Although 2015 will go down as the year when the United States began grappling with the problem of police violence, it ended with a trio of defeats for reformers.

First, a jury in Baltimore was unable to come to a verdict in the trial of Officer William Porter, one of several officers charged in the death of Freddie Gray. Several days later, a grand jury in Waller County, Texas, decided that there had been no crime committed in the death of Sandra Bland in a jail cell there. Finally, and most gallingly to many observers, on Monday a grand jury in Cuyahoga County decided not to indict two officers in the shooting death of 12-year-old Tamir Rice.

Taken together, these cases—and particularly the Baltimore and Cleveland cases—demonstrate yet again the difficulty involved in holding police accountable when civilians are killed. Even as there is greater awareness about the toll that police killings take, police are seldom prosecuted, and when they are, they are seldom convicted. That was the case before Michael Brown’s death in August 2014, and it remains true today. The reasons for that are various. Prosecutors are reluctant to bring charges against police, because they rely on officers to gather information and serve as witnesses in other cases. Juries tend to be deferential to officers.

There are also legal protections: In Graham v. Connor, the Supreme Court ruled that events “must be judged from the perspective of a reasonable officer, rather than with the 20/20 vision of hindsight.” Finally, even when the facts seem clear-cut, the law grants police wide latitude. Although many people who watched dash-cam footage of Bland’s arrest were horrified by Trooper Brian Encinia’s conduct, police experts who reviewed the footage, including some who criticized Encinia’s judgment in no uncertain terms, generally felt he had acted within his legal authority. Many departments employ “use-of-force matrices,” which detail what steps an officer may take during an incident, in some cases giving them the right to use more aggressive action than might be necessary or seem justified to an outside observer.

This was particularly apparent in the Rice case. The boy was shot by an Officer Timothy Loehmann just seconds after he arrived on the scene, sent by a dispatcher who told him there was a report of a man pulling out a gun and pointing it at people. Surveillance footage of the death galvanized and appalled the nation. The 12-year-old being gunned down by the officer so abruptly seemed to exemplify overuse of deadly force, while the ensuing events—Rice’s sister was prevented from reaching him, and officers did little to save his life—clinched the case as a signal injustice. As more information emerged about Loehmann, including his abbreviated, troubled career with another Ohio police department, there seemed to be widespread recognition that he shouldn’t have been wearing a badge and that he had acted inappropriately when he shot Rice.

The problem is that although Loehmann’s actions may have seemed obviously inappropriate to a layman, that doesn’t mean that they actually violated the law. Three independent reports, commissioned by Cuyahoga County Prosecutor Timothy McGinty all found that the Loehmann and his partner Frank Garmback had acted within proper protocols and rules for officers. (One report used particularly unfortunate and offensive language, likening Rice’s loss of life to the potential end of Loehmann’s career.) The grand jury’s decision not to indict is simply the latest evidence that no statutory crime may have been committed.

In announcing the grand jury’s decision, McGinty made that argument: What happened was terrible, but I can’t prove it was a crime. “The state must be able to show that the officers acted outside the constitutional boundaries set forth by the Supreme Court of these United States,” he said, and while Rice’s death was a “tragedy,” McGinty said, “it was not, by the law that binds us, a crime.”

That isn’t to say that McGinty couldn’t have procured an indictment—if not necessarily a conviction—if he’d taken a more aggressive strategy. The DA has come in for harsh criticism throughout the case. He took an extremely long time to bring the case before a grand jury—so long, in fact, that Rice’s family and activists dredged up an obscure Ohio law to get a municipal judge to issue at warrant for the officers, circumventing the McGinty’s process. (They received an unsatisfying split decision: A judge ruled that there was probable cause to arrest the officers, but that the law did not actually authorize him to issue warrants.)

Activists and other observers accused McGinty of issuing the three independent reports as a way to justify a future failure to indict—a suspicion that Monday’s announcement will only reinforce. McGinty also failed to convict Officer Michael Brelo in the 2012 deaths of Timothy Russell and Malissa Williams, two residents gunned down after a mistaken chase. In short, McGinty seems at best soft on police and at worst ineffective as a prosecutor. “It has been clear for months now that Cuyahoga County Prosecutor Timothy McGinty was abusing and manipulating the grand jury process to orchestrate a vote against indictment,” Rice’s family said in a statement.

But one tough lesson of the William Porter case is that however lethargic McGinty’s approach may have seemed, a more aggressive approach is no guarantee of different results. Whether prosecutors move glacially and timidly or quickly and boldly, it’s hard to hold police accountable because of the way the law is written and the system works. Baltimore City State’s Attorney Marilyn Mosby did practically everything differently: She moved with incredible speed to bring charges against the officers involved in the death of Freddie Gray, and she brought an aggressive slate of charges—including a depraved-heart murder charge against one officer. She quickly obtained indictments from a grand jury and prepared to take the cases to trial.

But once the first trial began, the difficulties facing prosecutors became clear. A central element of the state’s case was an accusation that Porter had failed to restrain Gray with a seatbelt. Yet Porter’s team mounted a convincing argument that although Porter may have violated written policies, what he did was in line with general practice for Baltimore cops. The law was murky enough that it was tough to obtain a conviction, and some analysts felt that prosecutors were lucky to get a hung jury rather than simply an acquittal.

Further evidence for the gap between legal standards for a crime and civilian judgment about proportionality comes from the fact that even though officers are seldom prosecuted for violence, many cities—including Baltimore and Cleveland—have made large payouts to victims and their families, either as a result of civil verdicts or as settlements to end or forestall civil suits. McGinty himself suggested that the civil system might provide some recourse for Rice’s family. “We, too, want justice for Tamir,” he said. The grand jury’s decision “doesn't mean the legal system is done,” he said, adding that civil courts might provide the accountability to the family “that they deserve.”

If the failure to indict in Cleveland is a defeat for police-reform activists and Black Lives Matter, it is also a defeat for some of the movement’s critics who have espoused “responsibility politics” for African Americans who criticize the justice system. In October, I noted Harvard Law Professor Randall Kennedy’s call for activists to set aside victims of police violence like Michael Brown, whose reputation could be more easily impugned because of his alleged involvement in a robbery before he was shot. Proponents of respectability politics “preferred to rally attention around Tamir Rice, the black twelve-year-old who was playing with a toy gun in a park when he was precipitously shot dead by a policeman in Cleveland, rather than a figure like Michael Brown,” Kennedy wrote earlier this year.

The fact that no one was indicted in Rice’s death shows the shortcomings of that approach. While there are many death at the hands of police, there are few perfect victims, victims whose deaths are so clear-cut, so appalling, so unjustified, that a conviction is a slam dunk. Many of the people who come into contact with the long arm of the law may have committed crimes; others may simply have acted unwisely, like Rice brandishing his toy gun with the orange tip removed. None of this justifies extrajudicial execution by the state, which is what Rice—along with Samuel Dubose and Walter Scott and many others—received.

Rice’s family and other activists are calling on the Department of Justice to step in and take up the case. But as the DOJ’s investigation into Michael Brown’s death showed—and as the department hinted Monday—federal civil-rights cases are extremely difficult, and tend to require an even higher standard of proof than a simple criminal prosecution, making it unlikely that Washington will deliver justice in this case.

Although the anger of police reformers and Black Lives Matters activists ran high in 2015, there was also, at times, a naïvely optimistic underpinning to their protests. They were built around the belief that if only there were reliable statistics on deaths in the hands of police, if only there was more awareness about deaths, if only more political pressure could be brought to bear on prosecutors, then the criminal-justice system could deal with police-inflicted deaths. That hope was encouraged by the Walter Scott case, in which Officer Michael Slager was fired and charged with murder almost immediately after a video of Scott’s shooting was made public.

The results this month in Baltimore and Cleveland give reformers a roadmap to follow in 2016. It’s not enough to just enforce the laws as they exist now—though that, too, is essential. Reducing police abuse will also require political efforts to change the laws and protocols that determine how police can act and what actions are justified. Until those changes are made, stomach-churning incidents like Rice’s death will continue to go without indictments or convictions.