HENRI CARTIER-BRESSON took the epigraph for his landmark collection of photographs from Cardinal de Retz: “There is nothing in this world that doesn’t have a decisive moment.” Cartier-Bresson’s American publisher settled on The Decisive Moment for the book’s title.

Whenever the media confront a horrific outcome, the search for the “decisive moment” dominates the coverage. Two very different recent tragedies—the fatal police shooting of a carjacking suspect in Boston’s Roxbury neighborhood and the deadly collision of an Air Canada jet and a firetruck on a LaGuardia Airport runway in New York City—support Cardinal de Retz.

Radio transmissions, body-camera footage, and surveillance videos captured “decisive moments” in these incidents. They were replayed, or cited in accounts of court proceedings and news conferences for days.

The striking contrast between the official responses in these two cases of decisive moments illuminates a weakness in our public safety architecture—a dangerous gap that we have a chance to fill.

In the Roxbury case the Suffolk district attorney charged Boston police officer Nicholas O’Malley with manslaughter in the death of Stephenson King, who was at the wheel of a stolen car. That prosecution may be fair enough. No decent justice system can survive without being willing to punish the misconduct of its own officials. After all, those officials are administering the state’s monopoly on violence. Whether O’Malley is guilty or not will now be for a jury to decide.

But the National Transportation Safety Board’s response to the LaGuardia crash shows us why, even if prosecuting O’Malley is the right move, it is a terrible place to stop.

The prosecutors charging O’Malley went “down and in.” They identified what they allege is a faulty human component failing at a decisive moment. The NTSB will go “down and in,” but it will also go “up and out” to assess any conditions that bent the probabilities toward the deadly outcome.

Although NTSB has clinching evidence that the LaGuardia air traffic controller errantly sent the firetruck across the runway, it will still ask whether the controller was set up to fail—evaluating the collision as a full system crash.

The controller made a horrific choice, but why? Was training a factor? The firefighters’ procedures? The lack of transponders on the firetrucks? Workloads? Communications? Distraction? The NTSB’s focus won’t be limited to blaming; the lens will be wider— learning and prevention will be prioritized.

The Suffolk district attorney initiated a performance review of the proximate cop. The NTSB will conduct a comprehensive event review of everything that took place at LaGuardia.

Criminal justice disasters, like air crashes and wrong-patient surgeries, are “organizational accidents.” Unnecessary police shootings and wrongful convictions can never be explained entirely by the independent acts of lone “bad apples.”

Yes, the cop fired recklessly, or the prosecutor hid the exculpatory evidence. But someone—many “someones”—hired them, trained them, acculturated them, supervised them, assigned them, incentivized them, and failed to intercept their errors. These remote actors all had their reasons. Why did they zig when they should have zagged? Will the same features remain in place to shape the decisions of the next cops or prosecutors who come along?

Deterrent punishment is only one means of prevention, and it is far from foolproof. A destructive byproduct of relying exclusively on punishment is that law enforcement and the public are arrayed in permanent conflict—one side shown seeking impunity, the other, vengeance. This arrangement perversely buries the fact that no one in either camp wants these events to happen. They would take steps to prevent them if they could see them coming.

We are beginning to see early adapters in the criminal system conducting “sentinel event reviews.” These all-stakeholder (and all ranks) reviews aim at learning, not blame. They complement the punitive accountability pursued by the Boston police prosecution with a “forward-looking accountability” that examines the influences that set the stage for disaster.

They give the public and the authorities something to do with each other rather than to each other. By now, a dozen of these reviews of diverse events—including deaths in custody, wrongful convictions, preventable domestic homicides, and civil disorders—have been successfully conducted, supported by funding from the National Institute of Justice. They have generated hundreds of consensus recommendations from the officials, specialists, and community members involved.

The deaths on the LaGuardia runway show that perfect safety has not been achieved in aviation. But the NTSB’s investigation of those deaths recognizes our reality: There are no permanent “fixes” possible; only dedicated learning from events and continuous improvement will bend the curve toward greater safety.

Massachusetts can recognize that this is also true in public safety. We can disentangle learning and punishment. (The Betsy Lehman Patient Safety Center, which offers statutory protection against use of its materials in litigation, provides one model vehicle for this approach.) If we learn to look, every “decisive moment” can teach us what came before it, what lay beneath it, and what we can do beyond it to prevent recurrence. And if we all learn to look together, we might generate some healing, too.

James Doyle is a Boston defense lawyer and author. He is former head of the public defender division of the Committee for Public Counsel Services and director of the Center for Modern Forensic Practice at the John Jay College of Criminal Justice.