Reynaldo Rodriguez was 19 with a young son, a good job and no criminal record when he shot and killed a man. As part of an ongoing family feud, someone — Rodriguez believed it was a man named Robert Cuellar — had shot at Rodriguez's mother and brother. Then Cuellar slapped Rodriguez's sister.
"I just blew a fuse," Rodriguez says now of killing Cuellar.
In 1977 he pleaded guilty to second-degree murder, and the judge gave him a choice: A sentence of 15 to 30 years would probably mean parole in 12. A life sentence would make him parole-eligible in 10 years.
Rodriguez chose life. At his sentencing, Saginaw County (Mich.) Judge Gary McDonald made it clear that this was "not the mandatory natural life imprisonment sentence" and said that if Rodriguez was a "model prisoner," McDonald would recommend release in 10 years.
Thirty-seven years later, Rodriguez is still behind bars.
America's prisons hold tens of thousands of people like Rodriguez — people primarily confined not by the verdicts of a judge or a jury but by the inaction of a parole board. Michigan is one of 26 states where parole boards are vested with almost unlimited power to decide who gets out of prison when, and why.
With more than 1.5 million people behind bars, the United States has the highest incarceration rate in the world, and the financial costs are staggering. As politicians from both parties seek alternatives to mass imprisonment, the parole process has emerged as a major obstacle.
A months-long Marshall Project investigation reveals that, in many states, parole boards are so deeply cautious about releasing prisoners who could come back to haunt them that they release only a small fraction of those eligible — and almost none who have committed violent offenses, even those who pose little danger and whom a judge clearly intended to go free.
A recent revision of the Model Penal Code, an influential document written by legal scholars, declared parole boards "failed institutions."
"No one has documented an example in contemporary practice, or from any historical era, of a parole-release system that has performed reasonably well in discharging its goals," a draft of the document says.
Rodriguez's parole file, obtained through a Freedom of Information Act request, includes a scoring matrix that suggests how much time he should serve. Around the country, almost every decision in the criminal justice system is now aided by such data-based tools. But many parole boards eschew them, basing decisions instead on political considerations and gut instincts.
In Michigan, for example, Rodriguez's prior criminal history (zero) was measured against the severity of his crime (6 on a 23-point scale) to produce a recommended sentence of 14 years. But Rodriguez wasn't even granted a hearing before the parole board until 1994, 17 years after he was sentenced.
By then, Rodriguez had earned his GED and worked his way up to head mechanic in the prison shop. He had a letter of support from Judge McDonald and glowing reports from his work supervisors. A prison psychologist noted: "prognosis while on parole is good."
But the board refused to release him and has considered his case only once in the years since.
"Nature of crime as described in public hearing causes further concern," the board wrote.
Because declining to consider a case is not considered a "decision," the board is not required to give any reason.
Rodriguez is now 59 years old.
Parole boards are vested with almost unlimited discretion to make decisions on almost any basis. Hearsay, rumor and instinct are all fair game. In New Mexico, the law directs the board to take into account "the inmate's culture, language, values, mores, judgments, communicative ability and other unique qualities."
The boards' sensitivity to politics stems in part from the heavy presence of politicians in the ranks of board members. At least 18 states have one or more former elected officials on the board. In 44 states, the board is wholly appointed by the governor, and the well-paid positions can become gifts for former aides and political allies.
While some state laws require basic qualifications, these statutes are often vaguely worded, with language that is easily sidestepped. Many states have no minimum requirements at all.
And unlike politicians, who are bound by open records and disclosure laws and are accountable to their constituents, parole boards often operate behind closed doors. Their decisions are largely unreviewable by courts — or anyone else.
"Not only are they closed, they're paranoid closed," said Janet Barton, the former operations manager of Missouri's parole board. "Closed to the extreme."
Few others in the criminal justice system wield so much power with so few professional requirements and so little accountability.
"I received your letter . . . and frankly was surprised you were not released at your last review," Judge McDonald wrote to Rodriguez in 1999, almost 22 years after he went to prison. "I am sure you know that I have no power over the parole board and there is nothing I can do in that regard."
In many states, the boards' most basic workings are shielded by law from public view. Boards are not obligated to give any but the most cursory reasons for their decisions, which include not only whether to release prisoners but also how long they must wait to be considered again or what they can do to increase their chances in the meantime.
In 24 states, boards need not disclose what material they relied upon to reach their decisions. Courts have consistently upheld this prerogative, ruling that inmates have no due process right to parole.
"I have no idea as to what is expected of me to gain meaningful consideration from the parole board again," Rodriguez wrote in an emotional 2003 letter to the board.
At the bottom, someone had stamped: "No response necessary."Paradox for parole boards
Prisoners like Rodriguez represent a paradox for parole boards: Older inmates who have committed the most serious crimes, and served the longest terms, are the least likely to commit new crimes upon release.
One Stanford University study of 860 murderers paroled in California found only five returned to prison for new felonies, and none for murder.
This is especially true for older prisoners. Recidivism rates drop steadily with age. And older prisoners are more expensive: The average annual cost per prisoner doubles at age 55 and continues to climb thereafter.
Still, these prisoners are consistently the least likely to be paroled. Though they pose a low risk of future violence, the political risk of releasing them is huge. Parole board members are routinely pilloried in the news media and chastised by the public. Many have lost their jobs for releasing people whose crimes were violent.
"There's some offense conduct you just can't outrun," said William Wynne, a member of the Alabama parole board.
Take Keith Drone, who has been denied parole five times — including once when the Missouri board granted parole and then took it away.
Drone was 17 when he and his friends tried to rob a motel. Although no one was killed or critically injured — except Drone himself, who was shot in the head — he received three life sentences, in part because an off-duty police officer, acting as a security guard, was shot in the leg during the melee. Under Missouri law, "armed criminal action" enhancements carry a life sentence, though Drone's plea bargain allowed him to serve the sentences concurrently.
Drone is now 44. "I'm not exactly an old man yet," he said. "I really don't want to be one of those guys."
After each parole hearing, Drone has received the standard line from the board: "Release at this time would depreciate the seriousness of the present offense based on A: circumstances surrounding the present offense."
Of the 10 factors parole board members weigh most heavily in their release decisions, five are related to the crime itself, according to a nationwide survey of parole boards in 2008. The top two are "crime severity" and "crime type."
Eight of the 10, including the offender's criminal history and gender, were known to the judge at the time of sentencing.
Those paroled before the end of their sentence are subject to supervision in the community. The parole board can dictate the terms of release, including how often they must meet with a parole officer and whether they must seek treatment for drug or alcohol abuse.
But those continually denied parole eventually serve their full sentence — they "max out," and the state loses its legal ability to supervise them. The people deemed too dangerous to release therefore become the very people released with no conditions and no supervision.
"When a person is parole-eligible, if they meet the qualifications, if they've done the programming, if they pose a manageable risk, then you want to parole them at the earliest point possible," says Catherine McVey, the former chairman of Pennsylvania's parole board.
Over the past two decades, however, the number of inmates who max out has more than doubled, according to a recent study by the Pew Charitable Trusts. In 2012, one in five state inmates was released from prison without supervision.Assessing the risk
Determining whether an offender poses a manageable risk is a major preoccupation in criminal justice circles. At many steps in the process — from sentencing to probation — offenders' histories are plumbed to produce data-driven, research-tested assessments of the threat they pose to public safety.
In the past five years, at least ten states have passed laws requiring parole boards to use risk assessment tools and other quantitative guidelines. Many other state boards also use them.
But the quality of the tools varies widely, and even high-quality tools are often ignored.
A Missouri parole board staffer filled one out for Drone in 2001, after he had been in prison for almost 13 years. Because Drone had no criminal history, his score was almost perfect: 9 out of a possible 11. A copy was attached to his parole rejection notice.
Fourteen years after that first risk assessment, Drone was finally granted parole, but his release was set for 2017. He has no idea why and no way to find out.
Some states require board members who depart from the guidelines to explain why. In New Hampshire, lack of a financial safety net or a safe place to live is "the most common reason we would hold somebody back with a decent score," said Donna Sytek, who chairs the New Hampshire board.
In other cases, it's impossible to know whether a departure from the guidelines was for a good reason.
"There's not a way to go back and look at that file to find out why they deviated from the parole decision guidelines," said one former member of the Georgia parole board, who requested anonymity to discuss the board's inner workings. "They don't have to answer to that. They don't have to answer to anybody."
Many state parole boards claim exemptions from state sunshine laws that would be unthinkable for other government officials.
In 19 states, some or all parole board hearings are closed to the public. In 24 states, board files and documents are sealed. In 18 states, parole files are not available to the inmates themselves.
Georgia, for example, classifies board records as "state secrets," immune even from subpoena. This spring, Gov. Nathan Deal (R) signed a measure intended to increase transparency, but lawmakers jettisoned a provision that would have required the board to explain its decisions.Leaving prisoners in the dark
Some state boards are more transparent. Connecticut broadcasts some of its parole hearings on public access television. In New Hampshire, Sytek scrapped a long-standing rule barring cameras from board hearings.
"In order to have confidence in their government, the public needs to know what their appointed officials are doing," Sytek said. "They don't trust what they can't see."
States that give inmates access to their files almost always make an exception for victim statements and other letters of protest.
"Someone can be kept in prison indefinitely and could never have a clue that the victim is saying something that may be untrue," said Barbara Levine, founder of a Michigan nonprofit devoted to reducing the prison population, and a member of the state's new Criminal Justice Policy Commission.
This may have happened to Roosevelt Price, a Missouri prisoner who last came up for parole in 2013. At his hearing, Price said he was stunned by a skeptical board member.
"I think you've been involved in other murders that you haven't been caught for," she told him.
By then, Price had served 20 years of a life sentence for shooting and killing his brother-in-law in a fight. His only other run-in with the law had also involved a long-standing feud with his wife's family. He had never been accused of another killing.
"I don't know where you're getting that from," Price said.
"There's things in your file I know about that I think you don't know," the woman responded.
Missouri officials declined to provide additional information.
"Several state statutes prohibit the release of this information," Missouri Parole Board chairman Ellis McSwain said in a written statement. He cited laws dictating that parole-related documents "shall not be disclosed" and that any parole-related meeting or vote "may be . . . closed."
When Judge McDonald sentenced Reynaldo Rodriguez to life, he expected Rodriguez to get out in 10 years. This was, for much of modern U.S. history, how criminal sentencing worked.
Judges would hand down an indeterminate sentence with a range of years, such as five-to-10 or 25-to-life. The lower number was generally meant to serve as the punishment portion of the sentence. The balance could be served on parole if the inmate could prove he had been rehabilitated.
"The idea was, you should only stay in as long as you needed to," says Peggy Burke of the National Parole Resource Center. "The parole board was in a good position to judge that."
Murderers released in 1981 served a median term of about five years before they were paroled or had their sentences commuted. The tough-on-crime 1990s changed that.
Criminologists began to tout the notion that rehabilitation was impossible. Therefore, the thinking went, the best option was to keep violent criminals off the streets as long as possible. "Abolish parole" and "life means life" joined the political lexicon.
Fourteen states and the federal government eliminated or severely restricted parole. These states turned instead to "determinate" sentencing schemes, in which the sentences handed down in court dictate exactly when an inmate will go home, and how long he or she will be supervised afterward.
Boards that retained the ability to release people, meanwhile, became increasingly reluctant to do so. In the early 1990s, the New York state board voted to parole more than 60 percent of those eligible. That rate then went into a two-decade decline, dipping below 20 percent in 2010.
When Rodriguez appeared before the Michigan parole board in 1994, the release of lifers in that state had slowed to a trickle.
In 1997, the Michigan board published a report trumpeting its transformation into "a Parole Board that is much less willing to release criminals who complete their minimum sentences — and much less willing to release criminals at all, forcing many to serve their maximum sentences."
Lately, the national mood has swung away from the punitive excesses of the 1990s. But those who study the issue say parole boards continue to release far fewer people than they safely could.
That conclusion is largely anecdotal: There is no national data on parole grant rates. Each board calculates its rate differently, and some states don't release it at all. There is also little hard data on recidivism among parolees vs. those who max out.
"Despite the best efforts of parole authorities, there is little empirical evidence to support the proposition that we can effectively distinguish those offenders who are truly rehabilitated from those who merely 'talk the talk,' " said a recent report by the New York state Sentencing Commission, which recommended that New York eliminate discretionary parole entirely.Crunched for time
Courts understand parole to be an act of grace rather than a right. Legally, boards can — and do — make decisions for almost any reason.
Board files may contain "observations of guards, counselors, and other corrections personnel. Even unsubstantiated rumors may appear," criminologist Neil P. Cohen wrote in his authoritative reference book "The Law of Probation and Parole."
Many boards deny parole based on crimes that were never prosecuted or facts that were never established before a judge or jury.
"Their offense on paper may be a property offense, but they may have had a series of other offenses that just overshot that," said Wynne, the parole board member in Alabama. "I've seen burglary thirds that were pled down, that [started as] a burglary first and a rape. You're going to treat that differently from a guy that walks in and steals a lawn mower out of a shed."
Time is one of the many factors that prevent board members from making thoughtful decisions. The 2008 parole board survey revealed that the average state board considered 8,355 inmates for release each year. That's about 35 decisions per workday for a board that usually has other responsibilities.
"I typically voted 100 cases a day. That was just an average day," said the former Georgia parole board member, who spoke on the condition of anonymity and who had urged a "system overhaul." "You're just talking about two to three minutes to make a decision. The public would be astounded at the short period of time that the board has to make decisions on life and death cases."
The parole process varies widely by state. Some boards review files and conduct interviews, either in person or by video; some only review files. Some meet to discuss cases; others simply pass files from one board member to the next until they accumulate enough votes to support a decision. That's the system in Georgia.
"Most families have it in their mind that the board sits down and has a hearing as it relates to a case. And that's not correct," the former Georgia board member said.
Janet Barton, who worked as operations manager of the Missouri board for 30 years, said some members never examined the files, basing their decisions instead on how others voted.
As a file changed hands, "a board member may not even look at that case, other than to say, 'I trust Richard. So I'm going to agree with him,' " Barton said. The next board member might say, " 'I don't trust a thing that guy does. He's a liberal decision-maker. I'm not even going to look at the case. I'm going to just deny.' "
For a time it was Barton's job to fill out the form that communicates the board's decision to the inmates. She became increasingly disheartened, she said, by how difficult it was for people with violent convictions to get parole, even decades after their crime.
Their forms would always say the same thing: "Release at this time would depreciate the seriousness of the present offense."
But that was "not always the truth. Sometimes I'd make that crap up. The real reason," Barton said, was "we don't believe in parole for people like you."
Members of the Missouri parole board declined repeated requests for comment.
"Our statute is that parole and probation records are closed," said Corrections spokesman David Owen.Lost confidence in parole
In December 2010, the Massachusetts parole board faced a scenario every board dreads: A man the board had voted unanimously to release went on to commit another terrible crime. During an armed robbery, Dominic Cinelli killed a police officer.
Headlines blared: "Massachusetts Cop Was Killed by Career Criminal Out on Parole Despite Three Life Sentences," Fox News wrote. Lawmakers and police called for decisive action.
Privately, Gov. Deval Patrick (D) assured board members they had done nothing wrong. He urged them to cooperate with investigators from his office, who ultimately blamed inadequate supervision by Cinelli's parole officer, rather than the board's decision to release him.
Still, when board members arrived at work days later, armed troopers escorted them to a conference room where they found Mo Cowan, the governor's chief of staff, distributing resignation letters, according to a wrongful-termination lawsuit filed by one of the board members.
Patrick still believed they had done nothing wrong, Cowan told them, but he was asking the entire board to resign nonetheless.
"The public has lost confidence in parole, and I have lost confidence in parole," Patrick said later.
In 44 states, parole boards are appointed by governors. Only three states recruit publicly for the job, which typically comes with a handsome salary and government benefits. Parole boards nationwide heed cases like Cinelli's as a warning.
In Massachusetts, where the average board member's salary is more than $100,000 a year, the new board got the message: The year before Cinelli killed the police officer, the release rate was 42 percent. The year after the shooting, it was 26 percent.
Mark Conrad, the chair of the parole board that oversaw Cinelli's release, was a former police officer with two degrees in criminal justice — and he was the governor's former driver. In many states, the most obvious qualifications of parole board members are their political connections to the governor.
A Louisiana board member, for example, served on the Governor's Commission for Marriage and Family and as a lobbyist for the Louisiana Family Forum, which has close ties to Gov. Bobby Jindal (R). A New Jersey board member was a Senate staffer, then chief of staff to former governor Richard Codey (D) before he appointed her.
One of the American Correctional Association's "essential" standards for parole boards is that at least two-thirds of members have at least three years of experience in criminal justice or a related position. But the ACA declined to clarify those guidelines. "We're not going to weigh in," said ACA spokesman Eric Schultz.
Without an objective standard, it's hard to say how many state parole boards meet the ACA's benchmark.
The boards in Arkansas (which is ACA-accredited), Delaware, South Carolina, Tennessee and Virginia seem to fall short. Members in these states included farmers, executives at the chemical company DuPont, an automotive broker and personal fitness trainer, a pastor, and an entertainment and event manager.
In at least seven other states, board members' official biographies do not make clear whether they have criminal-justice experience. Four states — Wisconsin, Oklahoma, New Mexico and Indiana — refused to provide board members' biographies.
Parole boards have recently begun to emerge as an area ripe for reform. In the past five years, at least 16 states have passed overhaul bills, including six that set out minimum qualifications for members, such as a bachelor's degree.
But many of those requirements are broad, allowing for experience in fields such as "education" that don't necessarily bear on the complexities of criminal behavior, risk assessment or the local prison system.Facing political pressure
Given the close relationship between politics and parole boards, it's not uncommon for board members to lose their jobs — or fear losing their jobs — for making decisions that are politically unpopular.
"You generally don't get reappointed if you take a controversial stand on a media case. And most cases involving law enforcement personnel become media cases," said Thomas Grant, a former member of the New York parole board. The New. York tabloids have a close connection with the police officers' unions, Grant said, and "they can generate a Page One story pretty quickly."
Grant should know. In 2009, the Daily News ran a story headlined: "Cop killers' pal: Parole Board's Thomas Grant keeps voting to turn 'em loose." In 2010, the paper asked: "Has state parole commissioner Thomas Grant ever met a cop killer he didn't want to put back on our streets?"
Grant was not reappointed after his first term.
Victims' rights groups, too, now wield considerable influence in states around the country.
"The heavy pressure for being super conservative is from your victims' groups," said McVey, the Pennsylvania board chair. "The advocacy groups are stronger than ever."
The vast majority of the nation's parole boards are required to hear victim input before making a decision, according to the 2008 survey; 40 percent said victim input is "very influential."
In Alabama, it's almost unheard of for the board to grant parole over victim opposition. The board also routinely receives letters opposing parole from the governor, the attorney general and other elected officials.
"That's going to impact the disposition when they protest, almost universally," Alabama board Chairman Robert Longshore said. "You've got a very politicized victim community in the state of Alabama."
In Maryland, Oklahoma and California, the governor has the power to reverse a parole decision, taking a process already implicitly fraught with politics and making it explicitly so.
Dorian Maddox was caught in that emotional grinder. After extensive psychological workups, the Maryland parole board voted to release him in 2005, only to have the decision reversed by Gov. Robert L. Ehrlich Jr. (R) in 2007.
When he pleaded guilty to first-degree rape and was sentenced to life in 1976, Maddox said, "the state's attorney and my attorney both agreed that I would accept this plea and after 20 years I would be paroled."
Thirty-eight years later, he is still waiting.
An earlier version of this story described the average time served for murder in 1981 using a number that included manslaughter convictions; the numbers have been updated to reflect only those convicted of murder.
This article originally misstated when Michigan Judge Gary McDonald retired. It was 2003.