A consensus is building around the need to seriously rethink the role of the prosecutor in the administration of justice. Power dynamics are unbalanced, sentencing guidelines are outdated, and old-fashioned human biases persist. And prosecutors—singularly independent agents in a justice system roiling in turmoil—have been facing growing criticism and public distrust for some time, and that disapproval is about to hit a tipping point. It’s time to curtail the power long held by these officers of the court as they promote justice, ensure fairness, and enhance public safety.

In all but four states, prosecutors are elected to office—about 2,400 of them—and many largely run unopposed in counties with strong political-party identification, places where politics are so ingrained that decisions are made well ahead of the voting booth. Party loyalty, tough-on-crime promises, or just electoral indifference—can result in prosecutors who amass and wield enormous amounts of power in the courtroom and beyond. And where there’s unchecked power, there’s often corruption, particularly in high-profile cases. In some instances, entire districts have been tainted by a reputation earned during a particularly bold district attorney’s tenure. These and other regional struggles between justice and authority have developed over decades of individual and cumulative legal acts in thousands of cities and districts.

In the meantime, the American prison population has ballooned, a phenomenon that until recently was blamed on socioeconomic factors, harsh federal and state guidelines, or other macro-level explanations. But, as Occam’s razor might suggest, the real explanation for a growing prison population is much simpler: growing prosecutorial power.

***

To get some context, I called John Pfaff, a professor of criminal law at Fordham University who studies mass incarceration. “We’re not admitting a few people to an ever-growing pool of long-term inmates; we’re basically cycling people in,” Pfaff said. “Every year, about half the prison population leaves, between 600,000 and 700,000 people. But we also admit about the same amount every year. It’s a giant churning process.” Between 2000 and 2012, the American prison population rose by about 100,000, from 1.5 million to 1.6 million, but that happened with a staggering 8.4 million admissions to prison, which included scores of both repeat and new offenders. “What my results show is that most people are in prison for a very short period of time, and if we simply cut back on admissions, it would lead to a very quick decline in our prison population,” Pfaff said. “As much as people sometimes seem surprised by that, it’s a factor hiding in plain sight.”

So what caused this giant increase in prison admissions? Pfaff started looking into it and realized that no one had answered the question well. He dug through reams of information and concluded that most academic and research work on the topic approached the question the wrong way: They tried to see what societal and sociological factors caused the prison population to grow. But they should have asked a more elementary question: Who was responsible for it in the first place? Many well-executed studies premised their findings on race, politics, and economics, among other sociocultural factors. Many an analysis looked at trends in crime, in arrests, and in prison admissions per crime. In many ways, those approaches were “concessions to the data,” according to Pfaff. Then he realized: “We have no data on prosecutors. The studies kind of skipped over that. That struck me as a problematic omission. I decided to figure out what role prosecutors play. I saw that we needed to add them back in.”

In a movie-worthy moment, Pfaff came across a data set that was, essentially, sitting in view for all to see—but no one had used it. By analyzing filings in state courts, “We discovered that once you add DAs back in, the picture changes completely,” Pfaff said. The data he studied came from about 35 state courts based on case filings by prosecutors. “When you look at it on a map, it looks fairly representative,” Pfaff said. “It’s got big southern states, small southern states, big northern states, western states, eastern states—it’s got a fairly good sample of the U.S.”

Over the 1990s and 2000s, prosecutors started charging a lot more arrestees with felonies. According to Pfaff’s data, in 1994 about one of every three arrests turned into a felony case, and by the end of the 2000s, it was two out of every three arrests. “No one had really noticed how central prosecutors were, just how important they were,” Pfaff said. He asked himself where did the increase come from. Using a Bureau of Justice Statistics set, Pfaff found, “Roughly speaking, half of this increase in felony filings comes from cases of misdemeanors being charged as felonies.”

So in most places, prior to the 1990s, huge swaths of arrests would not convert to felony cases and would never lead to prison time in the way they do now. “When I broke it down state by state,” said Pfaff, “what I found is that from almost all the states—with one or two idiosyncratic exceptions—they’re all following the same basic trends over the course of the 1990s and 2000s: this steady increase driven by felony cases.” And all of those trends could be traced back to prosecutors. “They can choose how harshly to go after someone, how lenient to go after someone,” Pfaff said. “They have tremendous power in that respect.”

***

Right now, New Jersey is the only state where district attorneys are appointed—and thus not so likely to have endless terms. They’re designated by the state’s attorney general, who is in turn, appointed by the governor. New Jersey also has plea-bargaining guidelines in drug cases. If a prosecutor wants to plead out someone to certain categories of serious drug offenses, she has to consult a chart that lists the arresting offense and the offender’s prior personal history to generate a range of pleas she can offer. Included in the calculations are mitigating and aggravating factors that can raise or lower the range. “As far as I know, New Jersey is the only state that’s done this, and no state even talks about this,” Pfaff said. “New Jersey has shown that if you want to, you can create guidelines that appear to regulate how DAs propose pleas.”

Former attorney general Eric Holder took an important step toward creating a federal standard in this regard by issuing U.S. attorneys new guidelines called Smart on Crime, an administrative umbrella under which multiple efforts are targeting reforms aimed at reducing the federal prison population. “Until 2010, federal prosecutors were required to pursue the most serious, readily provable charges in every case and for every defendant,” according to a Department of Justice brief. In short, the government expected prosecutors to throw the book at offenders. But Holder’s new instructions asked them to use more discretion when asking for mandatory minimums, even specifically instructing them to remove references to quantities of drugs that may trigger such harsh minimums.

“I’m not going to tell you that every single prosecutor out there would have written the Smart on Crime policy him or herself, but what I can tell you is we know they’re doing it. The stats show that,” Deputy Attorney General Sally Yates told me recently in her office at the Department of Justice. “Particularly for prosecutors who’ve been doing this for a long time, we’ve seen that mandatory minimums done the old way cast too broad a net because they focus just on one feature—drug quantity—and doing so doesn’t distinguish between the drug kingpin and the courier. Prosecutors who have been doing this for a long time have recognized that.”

The number of federal drug defendants in 2012 was 24,563. In 2015, it was 19,792, a drop of 4,771, according to the DOJ. There was a 14 percent drop in federal drug cases between 2012 and 2014, and another 6 percent drop from 2014 to 2015, which the department sees as “a steady downward trend” in low-level offenders and proof that Smart on Crime is having an impact. Conversely, while low-level federal prosecutions are down, when it comes to more serious crimes, the percentage of drug cases in which defendants had a weapon rose from 15.1 percent in 2012 to 16.4 percent in 2014, and to 17.3 percent in 2015. In other words, low-level federal defendants aren’t getting the book thrown at them as often and more time is spent prosecuting higher-level federal offenses. “We’re charging mandatory minimums 25 percent less frequently now than two years ago before Smart on Crime. That is a very significant shift. We’re seeing that all across the country,” Yates said.

Pfaff is familiar with the initiative, but he was quick to cast some doubt on its efficacy. “Smart on Crime only applies to the federal system. The federal system only has about 13 percent of [all U.S.] prisoners,” he said. “It’s a fairly small fraction, and it’s a weird system that focuses much more heavily on drugs than on violence, because that’s all the feds can do. The feds can’t get after murders and arsons, so 49 percent of all federal prisoners are in prison for drug crimes, but 55 percent of all state prisoners are in prison for violent crime. It’s a much different scenario.”

It’s also essentially voluntary. The DOJ has no power of enforcement after it has issued guidelines. “They’re expected to adhere to the new guidelines 100 percent of the time. This is a mandatory policy,” Yates said. But without enforcement measures and given that presidential administrations change hands every four to eight years, there’s little incentive for prosecutors to play by the Smart on Crime rules. Yates did readily concede that 100 percent compliance couldn’t be expected. She was also clear on the idea that using more discretion does not equate to more leniency or fewer prisoners. “It’s got to be sufficient for punishment and public-safety purposes, but also fair and just,” Yates said. “The more the public can see that that’s how we view our mission, the better the trust will be between the public we serve and their Justice Department.”

***

David Schleicher, a professor at Yale Law School, is researching whether crime rates or conviction rates have any effect on elections for district attorney. “Our preliminary finding is they do not,” he said. DA elections mainly happen during primaries in counties with low turnout, where voters are not very informed about the rest of the candidate field beyond the more prominent races. Even if they are informed, though, it may not matter: In Oregon, the local chapter of the American Civil Liberties Union took a look at the decade between 2004 and 2014, and determined that 78 percent of district attorneys were elected in uncontested races. David Rogers, the author of the study and the Oregon ACLU executive director, concluded that DA elections and appointments “lock in the criminal-justice status quo, preventing much-needed progress and public engagement.”

In New York City, “crime rates have a pretty big effect on mayoral popularity ratings and, therefore, on elections for executives,” Schleicher said. One of his findings is that in partisan executive races, voters follow the party line explicitly. An implication of this is that more-prominent executive races, such as for governor or mayor, have greater voter responsiveness than lower, down-ballot races. This kind of mindless voting at the DA level means that despite the appearance of being beholden to the electorate, district attorneys are not actually held responsible for their successes and failures. Schleicher argues that if district attorneys were appointed by mayors instead of being independently voted into office, it may have the ironic effect of making their role more accountable. Since the fate of the citywide executive would be directly tied to crime outcomes, district attorneys would be under serious top-down pressure to make smarter and less expedient decisions. That dovetails with the other advantage of a mayoral appointment: Resources could be reallocated for strategic crime-fighting. “If you actually cared about crime rates, you would think to yourself, ‘How should I allocate resources in order to reduce crime rates? That’s what’s going to get me reelected or not reelected,’” Schleicher said. “DAs have no incentives to do that.”

District attorneys do, however, have an incentive to prosecute and send people to state prison—because state prisons do not spend local county resources, so district attorneys’ budgets stay intact. California is putting solutions to that problem to the test. In 2011, it passed what’s known as realignment legislation, which reassigned some types of crimes so that those convicted would be sent to county jails instead of to state prisons to serve out their sentences. What happened? “DAs started charging these crimes less because they were affecting county budgets,” Schleicher said. He’s so interested in the potential of this idea that his latest research project, with Elina Treyger at George Mason University Law School, is exploring what impact a more substantial realignment effort might have on incarceration rates, budgets, and other resources. “You could imagine that effect would be supersized if it was housed in one office; if it had a collective budget,” Schleicher said. Though he’s emphatic that his work is very preliminary, he was comfortable alluding to at least one working hypothesis: “I think we can say, in theory, that having a DA being an elected—versus being an appointed—official is going to [make him] more responsive on exclusively criminal-justice issues.”

Even so, it would still be the case that prosecutors would have a lot of the powers they currently have—who they charge, why, for what offense—because state legislatures impose “a bunch of ridiculous laws,” according to Schleicher. What’s more, there is some evidence that prosecutors push back against reforms. An Urban Institute report analyzed South Dakota’s criminal-justice reforms and found that when certain low-level crimes were no longer eligible for lengthy sentences, prosecutions against crimes that were eligible for such sentences shot up. “They noticed that DAs actually started increasing their charging in an effort to get around reform,” Pfaff said. He said that if states make it difficult to charge offenders with one type of crime, DAs will choose another option and use it widely. Sometimes, that results in even tougher sentences than the DAs might have originally sought. “There is some evidence that DAs will, in the presence of reform laws, try to figure out ways” around those reforms, Pfaff said. “They often have the ability to find ways to circumvent efforts at reform if they really want to.”

And district attorneys still have more than enough authority to prosecute offenders as they see fit. “They remain the only actor who is subject to almost no regulation at all,” Pfaff said. “They have tremendous amounts of power. In states without sentencing guidelines, there are no rules about what sort of sentences they can impose. They can choose who gets charged, and who doesn’t, with no review. They can choose what charges to file.”

***

Prosecutor, heal thyself?

“Prosecutors’ offices are mostly a black box with little transparency,” according to Rogers and to most critics. “One exception is the Manhattan DA’s office,” Pfaff noted, “which actually allowed Vera Institute to do a study with their files. That’s what it takes. A DA has to basically open up his office to a researcher to look at this. You can’t just sit down and download the data yourself.”

Or can you? John Jay College of Criminal Justice recently announced its new Institute for Innovation in Prosecution, headed by former prosecutor Meg Reiss. A joint project from the Manhattan District Attorney’s Office and the college, the New York City-based institute will “develop programs designed to support innovation in the role of prosecutors in the American justice system.” Reiss, a jurist with two decades of experience, has great faith in what her former colleagues can accomplish. She also owns up to the negative perceptions typically tied to the role. “There’s a lot of mystery about what actually goes on in a prosecutor’s office, so people have never been able to really evaluate it and see exactly what it is they’re doing,” she said. But transparency alone won’t solve some of the problems. Most people don’t have a good set of criteria with which to judge the performance of their elected officials, beyond things areas have a material impact on them, so asking voters to be informed enough to select and unselect such a specialized position as district attorney is a stretch.

Reiss said part of the solution is giving prosecutors better tools with which to do their jobs—with “a lot more discretion and creativity.” She said some crimes should fall into categories like “alternatives to prosecution” and “diversion programs.” “Of course, you address violent crime appropriately, and no one’s saying that you shouldn’t,” Reiss said. “But I think everything needs to be carefully evaluated and understood. There isn’t a broad stroke that you use for every type of crime or every type of person.” She cites the intelligence-driven prosecution model out of the Manhattan district attorney’s office as a good place to start: “DA Cyrus Vance is holding a symposium, one of many they’ve done already, on a crime-strategies unit that he has set up in his office, teaching other officers around the country how to do the same.”

It might take that kind of colleague-to-colleague approach to change the prosecutorial culture in the United States. “The bottom line is people came to be prosecutors because they really wanted to ensure fairness and increase public safety,” Reiss said. “They have a real moment at this time to step up and make a big change, to really lead in this effort, to be really innovative and forthright in their intentions, to reduce mass incarceration, to address racial disparity in the system, to look for alternatives to oppressive sanctions. We missed so many things and now is the moment.”