A SPOUSE,  A SISTER, or a brother suffers an acute mental health crisis. Family members call 911 as a last resort. A Boston Globe analysis reports that these calls—there are a lot of them—initiate chains of events that can (and, on an average of five times each year in Massachusetts, do) culminate in a fatal police shooting.

Should we send the cops? Or send the social workers?  Or both?  Current public debates leave the impression that the answers to those questions are the only answers we’ll need.

But if these encounters were that simple no cop or social worker would have been summoned in the first place. Whoever we call will confront immediate, complex challenges.

These are emergencies.  The scenes are dynamic.  The actors and their circumstances are too various to be cataloged in advance and reduced to checklists.

No first responder will be in the position of an orchestral musician, able to follow a score. First responders are jazz players, forced to improvise and coordinate their moves on the basis of a shared foundation of implicit knowledge and experience. They can’t “de-escalate” by simply throwing a switch.  They will have to forecast what will happen next from the little they know now, and with lives at stake.

In effect, they are forced to do the predictive calculations that Diane Vaughan invokes in the title of her recent book:  Dead Reckoning: Air Traffic Control, System Effects, and Risk.

Vaughan’s painstaking 1997 anatomy of the space shuttle Challenger launch decision revolutionized our understanding of disasters.  She showed us that the shuttle tragedy was not the work of reckless, self-aggrandizing NASA calculators breaking safety rules, but, rather, the product of a “mistake embedded in the banality of organizational life.” In Dead Reckoning Vaughan turns her attention to a field of high-risk social-technical interactions that functions safely under enormous pressures. (In May 2023, for example, Logan Airport handled over 69,000 departures and arrivals.)

In her book, Vaughan asks: “How do all of these things go right”?

As different as the two contexts seem on the surface, the lessons Vaughan derives from air traffic control’s success could help us build the capability to “land” every patient safely too. That should be our goal. We should react to the possibility of a mentally ill person in crisis being killed by the police or killing a bystander as we would to two planes colliding on a runway—zero tolerance. We should announce that goal, and we should get to work on it.

To begin with, thinking about these street and in-home encounters with air traffic control in mind forces us to accept that we are dealing with more than a simple confrontation between an individual patient and a particular cop or social worker.  What we see here, as in landing a plane, are effects emerging from complex interactive systems.

After a tragedy our reflex (and, certainly, the media’s) is to conduct a performance review—to evaluate the last cop or social worker in line.  But just as the processes of blaming a pilot don’t capture the breakdowns in communications, equipment, and training that contribute to an air crash, tightly focused performance reviews after a police shooting—directed at who gets disciplined, or prosecuted, or sued—miss chronic weaknesses in hiring, training, dispatch and equipment. Horrific outcomes are more likely to emerge from the interactions of components than from within one component, human or structural.

Besides, the fatal mental health outcomes we are trying to forestall arise in an environment of systems nested within systems, each sub-system with its own history and culture. The medical system’s diagnosis and treatment capacities are implicated. Questions of continuity of care and interface with the family members and neighbors who provide a default system of care-giving (or neglect) are influential. Policing, courts, and social work are organized in their own silos, too.

Like the air traffic controllers moving planes from one altitude or one vector to another, everyone implicated in these events is engaged in “boundary work.”  Virtually nothing happens within one silo independently; all decisions influence (and are influenced by) the performances and capabilities of other components. Safety requires collaboration; and danger emerges from failures in collaboration.

When and how to pass a situation along to another actor or mode of response become life or death questions.

In air traffic control, the challenges—sudden shifts in weather, incompetent pilots’ choices, equipment failures, even terrorist attacks—are answered by mobilizing a deeply engrained cultural commitment to continuously learning from events to foster resilience. All the operators, in all the roles, know their personal jobs, and they know how they interact with others and the system at large.

But aviation’s components have the advantage of functioning under the integrated federal umbrella managed by the Federal Aviation Administration.  Air traffic controllers work in scattered, individualized centers, towers, and other facilities, each with its own “personality,” but all received the same training and answer to the same agency.   The pilots, airlines, and manufacturers that the controllers are working to position are all included in careful National Transportation Safety Board reviews of serious accidents and in a confidential, voluntary, non-punitive Aviation Safety Reporting System that analyzes “near miss” and “good catch” incidents. Lessons are learned, and the lessons are disseminated.  Bulletins are issued; training is modified.

Nothing like this capacity is available to the cops, EMTs, social workers, emergency department staffs, court personnel, and family members who must do the “boundary work” required to keep the mentally ill and their communities safe in times of acute crises.  There are hundreds of police departments in Massachusetts; hundreds of medical facilities; hundreds of emergency departments and 911 dispatch call centers.  There are thousands of family members and neighbors who have been involuntarily assigned responsibilities.  The  “When should I call?”  “Whom should I call?” are always lurking in the minds of the family members of the seriously mentally ill. Answering those questions is “boundary work” too.

We can’t live without disciplining dangerous conduct by individual system actors, but we can’t allow a laser focus on punishment to blind us to the need to develop a foundation for continual work on safety.  There are fledgling efforts that we could adapt and incorporate into a safety center.  Massachusetts’ Betsy Lehman Patient Safety Center provides a “safe harbor” for examining weakness in the medical silo.  Jurisdictions such as Tucson have conducted all-stakeholders  “sentinel event reviews” of in-custody deaths of people in psychiatric crisis that have generated dozens of recommendations.

What if we combined the two to examine fatal shootings—and successful de-escalations and diversions—and learn their lessons? What if we developed a platform for sharing those lessons? A Safety Center that could address these and other recurrent problems such as wrongful convictions?

Could we find a way to show that we think of the mentally ill not as dangers to be controlled or confined, but as friends and neighbors to be cherished?  Could we communicate to the mentally ill and their families—and to each other—that “landing” all of the patients is something we desperately want to do?

James Doyle is a Boston defense lawyer and author and formerly the 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.