Prosecutorial Discretion: What Each Position Is Protecting

A woman in Philadelphia watches her neighbor's seventeen-year-old son arrested for marijuana possession. It is 2018. The new district attorney, Larry Krasner, has issued a declination policy: his office will not prosecute marijuana possession cases. The son is released. He finishes school. She feels something she had not expected to feel about a DA's office — relief, and then a tentative hope that this particular machinery might not be as fixed as she had assumed.

A man in Chicago signs a plea agreement admitting to a theft he did not commit. His public defender has explained the calculation: trial would take months, and if convicted he faces two years. The plea is probation. He has a job; he cannot afford two years. The prosecutor offering the deal has never spoken with him. The charging document was reviewed for minutes. His entire fate — a permanent criminal record, the jobs he will not be hired for, the apartments he will be denied — flows from a decision made in a corridor before a hearing he was not present for.

A law professor in Washington named Angela J. Davis is reviewing charging data across jurisdictions and noticing that identical conduct produces wildly different charges depending on who is arrested and which office handles the case. Two people commit the same armed robbery. One is charged with robbery; one is charged with robbery, use of a weapon, assault, and three further counts. One receives a plea offer of two years; one receives an offer of eight. Both plead guilty. One goes home in two years; one does not come home for a decade. No court reviews this difference. No law requires it to be justified. It is discretion — the prosecutor's — and it is essentially unreviewable.

These three people are inside the same debate about the office that John Pfaff has called the most important and least understood in American criminal justice. Between 1994 and 2008, as crime fell and the national conversation focused on sentencing laws and mandatory minimums, prosecutors quietly doubled their rate of filing felony charges per arrest. They did not need legislation. They did not hold hearings. The prison population grew because prosecutors chose to charge more people more seriously — a decision made in thousands of individual offices with almost no public scrutiny. This map is about what each position in the resulting debate is actually protecting, and why the debate is harder to resolve than the positions usually acknowledge.

This map is distinct from the criminal justice map, which covers the broader question of incarceration — what prisons are for, who is in them, and what the abolitionist and reformist traditions are each protecting. It is distinct from the criminal sentencing map, which focuses on the moment a judge imposes a sentence and what purpose that sentence should serve. This map focuses on the office that makes most of those outcomes inevitable before a judge ever enters the room: the prosecutor's charging, plea, and declination decisions, and the debate about what that power is for, who it answers to, and whether it can be reformed.

What traditional prosecution is protecting

The traditional prosecution position holds that the prosecutor's job is to enforce the law as democratically enacted, to represent the state in the adversarial system, and to pursue convictions in cases where the evidence supports them. Discretion exists to serve justice in individual cases — to account for the specific facts, the specific defendant, and the specific circumstances that no legislature can anticipate. It does not exist as a tool for implementing social policy from the DA's office. When a legislature defines an offense and sets a sentencing range, a prosecutor who systematically declines to charge that offense is, in effect, nullifying law that was passed by elected representatives and signed by elected executives. The voters who enacted that law through their representatives did not ask for a prosecutor who would substitute their own policy judgment for the legislature's.

Traditional prosecutors are protecting the rule of law as a constraint on unilateral power. The same argument that progressives make about discriminatory police enforcement — that selective enforcement of the law is a form of abuse — applies to selective declination. A prosecutor who announces that they will not charge certain categories of offense is wielding veto power over democratically enacted statutes. They may believe their judgment is better than the legislature's; they may be right. But the structural problem remains: if prosecutors can nullify statutes they disagree with, the check that limits prosecutorial overreach — that the conduct must have been criminalized by a representative process — no longer functions. The power flows in both directions. A prosecutor who can decline to charge anything can also charge anything. The discretion that makes progressive prosecution possible is the same discretion that makes selective prosecution of disfavored groups possible. Constraining that power, rather than celebrating it when it reaches outcomes one favors, is the rule-of-law position.

They are protecting victims of crimes that progressive declination policies deprioritize. The communities most affected by retail theft, drug-related violence, and gun crime are often the same low-income, predominantly minority communities that progressive prosecution frames as its intended beneficiaries. When a DA's office announces that it will not prosecute retail theft below a certain threshold, or will not seek pretrial detention for nonviolent defendants, the people most harmed by the resulting under-enforcement are typically not wealthy. They are the small business owners whose margins cannot absorb repeated theft, the public housing residents who live near open-air drug markets, the families of homicide victims in neighborhoods that have seen murders go unsolved at high rates for decades. The argument that the prosecutor's office systematically over-polices minority communities is real; the argument that it under-protects those same communities is equally real. Serious traditional prosecutors hold both.

They are protecting the deterrence function that depends on credible enforcement. The case for criminal law rests on the assumption that people who commit crimes face a realistic possibility of consequence. When that assumption breaks down — when certain offenses are known to be uncharged, certain pleas known to be available regardless of the facts — the deterrence calculation changes. The concern is not theoretical. Some progressive declination policies have been followed by increases in the offenses they declined to prosecute, and the political backlash — Chesa Boudin's recall in San Francisco, Kim Foxx's decision not to seek reelection in Cook County — reflects a voter judgment that something was lost when the credible threat of consequence was withdrawn. Traditional prosecutors are protecting the function of law as a constraint on conduct, which requires that the constraint be real.

What progressive prosecution is protecting

The progressive prosecution position holds that the DA's office has always wielded enormous discretion, and the question is not whether to use it but how. Every charging decision is a discretionary act. Every plea offer is a policy choice. Every death penalty case filed or declined reflects a judgment about how the law should operate in practice. The reform is not to introduce discretion into a system that didn't have it; the reform is to use existing discretion differently — to reduce incarceration for low-level offenses, to end cash bail that keeps poor defendants in jail before trial, to build diversion programs that treat addiction and mental illness as health problems rather than criminal ones, and to establish conviction integrity units that review past wrongful convictions.

Progressive prosecutors are protecting the people the prior discretion was destroying. A teenager arrested for marijuana possession and charged as a felon faces consequences — the felony record, the ineligibility for federal student aid, the housing and employment barriers — that bear no relationship to the public safety threat posed by marijuana possession. The mandatory minimum sentencing era was not an era of prosecutorial restraint: it was an era in which prosecutors used mandatory minimums as leverage in plea negotiations, threatening defendants with decades of incarceration to secure guilty pleas in cases where the facts were ambiguous. The people who agreed to those pleas — many of whom were innocent or whose culpability was far more limited than the charges implied — made rational calculations given the threat they faced. Progressive prosecutors are protecting them: the people for whom the gap between what the law authorized and what justice required was the largest.

They are protecting the possibility that the office can be a tool for repair rather than only a tool for punishment. Prosecutor-led diversion programs — for drug offenses, mental health crises, domestic violence, youth cases — have produced better outcomes on recidivism than prosecution and incarceration in many cases. Larry Krasner's Philadelphia office published its own data: reduced incarceration years, reduced supervision years, reduced use of cash bail, and documented reductions in certain racial disparities, without the crime spike his critics predicted. Rachael Rollins in Boston, Alvin Bragg in Manhattan, and Kim Foxx in Cook County each built similar records — incomplete, contested, but present. The argument is not that progressive prosecution is perfect; it is that it is being evaluated by a standard that traditional prosecution has never been required to meet. Defenders of the old system do not typically present data showing that aggressive charging improved public safety. They assume it did. Progressive prosecutors are protecting the principle that the assumption should be tested, and that the evidence when tested is more favorable to a different approach than the critics acknowledge.

They are protecting the democratic mandate for local criminal justice. When Larry Krasner won the Philadelphia DA race in 2017, he won it running explicitly on a platform of ending mass incarceration and using prosecutorial power differently. He won. He was reelected. He won a third term. Whatever one thinks of his policies, the argument that his declination practices are anti-democratic requires explaining why a policy that a majority of voters repeatedly elected him to implement is less legitimate than the unilateral charging decisions of his predecessors, which were never subject to a public vote on anything like those terms. Progressive prosecution is, on its own account, a more transparent form of prosecutorial power — one that names its choices and submits them to voters — than the system it is trying to replace.

What accountability reformers are protecting

The accountability reform position, associated most closely with Rachel Barkow and John Pfaff, holds that the debate between traditional and progressive prosecution is the wrong debate. Both positions assume that the answer to prosecutorial power lies in who holds it: get the right person elected, and the power will be used well. What neither position grapples with is the structural reality that prosecutorial power in the United States is almost entirely unaccountable by design — to courts, to legislatures, to any independent review body, and even to voters, because voters have almost no information about how prosecutors are actually making decisions.

Accountability reformers are protecting the public's ability to know what prosecutors are actually doing. Most prosecutor's offices do not publish data on charging rates, plea offers, or outcomes by offense and defendant characteristics. They are not required to. The result, as Pfaff documents, is that prosecutors doubled their felony filing rate over two decades without anyone outside the system knowing it was happening. The reform case is not primarily about whether to charge more or less; it is about requiring offices to collect and publish the data that would allow voters, courts, and legislatures to evaluate what they are doing. Prosecutors are among the only government officials who exercise life-altering power with no systematic performance accountability. Barkow's institutional design argument — borrowed from administrative law — is that the solution is not to elect better prosecutors but to build oversight mechanisms that function regardless of who is elected.

They are protecting the victims of prosecutorial error and misconduct that the current system has no mechanism for catching. Prosecutors in the United States have absolute immunity for their actions in court. They have qualified immunity for their investigative actions. When a prosecutor withholds exculpatory evidence — a constitutional violation under Brady v. Maryland — the sanction, if any, is applied to the conviction, not to the prosecutor. Prosecutors are almost never personally disciplined for misconduct. Conviction integrity units — the in-house review bodies that Krasner and others have established to review past wrongful convictions — exist in fewer than a fifth of the largest offices, and their thoroughness varies entirely by the priorities of the incumbent DA. Building that function into the structure of the office, rather than making it dependent on the DA's enthusiasm, is the accountability reformers' argument.

They are protecting the separation of powers logic that the prosecutor's accumulation of discretionary authority has quietly undone. As mandatory minimums shifted sentencing authority away from judges, it shifted toward prosecutors: the charge determines the sentence, and the prosecutor controls the charge. Bibas documents how the entire adversarial system — jury trials, the right to confront witnesses, the requirement that the state prove guilt beyond a reasonable doubt — operates in a world where 95 to 97 percent of convictions come through guilty pleas that are negotiated before trial. In that system, the prosecutor is investigator, charging authority, plea negotiator, and effective sentencer. The constitutional safeguards built around the trial process apply to almost no one. What replaced them — plea bargaining — has no constitutional structure, no required standards, and no external oversight. Accountability reformers are protecting the principle that the power to deprive people of liberty should not be exercised without some institutional check, regardless of whether the prosecutor using it is traditional or progressive.

What structural critics are protecting

The structural critique holds that the problems with prosecutorial power cannot be fixed by electing different prosecutors or building better oversight mechanisms — because the problems are built into the structure of the office and the system it operates within. Angela J. Davis, whose career as a public defender at the D.C. Public Defender Service preceded her scholarly work, documents that the racial disparities in charging, plea offers, and sentencing recommendations do not require discriminatory intent to produce. They emerge from the ordinary exercise of discretion applied to facts that are themselves shaped by racially disparate policing, bail systems that disadvantage poor defendants, and a plea bargaining structure that coerces the innocent along with the guilty. A prosecutor with no conscious racial animus, operating by office norms, on cases that arrive pre-sorted by racially disparate enforcement, will produce racially disparate outcomes.

Structural critics are protecting the communities that have been processed rather than served by the criminal legal system for generations. James Forman Jr.'s meticulous history shows that the punitive turn was not imposed on Black communities against their will — it was, in significant part, embraced by the first generation of Black elected officials in American cities, who faced real crime crises and genuine constituent demands for safety. But embraced or not, the consequences — mass incarceration concentrated in specific neighborhoods, the permanent second-class citizenship of the felony record, the removal of fathers and sons and the disruption of community social structures — fell most heavily on the communities these officials were trying to protect. Structural critics are protecting the recognition that a system can cause harm through ordinary operation, that the harm is not accidental, and that the reforms being proposed — electing better prosecutors, adding data requirements — do not address the scale of that harm.

They are protecting the defendants who have no trial because trial itself has become a threat. Judge Jed Rakoff, writing from his position on the federal bench, describes the plea bargaining system as a constitutional evasion: the Sixth Amendment guarantees the right to trial; a system in which trial is punished with sentences three to five times higher than the plea offer does not abrogate that right formally, but it eliminates it practically. Innocent defendants plead guilty because the risk calculation is rational: even with a strong defense, the risk of conviction is real, and the difference between conviction and the plea offer is measured in years of life. The prosecutor who makes that offer is not committing misconduct; they are operating the system as designed. Structural critics are protecting the recognition that a system that processes millions of people without giving most of them anything resembling a contested hearing is not, regardless of who runs it, a system that the word "justice" readily describes.

They are protecting the possibility of alternatives that the current system's dominance forecloses. Community accountability programs, participatory defense organizations, restorative alternatives to prosecution — these exist, produce better outcomes in many of the cases they handle, and remain marginal because the default is prosecution, the default is plea, and the default is conviction. The resources go to the machinery of charging because the machinery of charging is what the system is built to do. A structural critique asks: if we began from a different baseline — from community-based accountability rather than state prosecution as the starting point — what would it look like, and what would it cost? The existing system has never been required to justify its resource allocation against the alternatives. It assumes its own necessity.

Where the real disagreement lives

All four positions are responding to real features of the current system. The fractures run deeper than the policy debate typically reaches.

Is prosecutorial discretion democratic or anti-democratic? Each position gives a different answer to the same institutional question. Traditional prosecutors say discretion is anti-democratic when used to nullify legislative choices — and that it should be disciplined by fealty to enacted law. Progressive prosecutors say discretion that produces mass incarceration without a vote is the anti-democratic choice — and that explicit policy commitments subjected to electoral scrutiny are more democratic, not less. Accountability reformers say both positions fail to address the fact that prosecutors are essentially unaccountable to any external check; democratic legitimacy requires transparency, data, and oversight, not just elections. Structural critics say that elections in which one candidate is a prosecutor and the other is a prosecutor provide limited democratic agency to the people most affected by prosecutorial power — who are also, in many jurisdictions, the people least likely to vote, least likely to be heard by campaigns, and most likely to bear the costs of either approach. The democratic question is not settled by pointing to elections.

Who are prosecutors accountable to — and who should they be? Prosecutors in the United States are elected locally, in most jurisdictions. They answer, in theory, to the voters of the county. In practice, they answer to the political coalition that elected them, the professional norms of the office, and the informal expectations of judges, police, and the bar. Victims have no formal role in charging decisions. Defendants have no right to have charges explained or justified. Courts review the sufficiency of evidence but not the wisdom of charging. Legislatures appropriate money to prosecutor's offices but rarely review how it is spent. The result, as Barkow argues, is that prosecutors are accountable in the formal sense — they face reelection — but not in the substantive sense: no one evaluates whether they are producing safety, proportionality, or justice by any measurable standard. The accountability reformers' argument is that the existing accountability structure is a form of non-accountability wearing democratic clothes.

Does racial disparity in prosecution reflect prosecutorial choice or the cases that arrive at the office? The structural critique holds that charging decisions translate upstream racial disparities — in policing, in bail, in defense quality — into outcomes, and that the translation is the prosecutor's responsibility to interrupt. Traditional prosecutors respond that the office prosecutes the cases that come in; if the cases arriving are racially skewed, the solution is police reform and bail reform, not prosecutorial declination of otherwise valid charges. Progressive prosecutors say that using declination and diversion to interrupt the pipeline is precisely what the office's discretion is for. Pfaff's data complicates all three positions: prosecutors chose to increase felony filing rates during a period of falling crime, not in response to case volume but as an active policy choice — which means case inflow does not fully explain outcome disparities, and the office's choices are more consequential than the "we just prosecute what comes in" narrative allows.

Is progressive prosecution transformation or substitution? The sharpest tension within the reformist space is whether electing progressive prosecutors changes the system or changes its occupants. Forman's history of Washington, D.C. is the cautionary account: a generation of Black elected officials, including Black prosecutors, implemented punitive policies while genuinely believing they were protecting their communities — and the communities bore the consequences. The structural critique asks whether Krasner's Philadelphia is different in kind or different only while Krasner holds office; whether the Boudin recall reveals that a progressive prosecutor who generates political opposition will be replaced by someone who reverses the policies; and whether the energy going into electing progressive prosecutors might produce more durable change if directed toward the institutional reforms — data requirements, independent oversight, sentencing transparency — that do not depend on a particular person holding power.

What sensemaking surfaces

The prosecutor's office is the least visible major institution in the criminal legal system, and the most powerful. It makes decisions that are unreviewable, unmeasured, and largely invisible to the public it is supposed to serve — and those decisions determine who goes to prison, for how long, in greater numbers than sentencing laws, mandatory minimums, or any of the levers that the policy debate typically focuses on. Understanding why is the beginning of understanding why reform is harder than it looks.

The traditional position's greatest contribution is its insistence on limits: the rule of law means something, democratic enactment means something, and a system where the prosecutor can nullify any statute they disagree with is one that has replaced one form of arbitrariness with another. The critique of progressive prosecution on these grounds is not simply political; it identifies a real structural problem that does not disappear because the progressive prosecutor happens to be right about marijuana. The discretion that makes good prosecution possible is the same discretion that makes bad prosecution possible, and celebrating the one without designing against the other is not a durable position.

The progressive position's greatest contribution is its empirical challenge to the default. The criminal legal system of the last fifty years was not a neutral baseline. It was a set of choices — to charge felonies at twice the prior rate, to use mandatory minimums as plea leverage, to incarcerate at levels that no other democracy has approached — that were not subjected to the same scrutiny that progressive declination policies now face. Requiring the reform to prove itself against a standard the status quo was never asked to meet is not intellectual rigor; it is a status quo bias. The progressive prosecutors have produced evidence. The question is whether it is being fairly evaluated.

The accountability reformers' greatest contribution is the institutional imagination to see that the debate between traditional and progressive prosecution is a debate within a structure that neither side questions. Both depend on elected prosecutors using discretion well. Neither builds anything that would continue to function when the prosecutor uses it badly. The administrative law tradition — agencies with data requirements, performance metrics, independent oversight, and review mechanisms — has produced more durable accountability in every other domain of powerful government action. There is no structural reason why prosecutorial power should be exempt. The fact that it has been is a design failure, not an inevitability.

The structural position's greatest contribution is the question it refuses to stop asking: who pays? The woman in Philadelphia was relieved that her neighbor's son was not prosecuted. The woman whose neighbor was robbed is waiting for someone to tell her why the case is not being prosecuted. The man who pleaded guilty to something he didn't do is trying to find a job with a felony record. These are not three separate problems requiring three separate reforms. They are the same system producing different outcomes for different people, and no account of prosecutorial reform is complete if it can speak clearly to only one of them. The office of the prosecutor sits at the center of that system — deciding who gets charged, who gets a deal, who is diverted, who goes home, who does not. What it is for, who it answers to, and whether it can be something other than what it has been are questions that the debate between traditional and progressive prosecution has barely begun to address.