesses said they’d been mistaken and the suspect said he’d lied about the weapon. Late in the trial, prosecutors produced a new witness who identified Pettiford. Pettiford had an alibi and no motive.
According to The Baltimore Sun, before the judge sentenced Pettiford to life plus 20 years, he said: “I don’t care if every witness that appeared in the trial — including the detectives — come back here and say it was all a farce and it was all false and it was all wrong. I think justice was done.”
New Evidence Later Discovered
A year later, a separate federal drug investigation led to a different suspect in the murder, who pleaded guilty in federal court and told investigators that Pettiford had nothing to do with the crime.
There was also evidence that had never been given to the defense: a three-page statement from a friend of the victim that said he was the intended target and pointed to the same suspect prosecuted by the feds; a police bulletin that named that same suspect in connection to the murder; a statement from an eyewitness who identified the second shooter as someone the federal prosecutors thought was involved; and a police report naming that second person as a suspect.
Baltimore City prosecutor Nancy Pollack, who had handled the trial, didn’t act on the information federal prosecutors gave her suggesting Pettiford was innocent. Michelle Martz, Pettiford’s lawyer, said she went repeatedly “to beg and plead for [prosecutors at the time] to do something. I was floored the state wouldn’t be more concerned that they might have the wrong guy.”
How the Deal Happened
At the end of a post-conviction hearing, at which a detective revealed the existence of the three-page statement implicating someone else, the judge ordered Pollack to turn over everything in her files. Pollack agreed to a new trial and offered the plea.
Why Defendant Agreed to Deal
Pettiford, scared of what the prosecutors might do during a second round, had only one question: “Do I have to go back to prison if I take it?” He accepted the Alford plea, walked down the courthouse steps and into his family’s waiting car.
How He Was Later Exonerated
A year after the Alford plea, The Baltimore Sun newspaper exposed that the state had suppressed even more evidence and that a detective had misled the defense. In response, the judge vacated the Alford plea, saying it had been “a miscarriage of justice,” and the state declined to prosecute again. Pollack, who declined to comment, had already resigned, but the Baltimore Police Department found that the detective did nothing wrong. That detective was also named in a lawsuit filed by Sabein Burgess, who was wrongfully convicted in 1995 and exonerated in 2014.
By many measures, the U.S. has become the most dangerous place to give birth in the affluent world. Each year 700 to 900 American women die from pregnancy or childbirth-related causes – up to 60 percent of which are preventable – and some 65,000 women nearly die.
ProPublica and NPR have shined a light on this issue through the joint investigative series Lost Mothers, shifting the national conversation on maternal mortality from one of private tragedy to public health crisis. Now the news organizations are teaming up, in partnership with the Brooklyn Public Library’s BPL Presents, to host a community forum about protecting more women from harm.
strategies for self-advocacy and conveying levels of pain
changing the culture of the “perfect birth story”
preparing for an emergency
paying attention to symptoms even after the delivery
The expert panel will also take questions from audience members seeking answers and support, and refreshments will be provided. This one-of-a-kind event encourages women and families to share their stories and connect with one another, and elevates a much-needed national dialogue.
Maternity care is disappearing from America’s rural counties, and for the 28 million women of reproductive age living in those areas, pregnancy and childbirth are becoming more complicated — and more dangerous. That’s the upshot of a new report from the Rural Health Research Center at the University of Minnesota that examined obstetric services in the nation’s 1,984 rural counties over a 10-year period. In 2004, 45 percent of rural counties had no hospitals with obstetric services; by 2014, that figure had jumped to 54 percent. The decline was greatest in heavily black counties and in states with the strictest eligibility rules for Medicaid.
The decrease in services has enormous implications for women and families, says Katy B. Kozhimannil, an associate professor in health policy who directs the Minnesota center’s research efforts. Rural areas have higher rates of chronic conditions that make pregnancy more challenging, higher rates of childbirth-related hemorrhages — and higher rates of maternal and infant deaths. And because rural counties tend to be poorer, any efforts to revamp or slash Medicaid could hit rural mothers especially hard. We spoke with Kozhimannil about the new study and the implications for maternal care. (The conversation has been edited and condensed.)
You and your colleagues have been looking at maternal health issues for several years. What’s the most surprising part of this new study?
I was surprised about the findings on race. Being aware of structural racism in U.S. health care, I shouldn’t have been. But we found that hospitals are more likely to close their doors entirely or close their obstetric units in communities that have more black residents. Rural black communities also experience some of the poorest birth outcomes in the country, especially in the Southeast.
I think [the race findings] are new and really important. In all the discussions I’ve had around maternity care access, I think there’s often a false association of “rural” with white communities and with farming, but that doesn’t represent the demographic reality of rural America, which is very diverse. There are 10 million people of color in rural America, that’s about 20 percent of all rural Americans.
What has led to the decline in rural obstetric services more broadly over this 10-year period?
We didn’t choose this period because we thought it was particularly unique. We chose it because it was the most recent decade of data we could get. That said, this was a period when there was a substantial shift in the health care delivery system. The debates around Obamacare, the implementation, the threats to repeal — all that really created instability with respect to what hospitals and clinicians were expecting around payments.
And the role of finances is key. If hospitals want to offer obstetric services, they need to be ready for a baby to be born at any time — they need to have a bed available, the equipment available for mom and for baby, clinicians and staff available that have the necessary skills. That’s a substantial expense. If a hospital’s revenues are limited because it has a low volume of births — as many rural hospitals do — or if revenues are unpredictable, that creates a really difficult administrative problem.
How does Medicaid play into this?
Medicaid funds about half of all births in the United States, and an even greater percentage of births in rural hospitals. Medicaid funding for births is incredibly important and it’s one factor in hospitals’ decisions around whether to keep obstetric services. We found that rural counties in states with more generous Medicaid programs — with higher income eligibility limits for pregnant women — were less likely to lose hospital-based obstetric services.
Meanwhile, there’s talk of allowing states to impose new rules that could restrict access to Medicaid.
Changes to the financing of Medicaid would likely have big negative effects on the availability of obstetric services in rural areas. Based on our study, the generosity of a state’s Medicaid program seems directly linked to access to maternity care in rural counties. As such, any new reductions or restrictions on Medicaid funding or services may affect rural hospital financing.
What is it like to be pregnant in a rural area that doesn’t have adequate maternity care? What do women do?
For some women, there may be a nearby clinic or their general practitioner may be able to see them for prenatal visits if they have a low-risk pregnancy. But then they need to give birth in a more distant area with a different set of providers.
That may not even be a choice for women who live in communities that don’t have any providers that see pregnant patients, or for women that have higher risk complications that require more specialized care.
I remember talking to one woman who lived in rural northern Minnesota and who had a preterm birth with her first pregnancy. For her second pregnancy, she had to drive two hours to the nearest hospital with a high-risk obstetrician. With one child at home already, and a full-time job and a partner who worked, it was almost untenable. It would take a whole day for her to drop her child off at daycare, drive all the way to the hospital, wait for a 15-minute visit that felt rushed, then drive all the way back.
I just heard on the radio this morning that a truck ran into a railroad bridge that goes over the highway that this woman would take to go back and forth to the hospital. So if she was pregnant right now, there’s a 27-mile detour on three dirt roads to get around this broken bridge. That adds probably another 45 minutes to an already two-hour drive. Things like that can happen, you know, all the time.
What about giving birth? How does living in a remote area affect the kinds of choices doctors and women make?
In a typical childbirth education class in an urban area, childbirth educators say things like, “Go to the hospital when your contractions are five minutes apart.” None of that makes any sense in a rural context where women give birth far from home.
For rural moms, a lot of the conversation in childbirth education and in prenatal care revolves around logistics and transportation: “Do you know how you’re going to get to your appointment? Do you have access to a car? Is your car reliable? Do you have money for gas? Do you have a backup plan if your car doesn’t start? Do you have someone that you can call if you need to go in quickly?”
Anecdotally, I hear a lot about labor induction. The rural physicians I’ve talked to are like, “I can’t believe I am trying to talk patients into having an induction.” They believe in letting labor start naturally, but given the long drive, induction is often better for patients clinically. So that if complications come up, someone’s there, monitoring your blood pressure and vital signs. It’s not, you know, your partner or friend desperately driving down dirt roads as fast as they can while you yell in the back seat.
How does all this affect outcomes for babies?
We have good information from Canada that the women who have to drive long distances to give birth have higher rates of the babies being in the neonatal intensive care unit, and even of infant mortality. And so we know that distance is associated with outcomes of care. When rural hospitals close the doors of their maternity units, women have to drive longer distances.
These seem like pretty huge hurdles for rural mothers and babies. Is there any way to address these problems to improve maternity care?
One idea is programs to support pregnant women and families, especially with respect to their housing and transportation needs when they live far away from where they’re going to give birth. Alaska has actually done a tremendous job of this.
Another is for states to allow midwives and nurse practitioners to play a greater role in offering prenatal and postpartum care, without having to be under a doctor’s supervision. That would be useful. Our prior research shows that midwives, for example, attend births at about one-third of all rural hospitals, and that hospital administrators would like to expand the role midwives play.
State and federal programs to support the rural maternity workforce are crucial. There ought to be programs to support training in emergency births in rural communities that lose obstetric care, and to support the costs of providing maternity care in communities where there are willing providers.
Demócratas de alto nivel en los comités tanto de asuntos extranjeros como los judiciales de la Cámara de Representantes exigieron eljueves una investigación de los operativos liderados por la DEA en Honduras y México que resultaron en las muertes de docenas, posiblemente cientos, de personas que no tenían nada que ver con el narcotráfico.
La carta, dirigida al Fiscal General Jeff Sessions y el Secretario de Estado Rex Tillerson, se basó en una contundente investigación por el inspector general del Departamento de Justicia de un operativo de la DEA en Honduras en 2012 — conocida como Operación Anvil (Yunque) — contra las redes de narcotráfico que operan en la costa caribeña del país. Durante un operativo fallido, oficiales de una unidad de la policía federal hondureña escogidos, entrenados y comandados por la DEA dispararon contra un taxi acuático en el que viajaban personas aparentemente desarmadas y no vinculadas al narcotráfico. Cuatro personas murieron y otras cuatro fueron heridas.
La carta también menciona una investigación de ProPublica de una masacre cometida en 2012 por el cartel de los Zetas en el estado mexicano de Coahuila que se desencadenó después de que información sensible compartida por la DEA con una unidad de la policía mexicana bajo su control llegó amanos de los jefes del cartel, quienes ordenaron una ola de represalias contra gente sospechada de ser traidora.
La masacre dejó docenas y potencialmente cientos de personas muertas y desaparecidas dentro y alrededor del pequeño y tranquilo pueblo de ganaderos, Allende, que está a 40 minutos en auto de la frontera con Texas. El reportaje fue publicado en colaboración con National Geographic en junio.
“Creemos que estos y otros asuntos problemáticos merecen aún más investigación,” dice la carta, escrita por el representante Henry C. “Hank” Johnson, Jr., de Georgia, y firmada por el representante John Conyers de Michigan, el miembro de más alto rango del comité de asuntos judiciales, el representante de New York Eliot Engel, el miembro de nivel más alto del comité de asuntos extranjeros, y 11 representantes más.
“¿Dado que tanto el informe (del Inspector General) como un reciente reportaje de ProPublica han revelado las deficiencias de las unidades policiales bajo escrutinio de los Estados Unidos en Honduras y México, están ustedes llevando a cabo una evaluación completa del sistema de unidades especialmente entrenadas por nuestro gobierno para asegurar que se realizan mejoras?” preguntó Johnson en la carta.
Durante dos décadas, el llamado “Programa de Unidades Investigativas Sensibles” (SIU por sus siglas en inglés) ha sido la solución alternativa con que la DEA ha manejado el problema de combatir la droga al lado de un socio extranjero del que no se confía. La agencia establece una unidad bajo su mando, seleccionando agentes de las fuerzas policiales del país anfitrión. Después entrena a estos agentes extranjeros — frecuentemente en Estados Unidos — les somete a pruebas de polígrafo y, en algunos casos, trabaja a su lado en la calle.
La agencia ha establecido “SIUs” en alrededor de 13 países en el mundo. Administradores de la agencia los han avalado como el “pan nuestro de cada día” de las actividades de la DEA en el extranjero.
En México, sin embargo, la SIU ha estado plagada por la corrupción desde el inicio. Desde el 2000, al menos dos supervisores han sido asesinados después de que sus identidades y ubicaciones fueran filtradas a narcotraficantes por miembros de la SIU, según las acusaciones de agentes actuales y antiguos de la DEA que han trabajado en México. Este año, otro supervisor de la SIU, Iván Reyes Azarte, voló a Chicago y se entregó a las autoridades estadounidenses quienes le acusaron penalmente de colaborar con narcotraficantes.
LA DEA no niega la corrupción entre los rangos de su unidad mexicana. En entrevistas, agentes dijeron que parte del “juego” de trabajar en México consiste en entender que la unidad especial —
y cualquier unidad policial mexicana — puede filtrar información a un cartel en concreto y ayudar de buena fue a perseguir a otro. Los agentes dijeron que se trataba de determinar a cuál cartel la unidad especial estaba ayudando, y entonces usar la unidad para perseguir a los rivales de este cartel.
Pero las investigaciones hechas por ProPublica y el inspector general han dejado claro que a veces el juego tiene consecuencias trágicas. Y mientras la DEA rápidamente se atribuye el mérito cuando ayuda a sus socios extranjeros a capturar a capos importantes de la droga, se mantiene silenciosa o proclama su inocencia cuando se pierden vidas como resultado de sus operaciones.
En el caso de la masacre en México, oficiales de la DEA no hicieron una evaluación interna del papel de la agencia en el ataque, y no suspendieron la relación de la agencia con la unidad especial ni ofrecieron asistencia a las víctimas de la filtración. Y en Honduras, el Inspector General Michael E. Horowitz concluyó que la DEA no solo omitió hacer una revisión del tiroteo, sino que a sabiendas despistó al Congreso en un intento de encubrir su papel.
Entre las revelaciones del informe del inspector general: oficiales de la DEA y el departamento de estado dijeron al Congreso que sus socios policiales hondureños dispararon contra el taxi acuático porque los pasajeros del taxi dispararon primero, pero no había “pruebas creíbles de que los individuos en el barco de pasajeros abrieran fuego primero.” Las pruebas disponibles “suscitan interrogantes serios sobre si hubo disparos desde el barco de pasajeros en algúnmomento,” dijo el informe.
Aunque la DEA ha insistido que solo jugó un papel de apoyo en el operativo hondureño, el inspector general determinó que fue un agente de la DEA quien ordenó a un artilleroen el helicóptero abrir fuego contra el taxi acuático.
En un comunicado de prensa, Johnson escribió que el informedel inspector general “confirma nuestros peores temores,” y deja muchas preguntas sin respuestas. “¿La pregunta más importante de todas es: que está haciendo nuestro gobierno para corregir esto y asegurar de que, de ahora en adelante, cualquier agente de los Estados Unidos involucrado en la pérdida de vidas inocentes en el extranjero tenga que rendir cuentas?”
Over the past decade, the DNA laboratory in the office of New York City’s chief medical examiner emerged as a pioneer in analyzing the most complicated evidence from crime scenes. It developed two techniques, which went beyond standard practice at the FBI and other public labs, for making identifications from DNA samples that were tiny or that contained a mix of more than one person’s genetic material.
As its reputation spread, the lab processed DNA evidence supplied not only by the New York police, but also by about 50 jurisdictions as far away as Bozeman, Montana, and Floresville, Texas, which paid the lab $1,100 per sample.
Now these DNA analysis methods are under the microscope, with scientists questioning their validity, ProPublica has found. In court testimony, a former lab official said she was fired for criticizing one method, and a former member of the New York State Commission on Forensic Science said he had been wrong when he approved their use. The first expert witness allowed by a judge to examine the software source code behind one technique recently concluded that its accuracy “should be seriously questioned.”
Earlier this year, the lab shelved the two methods and replaced them with newer, more broadly used technology.
A coalition of defense lawyers is asking the New York State inspector general’s office — the designated watchdog for the state’s crime labs — to launch an inquiry into the use of the disputed analysis methods in thousands of criminal cases. While the inspector general has no jurisdiction over the court system, any finding of flaws with the DNA analysis could prompt an avalanche of litigation. Previous convictions could be revisited if the flawed evidence can be shown to have made a difference in the outcome.
The medical examiner’s office “has engaged in negligent conduct that undermines the integrity of its forensic DNA testing and analysis,” the Legal Aid Society and the Federal Defenders of New York wrote the inspector general on Friday. Because the lab has kept problems with its “unreliable” testing and “unsound statistical evidence” secret from the public and the courts, they continued, “innocent people may be wrongly convicted, and people guilty of serious crimes may go free.”
In addition to those convicted using the disputed methods, many defendants may have chosen to plead guilty when they learned prosecutors had DNA evidence against them. Their cases face significant barriers to reconsideration.
The medical examiner’s office stands by its science. Its chief of laboratories, Timothy Kupferschmid, said that the discarded techniques were well-tested and valid, and that the lab was adopting newer methods to align with changing FBI standards. He compared it to a vehicle upgrade.
“So just because we’re switching to the new model, I mean, our old pickup truck worked great, but my new pickup truck is so much better,” he said.
One case that hinges on the disputed DNA techniques stemmed from the beating of Taj Patterson in December 2013. A group of Hasidic men attacked Patterson, a black student, in the Williamsburg section of Brooklyn. Prosecutors blamed the attack on the Shomrim, a Hasidic group that patrols Williamsburg, a neighborhood where tensions between Orthodox Jews and blacks have long simmered.
Six days after the attack, the police found one of Patterson’s black Air Jordan sneakers on a nearby roof.
The police sent the sneaker to the DNA lab, where a technician swabbed a 3-inch by 6-inch area of its heel — and recovered 97.9 picograms of DNA from at least two people. A picogram is one trillionth of a gram.
The sample bore Patterson’s DNA. Using software developed in-house, the lab calculated that it was 133 times more likely than not that the remainder belonged to Mayer Herskovic, a young father who lived and worked in Williamsburg and had no criminal record.
“I don’t believe that this is DNA,” Herskovic told ProPublica. “A mixture, like you take milk, orange juice and water and you mix it, what is it? Is it still milk? Is it still orange juice? I don’t know.”
“DNA is the magic word,” he added. “If you throw it into a trial, they eat it up. For me, it’s not magic at all.”
No other physical evidence linked Herskovic to the attack on Patterson, who was blinded in his right eye. Neither the victim nor those who witnessed the crime identified Herskovic at trial, nor was he seen on surveillance video. Herskovic said he has never been part of the Shomrim, and deplored the assault on Patterson.
Three years ago, Barry Scheck, a co-founder of the Innocence Project, a nonprofit that uses DNA evidence to exonerate wrongly convicted prisoners, yelled at his colleagues on the state forensic commission about the potential perils of the DNA work at the city’s lab.
“The day of reckoning is going to come,” Scheck told his fellow commissioners, some of whom rolled their eyes, a video of the meeting showed. “Someday people are going to review this,” he continued. “It’s an Ebola. It is a cancer here that could be spreading. We are all on notice.”
For three decades, forensic DNA evidence has been a valuable tool in criminal investigations, incriminating or exonerating suspects. Matching a defendant’s genetic material with a sample found on a weapon or at a crime scene has proved extremely persuasive with judges and juries.
But not all DNA evidence is equal. Sometimes it’s clear: blood or semen identifies a single person. If it’s just a few skin cells left on an object, or if it contains more than one person’s genetic material, it can be more ambiguous. In such situations, labs used to report that the results were inconclusive, or the defendant could not be excluded from the mix.
New types of DNA analysis have been introduced in recent years to interpret trace amounts or complex mixtures, spawning an industry of testing tools, chemical kits and software. As analysis has become more complex, the techniques and results are coming under fire nationwide.
In the past three years, flaws in DNA methods have temporarily shut down testing in public crime labs in Austin, Texas, and Washington, D.C. Lab analysts “make it seem like it’s a completely objective process,” said Bicka Barlow, a lawyer in California with a master’s degree in genetics and molecular biology. “But I’m 100 percent convinced that there are many people who are incarcerated who were convicted with DNA evidence who are innocent.”
The two techniques that New York’s lab introduced were the “high-sensitivity testing” of trace DNA amounts, and the Forensic Statistical Tool, or FST, in which software calculates the likelihood that a suspect’s genetic material is present in a complicated mixture of several people’s DNA. By its own estimate, the lab has used high-sensitivity DNA testing to analyze evidence samples in 3,450 cases over the past 11 years, and FST in 1,350 cases over the past six. Cases in which both methods were used may be counted in both totals.
In February 2012, responding to a 911 call about gunshots near East Tremont Avenue, police officers from the 45th Precinct in the Bronx saw a passer-by make a motion as if he was dropping an object under a parked car.
His was a familiar face: Johnny Morgan, who had been arrested 75 times. The police found a .40-caliber Glock 23 beneath the car. Morgan was charged with gun possession, based both on DNA evidence and witness testimony. But the amount of DNA recovered from the gun was extremely small; the lab initially said it was unsuitable for testing.
After the prosecutor and the police requested a high-sensitivity test, analysts said Morgan’s DNA was a match. He was convicted.
Public crime labs assessing DNA evidence, including the FBI’s lab, “amplify,” or copy, the material 28 times to conduct their analysis. Under the high-sensitivity testing method developed by Theresa Caragine, a forensic scientist, and implemented in 2006, New York’s lab began to push very small amounts through three more cycles, bringing the total to 31. This approach provided more material to look at — as much as eight times the standard approach. But, like turning up the volume on a radio, those additional cycles amplified small imperfections from missing or contaminated DNA.
To reduce potential problems, the lab decided not to amplify samples smaller than 20 picograms, or about three cells’ worth of DNA, its then-director, Mechthild Prinz, said in 2005 during the state’s approval process for the test. She declined to comment for this article.
“The scientific community has been asked to test more and more evidence with less and less amounts of DNA,” Prinz explained in 2009 to the DNA Subcommittee of the state forensic science commission, which approves all forensic methods used in New York State.
“A couple of years ago, DNA testing was limited to body fluids — semen, blood and saliva. Now every laboratory in the country routinely receives swabs from guns,” other weapons, burglary tools and cash registers, she said.
After several years of high-sensitivity testing of small amounts of DNA, the lab developed a second method: a piece of software to interpret complex mixtures.
Invented by Caragine and Adele Mitchell, a geneticist with a specialty in statistics who joined the lab in 2008, the Forensic Statistical Tool, or FST, considers the overall amount of DNA in the mixture, how many people are in it, how much information is probably missing or contaminated, and the frequency with which each piece of DNA appears in different racial or ethnic groups. Then it compares the defendant’s DNA profile to the mixture, and calculates a likelihood ratio, which it expresses as a single number.
The bigger that number — and it’s sometimes in the millions or even trillions — the more likely that the defendant’s DNA is present. Caragine and Mitchell testified in 2012 that about a third of all test results were favorable to defendants, by indicating that their DNA was probably absent.
Only a small proportion of cases using the Forensic Statistical Tool went to trial. Most defendants faced with unfavorable FST results pleaded guilty, defense lawyers say. “Just the prospect of those numbers going in front of the jury could really warp the plea bargaining process,” said Brad Maurer, a lawyer and DNA specialist at New York County Defender Services.
Eric Rosenbaum, an assistant district attorney and head of the DNA Prosecutions Unit in Queens, described FST as an “extremely powerful tool because it is devastating in court.”
In December 2012, The New York Times profiled Mitchell and Caragine in the article “Helping Decide Guilt or Innocence,” which described their fruitful collaboration, but also hinted at a brewing controversy. The Legal Aid Society was gearing up for an extensive fight against admission of FST results in court.
One interested reader was Eli Shapiro, the former mitochondrial DNA technical leader in the DNA lab. One reason for his early retirement, he later testified, was the stress over having to sign off on lab reports generated by the software. Even in the lab, few people knew the science behind it.
Shapiro later said in court that he found the FST process described in the article “very disturbing.” He reached out to his former boss and colleagues to express his alarm. “They were not concerned,” he testified.
So, in early 2013, Shapiro offered his help to Legal Aid, which had just formed a unit specializing in DNA evidence. Under a judge’s order, the lab had given Legal Aid the results of its validation studies — internal tests of FST’s accuracy. Shapiro helped decipher the data.
“He knows the math,” said Clinton Hughes, a Legal Aid lawyer. “For relaxation, he does long division on the beach with a pencil.”
From 2012 to 2014, a hearing in Brooklyn before Judge Mark Dwyer focused on DNA evidence in two cases: it had been recovered from the handlebars of a bicycle after a shooting, and from the clothing of a sexual assault victim. With the help of testimony from Shapiro and some of the world’s most renowned DNA experts, Legal Aid hoped to persuade the judge to throw out the evidence.
The defense experts were denied access to FST’s software code, which would later come under scrutiny. Instead, they criticized the way that Caragine and Mitchell designed and tested FST.
Bruce Budowle, an architect of the FBI’s national DNA database, testified that New York’s statistical methods were “not defensible.”
He said that FST was designed with the incorrect assumption that every DNA mixture of the same size was missing information or had been contaminated in just the same way. He also criticized the lab’s overreliance on “pristine” saliva and samples to test its methods, which do not mirror the ways real crime-scene evidence is degraded by time and weather. The lab underestimated the challenges, he testified.
“Five-person mixtures can look like three-person,” he said, “four contributors can look like two-person mixtures. It’s almost impossible to actually be accurate.”
The software’s inventors acknowledged a margin of error of 30 percent in their method of quantifying the amount of DNA in a sample, a key input into the FST calculation. They acknowledged that FST didn’t consider that different people in a mixture, especially family members, might share DNA.
In April 2013, weeks after testifying, Caragine was forced to resign from the lab after New York’s inspector general found that she had violated protocol by changing her colleagues’ FST results in two cases. Her defense was that she was correcting their mistakes. Mitchell left in 2014. Caragine declined to comment for this article, and Mitchell did not respond to repeated requests for comment.
Perhaps the most dramatic testimony in the hearing came from Ranajit Chakraborty, who had developed the FBI’s policy on DNA in the 1990s and, as a member of New York’s DNA Subcommittee, voted to approve both high-sensitivity testing in 2005 and FST in 2010. What he had since learned about FST bothered him.
“What would your vote be today?” Jessica Goldthwaite, a lawyer for Legal Aid, asked Chakraborty on the stand.
“My answer would be no,” he said. In November 2014, Judge Dwyer sided with the defense, excluding evidence produced by both high-sensitivity testing and FST. He was the first state judge to do so, and so far the only one.
Appointed to the state forensic science commission when it formed in 1994, Scheck didn’t vote for either of the lab’s methods. His misgivings grew when he learned that the DNA sample used to convict Morgan in the Bronx gun case was only 14.15 picograms. That was below the 20-picogram minimum for high-sensitivity testing the lab had promised to set during its approval process back in 2005.
At the October 2014 commission meeting, Scheck pounded the table as he proposed to compel the lab to turn over any validation studies it had conducted for high-sensitivity testing of especially small samples. He accused lab officials of not having performed the necessary studies, despite their assurances otherwise. While Scheck’s motion failed, it drew a vote from an unexpected supporter: Marina Stajic, who then worked for the medical examiner’s office as the director of the toxicology lab. She supported the motion, she later testified, because she believed that the DNA lab should be transparent with its data.
Her boss, Dr. Barbara Sampson, the chief medical examiner, heard about Stajic’s vote the next morning. She expressed her anger in an email to a colleague, “Hold me down.”
Mimi Mairs, then a lawyer for the DNA lab, emailed, “She sucks.”
A spokeswoman for the medical examiner’s office declined to comment on the correspondence, as did the Manhattan district attorney’s office, where Mairs is now a prosecutor.
In April 2015, Dr. Sampson and Kupferschmid fired Stajic, who had worked at the lab for 29 years. Kupferschmid then called a commission member to inquire whether Stajic would also be removed from the oversight group, according to court documents.
In February 2016, Stajic sued Dr. Sampson, Kupferschmid and the city for allegedly violating her First Amendment rights. The defendants’ lawyer contends Stajic can’t prove why she was fired, and that her vote wasn’t constitutionally protected speech. Her case is pending.
The case that finally revealed FST’s source code began with a few drops of cooking oil.
Kevin Johnson and his ex-girlfriend Octaviea Martin shared custody of two sons, and he sometimes stayed over in her Bronx apartment. One night in April 2015, he was cooking cheeseburgers when some oil spilled. He and Martin argued about cleaning it up.
Her daughter got upset and called 911, telling the dispatcher that Johnson was pointing a gun at Martin. A police search of the apartment turned up two socks wedged between the refrigerator and the wall. In one sock was a black pistol; in the other, a silver revolver.
Johnson, who had been convicted on a previous weapons charge, was arrested.
The lab found that one gun contained two people’s DNA; by FST’s calculation, it was 156 times more likely than not to contain Johnson’s DNA. The second gun had three people’s DNA and a formidable likelihood of 66 million.
Hoping to cast doubt on the DNA results, his lawyers, Christopher Flood and Sylvie Levine, asked for the FST source code, which other lawyers had sought in vain.
Again, the government refused to hand it over on the grounds that it was a “proprietary and copyrighted” statistical tool owned by the City of New York.
The federal judge granted the defense access to the FST code in June 2016 under an order that bars wider disclosure. (The medical examiner’s office denied ProPublica’s public records request for the code, citing its “sensitive nature.”)
Nathaniel Adams, a computer scientist and an engineer at a private forensics consulting firm in Ohio, reviewed the code for the defense. He found that the program dropped valuable data from its calculations, in ways that users wouldn’t necessarily be aware of, but that could unpredictably affect the likelihood assigned to the defendant’s DNA being in the mixture.
“I did not leave with the impression that FST was developed by an experienced software development team,” Adams wrote in an affidavit. Pending more rigorous testing, “the correctness of the behavior of the FST software should be seriously questioned.” Characterizing Adams’ criticisms as merely stylistic rather than substantive, the lab told ProPublica that FST provided reliable calculations.
Technology consultants wrote the software code for FST, according to a spokeswoman at the medical examiner’s office. Few, if anyone, at the lab or on the state’s DNA Subcommittee had the expertise to double-check the software, said a scientist in the lab who worked on the techniques who asked to remain anonymous for fear of career repercussions. “We don’t know what’s going on in that black box, and that is a legitimate question,” the scientist said, adding that evidence in older cases should “absolutely” be retested in light of growing questions about FST. “As a scientist, I can’t say no.”
The U.S. attorney’s office withdrew the DNA evidence against Johnson days before the hearing about its admissibility was scheduled to begin.
Nevertheless, Johnson pleaded guilty this past May. On Aug. 28, he was sentenced to 28 months in prison, almost all of which he has already served. His lawyers declined to make him available for an interview.
As Johnson’s case proceeded, the lab circulated a memo to clients in September 2016, notifying them that it would replace both high-sensitivity testing and FST on Jan. 1. A new chemical kit would make the additional amplification cycles of the high-sensitivity method unnecessary. The lab would retire FST in favor of STRmix, a commercially available and FBI-endorsed software program for DNA mixtures that dozens of public labs use.
The medical examiner’s office “is fully committed to staying on the cutting edge of new technology to best serve the City of New York,” Kupferschmid wrote in the memo. He added that the lab would raise the minimum sample size for testing to 37.5 picograms — almost twice the initial floor of 20 picograms.
The change in policy is scant consolation to those who were convicted based on the discarded DNA techniques, like Mayer Herskovic. After the gang attack on Patterson, two confidential informants gave Herskovic’s name to a police detective. Herskovic was then arrested and swabbed for DNA. Neither informant testified against him at trial.
Sitting at a table in his apartment in Williamsburg, Herskovic discussed the DNA evidence, first calmly and then indignantly. The white walls were bare except for a small mirror, a clock and a portrait of his children, who were scribbling in coloring books on the kitchen floor. He recalled how, when the police asked him to give a DNA sample, his lawyer cautioned him not to, but Herskovic went ahead and did so.
“I was the first one to give DNA,” Herskovic said. “He told me they needed it, I said, ‘Go ahead, take it! It will be better.’”
The DNA on Patterson’s sneaker was pivotal to the case against Herskovic. Patterson testified that whoever pulled off his shoe had punched and kicked him. Although four other suspects were arrested, and several other men were identified by witnesses, seen on surveillance video, or had their license plates photographed at the scene, only Herskovic has been tried or sentenced to prison. Two people pleaded guilty to misdemeanors and were given probation; charges were dropped against the other two.
Herskovic’s four-year sentence was stayed pending appeal. He’s working at an hourly job for a heating, ventilating and air-conditioning company to support his wife and two young children. His appeals lawyer, Donna Aldea, plans to argue that FST was never tested on a population as insulated as the Hasidic Jews of Williamsburg, who very likely share many of the same ancestors, and therefore much of the same DNA.
“This case is a poster-child for how ‘DNA evidence’ can literally be fabricated out of thin air, and how statistics can be manipulated to create a false impression of ‘scientific evidence’ of guilt,” Aldea said. “This must be exposed.”
As floodwaters from Tropical Storm Harvey recede this week, many neighborhoods in the Houston area are reliving a depressingly familiar scenario. Houses in high-risk areas that have flooded repeatedly over the last three decades have likely flooded again, and will require taxpayer funds for their owners to either rebuild, get flood-proofing or be bought out.
Most homeowners who live in risky areas are required to buy flood insurance. Since private options are limited, most get subsidized policies through the National Flood Insurance Program, the federal government’s plan, which covers about 5 million American property owners. But critics say the program’s subsidies encourage people to live in flood-prone areas since they are spared the full cost of insuring them. The program is $24.6 billion in debt and is scheduled to expire this month, pending congressional reauthorization.
This map of Harris County, which includes Houston, shows the number of homes insured by the National Flood Insurance Program in each ZIP code that have filed enough claims from 1978 to 2015 to be placed on a list of severe repetitive loss properties.
More severe repetitive loss properties →
FEMA 100-Year Flood Zone
ZIP Code: 77096
ZIP Code: 77037
ZIP Code: 77034
The data, obtained from FEMA by the Natural Resources Defense Council, shows that the government has shelled out $265 million for flood claims on 1,155 severe repetitive loss properties in the program in Harris County.
In the ZIP code that encompasses the Greenspoint section of Houston (a low-income neighborhood that has flooded at least five times between 2001 and 2017), FEMA has spent $12 million in repetitive loss claims on 71 properties in the program.
For 389 properties in Harris County, payments have cumulatively added up to more than twice the current value of the property. In one extreme case, a house in northeast Houston has received 18 flood insurance payments totaling $1.8 million—more than 15 times the property’s market value.
Properties on the severe repetitive loss list are eligible for voluntary buyouts or mitigation funds that pay to raise structures to higher elevations. But demand far outpaces available funds, so the program usually pays to simply rebuild homes in the same risky spots, without making them more flood-proof.
The map represents the most up-to-date severe repetitive loss information available, but undoubtedly underestimates how many Harris County homes are now on the list. It doesn’t include those added after the massive April 2016 floods — or those that will qualify because of Tropical Storm Harvey.
Map sources: FEMA via Natural Resources Defense Council, FEMA