When a federal judge in Georgia announced the fate of death row inmate Troy Davis on August 23, the long-awaited decision was not what Davis or his supporters had prayed for. He’d become a cause célèbre for organizations such as Amnesty International and the NAACP, which decried his conviction as baseless and racist and had deployed the usual campaign of online petitions, protests, T-shirts, and pins. Former president Jimmy Carter, Archbishop Desmond Tutu, and Pope Benedict XVI had lent their support to the cause.

But in the end, Davis lost. Sometime in the coming months, he will be executed by lethal injection, though he still claims to be innocent of the charge that he killed a police officer two decades ago. How many chances should we give to someone to prove his innocence? Just one? Five? An infinite number? This bedeviling issue—“actual innocence,” in legal parlance—remains one of the giant open questions of modern constitutional law. Davis’s fate may no longer be in the balance, but sooner or later the courts will be confronted with a person who is scheduled for execution and yet can prove his innocence.

Davis’s saga began in the early hours of August 19, 1989, when a group of African-American men, including Davis, were seen attacking a homeless man near a parking lot in Savannah, Georgia. Off-duty police officer Mark MacPhail responded to the altercation and was shot in the chest and head. He died before help arrived. One of the attackers named Davis as the killer, and other witnesses confirmed that story at trial. In 1991, a Georgia jury convicted Davis of the murder, and he was sentenced to die. Since then, he has tried every avenue legally available to him, never wavering from the claim that he is innocent.

Davis was convicted on the basis of the testimony of nine witnesses. No physical evidence conclusively linked him to the crime, and no murder weapon was ever found. Later, Davis claimed that seven of the nine witnesses had recanted or contradicted their prior testimony. One, Darrell Collins, who was 16 at the time of the crime, said that he had been threatened with being charged as an accessory to murder if he did not name Davis. Another, Kevin McQueen, had originally claimed that Davis confessed to him while the two were doing time together in prison. Later, McQueen admitted that he had been motivated to say this by a prison yard argument with Davis. (He had received a reduced sentence for his testimony against Davis.) The federal judge decided that several of the recantations Davis presented were not credible, and the remainder did not fundamentally undermine the evidence against him.

At the core of Davis’s case is the question of what should happen when a fair, lawful trial is still alleged to have led to the wrong result. Under the Constitution, can we legally execute an innocent person?

The Supreme Court declined to answer that question when it ordered a new hearing in Davis’s case last summer, but some of the justices wrote separately to address it. Justice John Paul Stevens argued that a person “who possesses new evidence conclusively and definitively proving, beyond any scintilla of doubt, that he is an innocent man” surely could not “be put to death nonetheless.”

What if someone goes through every possible legal procedure, and after all is said and done, still claims to be innocent?

But Justice Antonin Scalia, in a dissent joined by Justice Clarence Thomas, argued that a new hearing for Davis was pointless because it no longer mattered whether he had new evidence of his innocence. Even assuming that Davis could prove he was innocent, Scalia wrote, “this Court has never held that the Constitution forbids the execution of a convicted defendant who has had a full and fair trial but is later able to convince a habeas court that he is ‘actually’ innocent.” Indeed, he wrote, the Court’s prior decisions had “expressed considerable doubt that any claim based on alleged ‘actual innocence’ is constitutionally cognizable.”

At this point, anyone whose common sense has not been deadened by three years of law school might scream: How can it be an open question whether it is constitutional to execute the innocent? But the issue of “actual innocence” is more complex than our intuition suggests.

At a trial, the government is required to prove beyond a reasonable doubt that the defendant is guilty. If he is acquitted, that is the end of the matter. If not, he can appeal to higher courts, and ultimately ask the U.S. Supreme Court to review his case. If those appeals fail, he can challenge his conviction again by seeking a writ of habeas corpus (a form of court-ordered release) in both state and federal courts. The defendant can argue that the evidence presented at trial was insufficient to prove guilt, and in some limited circumstances (which vary from state to state and case to case) he can also present new evidence. If his case is rejected, he can appeal yet again: In the state court systems, he can generally appeal to one or more higher state courts, then seek review again from the U.S. Supreme Court. In the federal courts, he can request an appeal to a federal appeals court, then seek Supreme Court review yet again.

These challenges can drag on, but eventually they come to an end.

Yet what if someone goes through every possible procedure and after all is said and done still claims to be innocent? What if another court were to actually find him innocent? No belated claim of innocence has yet been found so compelling as to force the issue. In two previous death-penalty cases (in 1993 and 2006), the Supreme Court heard arguments from prisoners who had exhausted their appeals, yet claimed to be innocent and asked the Court to stop their executions. In both cases, the Court concluded that there was not enough evidence that the prisoners were innocent. (One of those prisoners, Leonel Herrera, was executed; the other, Paul House, was later freed after the Court remanded his case to a lower court on other grounds, and the prosecutor eventually dropped the charges.) The Court also touched on the question of actual innocence in a 2009 case in which it decided that an Alaska prisoner did not have the right to circumvent state law that might bar him from testing old evidence for DNA. In that case, the Court assumed that an actual innocence right existed for the sake of argument, but said the question wasn’t relevant to his situation. (DNA evidence has exonerated scores of people in recent years, but these cases did not involve actual innocence proceedings because governors or prosecutors voluntarily agreed to release prisoners or because there was a statute allowing them to be freed.)

Congress, for its part, has said that a convict has only a limited number of appeals and opportunities to attack his conviction in federal court, even if he has new evidence. (While the rules differ from state to state, many also impose such limits.)

The question is whether Congress’s prescription is constitutionally permissible. Why shouldn’t we try as hard as we can to make sure we get it right? Yet perfect accuracy is not the goal of the criminal justice system. For one thing, there are practical concerns with never-ending review. Jury trials followed by some form of judicial review have long been our traditional method of determining guilt or innocence. So what procedures would we use to retry the trial, and who would decide whether those new procedures were accurate? And once a judge was convinced that a convicted prisoner was actually innocent, could that determination be reviewed again by the prosecution?

Normally these questions are answered by the legislature that creates the appeal or habeas procedure. But because actual innocence claims are pursued outside established procedures, there are no ready answers to these questions. And judges cannot simply answer them by saying that there is a duty to get it right, regardless of how many proceedings and how much time it takes, because the judicial system’s resources are finite. Indeed, some advocates of an actual innocence right would not limit it to death penalty cases. If such a right meant that courts must allow every prisoner to perpetually pursue claims of innocence, it might push an already overburdened judicial system to the brink.

But these practical problems do not really go to the heart of the matter. One could imagine a court inventing a rough solution to some of these problems, as happened in Davis’s case. There is a deeper, more theoretical problem with recognizing an “actual innocence” right.

The principle that courts should seek justice sits alongside a principle of judicial finality—at some point, legal disputes must be settled. In nearly every case, whether civil or criminal, the losing side must eventually accept the authority of the court. In criminal cases, there is a safeguard: the executive’s power to pardon, one last chance for a case that has slipped through the cracks. An unending right to keep challenging that decision would make the legal system pointless.

Moreover, judges cannot decide the limits of their own power. They hear cases that the legislature has decided are within their purview. This legislative role is part of the balance of powers: Judges exercise great authority within their jurisdiction. Their rulings can bind very important people who disagree with them, including the president. Because judicial power is so great, it must also be circumscribed. By expanding their role in “actual innocence” cases beyond what the legislature had given them, judges would be straining against judicial finality and against the principle that courts must not define the scope of their own power. It is intolerably dangerous to give judges the unreviewable power to decide how powerful they are.

Such an assertion of authority would be costly in other ways. Indeed, the tradition of judicial finality is one of the chief justifications for the courts’ ability to invalidate unconstitutional laws through judicial review. That finality is what forces other branches to obey the courts’ judgments, right or wrong.

This concern with concepts such as finality, jurisdiction, and the balance of powers may sound technical, lawyerly, and highly abstract. But so is the criminal justice system. Crimes are messy and the facts are often disputed, but the law must provide simple answers: innocence or guilt, freedom or imprisonment, life or death. It does that through a system of rules animated by abstract principles. Indeed, the reason so much power is given to judges is because they are presumed to be expert at technical, lawyerly questions.

This is not to deny the potential for injustice. But we should not look to the courts for a solution. Legislatures create the procedures used to challenge criminal convictions. If our current ones are inadequate, lawmakers can create more generous rules for presenting new evidence of innocence. Indeed, in many states they have done exactly that in creating new procedures to accommodate DNA testing. Similar procedures could be created for other forms of new evidence.

The mistake is in thinking that judges are the only ones who can or should fix this injustice. If we care so much that actual innocence claims get into court, we should be lobbying the democratically elected branches, which have the power to create new procedures. If we are unwilling to demand better systems for assessing innocence from them, we should not be surprised that the courts are reluctant to invent one.

6769153235_c770fdb730_b.jpg

A little-known legal doctrine confounds the most basic understanding of justice—whether it matters if a convicted person is actually innocent.

William Baude