Shondel Church was arrested in Kansas City, Missouri, last July, accused of stealing a generator and a tool box from his stepmother. He sat in Lafayette County Jail for six weeks before his first conversation with a public defender, Matthew Gass. Gass was reportedly hopeful that he could win the case at trial, but explained that the intensity of his workload meant he would need six months to prepare—six months during which Church would remain jailed. As a father of four and his family’s primary breadwinner, Church felt he couldn’t wait that long and instead pled guilty to a misdemeanor. He received two years of probation and a $2,600 bill for his stay in pretrial detention.

Ninety-seven percent of federal cases are settled the way Church’s was, by plea bargain. State-level data suggest similar numbers nationwide. Though access to a public trial is enshrined in the Sixth Amendment, taking a plea forecloses that possibility. “This constitutional right, for most, is a myth,” U.S. District Judge John Kane wrote in 2014—one voice among a chorus of jurists, advocates, and academics all calling for reform. Some want tweaks to the regulation and oversight of pleas; others urge more ambitious overhaul of the way trials are conducted, streamlining the process to make it accessible to greater numbers of people.

Plea bargains were almost unheard of prior to the Civil War. Only in its aftermath, as waves of displaced Americans and immigrants rolled into cities and crime rates climbed, did appellate courts start documenting exchanges that resemble the modern practice. The plea became a release valve for mounting caseloads. Appellate courts “all condemned it as shocking and terrible” at the time, said Albert Alschuler, a retired law professor who has studied plea bargains for five decades. The courts raised a range of objections to these early encounters, from the secretiveness of the process to the likeliness of coercing innocent defendants. Pleas, wrote the Wisconsin Supreme Court in 1877, are “hardly, if at all, distinguishable in principle from a direct sale of justice.”

The practice nonetheless continued, and, by the turn of the century, a minor economy had settled in its orbit. “Fixers” could be hired to arrange for alternatives to a prison sentence. Police regularly toured jails to “negotiate” with the inmates. One New York City defense attorney and friend to local magistrates loitered in front of night court hawking 10 days in jail for $300, 20 days for $200, and 30 days for $150. By the 1920s, as violations of the federal liquor prohibition flooded court dockets, 88 percent of cases in New York City and 85 percent in Chicago were settled through pleas. When the Supreme Court in 1969 finally heard a case concerning the legality of the issue, it unanimously ruled that pleas are constitutionally acceptable. They are “inherent in the criminal law and its administration,” the Court declared.

A few justifications are used to explain the widespread use of pleas. In cases that involve organized crime, prosecutors can use plea bargains to advance the case, extracting information from low-level offenders and pushing further up the criminal hierarchy. Pleas can also provide genuinely good deals to people facing long prison sentences. Most fundamentally, basic economics supports their use. Trials are expensive and protracted. Two rational parties, goes the logic, can more cheaply and quickly come to an agreeable outcome through stripped-down bartering: The prosecutor offers a lenient charge if the defendant foregoes trial and admits guilt.

This final rationale raises tough moral questions, which were perhaps best articulated by Chief Justice Warren Burger in 1971: “An affluent society ought not be miserly in support of justice, for economy is not an objective of the system,” he wrote. The court, in other words, should prioritize its profound responsibility to sort the guilty from the innocent over the efficient dispatch of criminal defendants. (“Miserly” may be how Church would describe the state’s dealings with him in Missouri; he’s involved in a class-action lawsuit that argues its understaffed public-defender system doesn’t provide sufficient legal counsel.)

But there is also a central practical concern reformers want to mitigate: that spare oversight of the process invests prosecutors with broad, opaque powers. Judges are not regularly allowed to take part when a plea deal is made, and written records of a deal are almost never required. Though jury trials demand proof of guilt beyond a reasonable doubt, pleas follow no standards of evidence or proof; the prosecutor offers a break in exchange for a guilty plea, the defendant decides whether to take it without knowing the merits of his case.

Indeed, the only bargaining restriction placed on prosecutors is that they cannot use illegal threats to secure a plea. “So if a prosecutor says, ‘I’ll shoot you if you don’t plead guilty,’ then the plea is invalid,” Alschuler explained. “But if he threatens to charge someone with a crime punishable by death at trial and the defendant pleads guilty, then the plea is lawful.” Assuming they have probable cause, prosecutors can even threaten to bring charges against a defendant’s family in order to extract a plea. For instance, if a defendant’s spouse or sibling is complicit in drug trafficking—perhaps they took a call related to the case—a prosecutor can offer to reduce or dismiss charges against the family member if the defendant pleads guilty.

This dynamic, combined with national trends over the last 30 years favoring lengthy mandatory sentences, gives prosecutors inordinate leverage. If a defendant considers going to trial, a prosecutor might hang overhead some charge that carries a mandatory life sentence. A plea of guilty might instead get eight years, or 10 years, “or pick a number,” said Matt Sotorosen, a senior trial attorney at the Office of the San Francisco Public Defender. “Even if you have an innocent client, most don’t want to take that chance. They’ll just take eight years. What if things go south at trial?” The results of this lopsided calculus are evident in data from the National Registry of Exonerations: Of 2,006 recorded exonerations since the project started keeping track in 1989, 362 of those, or 18 percent, were based on guilty pleas.

In theory, abolishing the use of plea bargains wouldn’t take much: Prosecutors would simply stop offering deals. That would be that, though the massive influx of trials would jam courts. (Michelle Alexander, author of The New Jim Crow, discussed defendants’ deliberately going to trial and “crashing the courts” as a form of resistance to mass incarceration.) But both sides of the debate agree the odds of this happening are infinitesimal. Even Alschuler, who throughout his career remained one of the staunchest critics of plea bargaining, admitted in 2013 that “the time for a crusade” had passed. Instead, he suggested people work to make the criminal-justice system “less awful.”

Consistent with this, reformers are exploring two avenues to make plea bargaining either more accountable or less common: The process could be altered to afford defendants more protection, or the jury trial could be simplified to ensure more people take advantage of this right.

“Plea bargaining in the United States is less regulated than it is in other countries,” said Jenia Turner, a law professor at Southern Methodist University who has written a book comparing plea processes in several U.S. and international jurisdictions. As a result, states are independently adopting measures to inject the process with more transparency here, more fairness there. In Connecticut, for example, judges often actively mediate plea negotiations, sometimes leaning in with personal opinion on an offer’s merit. In Texas and North Carolina, along with a few other states, both sides share evidence prior to a plea.

Turner suggests that replicating some of these practices across state lines, or standardizing the plea process nationally, could go a long way to equalizing the power between defendants and prosecutors. She also argues that agreements should be recorded in writing, and that sentencing discounts for pleading guilty should be nonnegotiable. In the United Kingdom, for instance, sentence reductions in exchange for a guilty plea follow strict schedules based on when the plea is entered.

There is no obvious recipe for fomenting this kind of reform. The drivers vary “greatly from one jurisdiction to the next,” Turner said. But she did concede one common thread that unites jurisdictions invested in changing the plea process: They must be motivated by some overarching values besides efficiency, “like seeking justice,” she said, “however that’s defined.”

The alternative to improved pleas is more trials. A half-step in this direction has long been practiced in Philadelphia, where bench trials—before a judge but no jury—are common. By avoiding the jury-selection process, known as voir dire, bench trials dramatically shorten the length of the proceedings while a defendant’s guilt must still be proven beyond a reasonable doubt. In 2015, excluding cases that were dismissed, only 72 percent of criminal defendants in Philadelphia pled guilty, as opposed to 97 percent federally; 15 percent pursued a bench trial.

“The solution in Philadelphia is a very good one given the alternatives,” said Keir Bradford-Grey, the chief public defender for the city. “We firmly believe in putting evidence to the test and litigating cases. This program allows for far more trials than we see in other jurisdictions.”

John Rappaport, a law professor at the University of Chicago, proposes a more radical idea: If pretrial bargaining with the prosecutor is going to take place, it should embrace more than the basic exchange of guilt for leniency. Defendants should be able to bargain across the trial process itself, offering simplicity in exchange for a lesser charge. What if a defendant agreed to a trial before six or three jurors, instead of 12? Or what if the standards of evidence were downgraded, from beyond a reasonable doubt to a preponderance of the evidence?

“It’s all fairly straightforward, and wouldn’t require any real administrative framework, but it’s foreign,” Rappaport said. “If a defense lawyer approached a prosecutor and said, ‘Hey, let’s do away with voir dire and take the first 12 jurors who walk in the room,’ the prosecutor would be taken aback.”

He suggests that reforming the plea system to incorporate more trials would expose other problem areas. “Trials are an important window into how the system is functioning—they’re a form of audit,” Rappaport said. “They shine light on investigatory and prosecutorial behavior and air them publicly.” If the police behave badly, this remains buried when defendants take a plea. In this regard, even a heavily pruned trial is favorable to no trial at all. And such a bargaining process would not exist without limits. “The outcome of the trial still has to stem from the application of general legal principles to facts of individual cases,” he said. A defendant could not agree to a coin flip, for example, as the determinant of guilt.

Though plea-bargaining started in shadow—fixers, cops twisting inmates’ arms—it has since risen to become judicial custom. It is the daily bread of every criminal court in every jurisdiction in the country, and virtually all in service to economics. “We put together the most cumbersome and expensive trial system that the world has ever seen, and then we decided we can’t do it for all but a tiny, tiny portion of people,” Alschuler said. He reached for a metaphor that he first used almost a quarter-century ago, in an article that sought alternatives to plea deals. His frustration seemed undiminished with time: “It’s like trying to solve the transportation problem by giving Cadillacs to 2 percent of the population and making everybody else walk.”