On January 13, 1982, when I was nine years old, my parents turned on the television to watch a horror. Air Florida Flight 90 had crashed into the 14th Street Bridge near Washington, D.C., and plunged into the ice-choked Potomac River below. We were watching the frantic rescue of a handful of survivors in appalling conditions: below-freezing temperatures, a driving snowstorm, and a frozen river. I remember watching Larry Skutnik, a Congressional Budget Office employee, dive into the river to help save Priscilla Tirado, who was too weak to hold onto a life preserver dropped from a helicopter. The 737 crash killed 78 people, including four motorists on the bridge. Five people survived. What had been a routine flight between Washington National Airport and Ft. Lauderdale instead became the horror we watched at home.
There are a few watershed moments in the history of commercial aviation, accidents that lead to a fundamental change or rethinking of flight safety, training, maintenance, operations, and procedures. Air Florida joined the Tenerife disaster, the Sept. 11, 2001 hijackings, the de Havilland Comet stress fatigue case, and the Tissandier high-altitude balloon flight of 1875 as object lessons for science and the industry. In the case of Air Florida, improvements in flight training, cold-weather procedures and operations, and crew resource management spread across the injury and, happily, made commercial flight safer and more reliable.
Aircraft accidents are investigated by the National Transportation Safety Board. The NTSB is not a regulatory body. It is not a criminal investigative agency. It issues no legally enforceable orders. It prosecutes no one. And yet despite that the NTSB is the leading driver of continuous improvements in aircraft safety. This is because the NTSB is a legal safe space: by statute, nothing published by the board can be used in any criminal or civil legal proceeding. This practice rewards openness and transparency in the investigation for root causes – the more “evidence” provided, the less likely something will remain for prosecution or class action – and defuses the adversarial trial process where there is just one winner and the loser goes to jail.
Flight 90 was, despite its legacy, in almost every way a very typical case. The vast majority of industrial accidents and airplane crashes are the result of a cascade of failures (sometimes called “normal accidents”), usually combining equipment or sensor failures with lack of training, inexperience, or misjudgements by the crew. These are collections of smaller failures that individually would not cause or even contribute to the accident. But a sequence of these small failures almost certainly will cause an accident. In other words, these individual failures are individual risk factors that if checked one by one can dramatically reduce the probability of an accident.
The NTSB listed three “probable causes” in the Air Florida crash: crew failure to de-ice engines, crew failure to de-ice wing leading edges, and the captain’s failure to heed the first officer’s warnings during take-off. The board found several other contributing causes, including crew inexperience in freezing conditions, the control tower’s clearance for take-off despite known problems with 737s in freezing temperatures, and prolonged time between de-icing in a freezing snowstorm.
Underneath this “probable cause” hid a cascade of other risk factors: nonstandard de-icing procedures by ground crew, improvisational de-icing by lining up behind a forward aircraft’s jet exhaust (which actually made the icing worse), a hierarchical cockpit culture where the captain is always right, and instrument failure that gave incorrect readings of engine power during take-off.
NTSB made 10 recommendations to improve safety operations, most of which related adverse weather conditions. But the most important outcome of the Air Florida disaster wasn’t mentioned in the report at all. Crew Resource Management dramatically reformed cockpit culture, forcing air crews to work as a team rather than as captain and subordinate. Two people working on the same problem have better chances than one ignoring the problem. Accident investigation now rate crew resource management practices in the cockpit as important contributors to flight safety.
NTSB’s mandate is an important distinction because if every plane crash resulted in a criminal investigation the result wouldn’t be safer flying but prosecution of pilots and ground crew for human error, driving the best qualified people out of the industry, without improving the state of the art. Treating an accident as a series of risk factors is a much more productive process than treating the accident as a crime for which someone must be punished or blamed.
This punishment approach is still the case with the vast majority of police use of force incidents resulting in fatalities. These are most frequently investigated either internally by the police force itself or by prosecutors. In most cases (such as the one I am about to use as an example), the use of force is determined to be “justified” under the circumstances. The victim’s family is left with little recourse than to pursue a liability case against the department or city government. The officers and department are spared criminal liability.
But nothing changes because there was no “problem” found in the first place. And the cycle repeats itself far too often.
If we treat a police encounter like an aircraft – a dynamic combination of experience, technology, sensors, and judgment – that under normal conditions should not result in fatalities, then we can crack open up cases and expose them to much more scrutiny as we look for risk factors that, individually, may not cause a fatal encounter, but strung together or uninterrupted appear to lead inevitably to use of force resulting in death.
The case of Rayshard Brooks, who was shot twice by Atlanta police officers following a traffic stop in 2020, is a high-profile incident coming as it did two weeks after George Floyd’s fatal encounter with Minneapolis police. A lot of information about Brooks’ case is in the public domain, including dashcam video, bodycam footage, the special prosecutor’s statement exonerating the officers, and the medical examiner’s report (to my knowledge this is the first public release of the complete report). The police, public, and Brooks’ family all focus almost entirely on the 45 seconds between the arrest attempt and Brooks’ death. But even a cursory examination of the series of events over a 45-minute encounter demonstrate several clear risk factors contributing to the fatal incident. Let’s revisit the horror.
The medical examiner’s report describes Brooks’ fatal wound and apparent attempts at surgical intervention to save his life. The bullet cut through Brooks’ torso, hitting several organs. The medical examiner determined the cause of death to be this shot to the organs. However, it also notes that the bullet pierced the aorta and the inferior vena cava. These are, respectively, the main artery and main vein in the human body. This detail helps make sense of the examiner’s finding that during the examination – after police intervention, ambulance transport, and emergency surgery – Brooks still had more than half a liter of blood pooling inside his body.
While such an injury is almost certainly and immediately fatal, it still begs the question of why the police administered CPR to a gunshot victim. Chest compressions are at best useless in the case of severe haemorrhage. There are no public records about Brooks’ treatment by EMTs or in the hospital emergency room that I can find. (The medical examiner indicates aggressive surgical attempts to stem the bleeding, including sutures in the damaged aorta. These measures were futile.) While the focus on the shooting itself is clearly important, there has been no evaluation of decisions made along Brooks’ chain of care from the street to the hospital that might have resulted in his survival.
Let’s return to the beginning of the incident. Bodycam video of the first officer on the scene immediately introduces two interpersonal factors that will pervade the entire encounter. The first is Brooks’ noncompliance. He does not comply with the officer’s first request to move his vehicle from the drive-through to a parking space and has to be asked several times. Additionally, the officer’s apparent attempt to be polite means he asks, rather than orders, Brooks to park his car and remain in the vehicle. This faux camaraderie masks the officer’s suspicions but may have hampered clear communication.
Even more telling is the officer’s remark to himself whether he wants to run this stop to ground. He clearly considered the possibility of leaving Brooks in the parking lot, a decision that would have, of course, led to a completely different outcome.
Another factor that may be missed because it is obvious even on video: Brooks is clearly intoxicated by alcohol and this frames the rest of the police encounter. However, the medical examiner also found cocaine and a type of amphetamine in his body. While it is impossible to say whether the officers could determine Brooks had taken stimulants, the amphetamines could help explain how Brooks was not subdued by the officers’ first TASER shot and also managed to fight off two trained adult men trying to subdue him when he attempted to bolt from the scene. If the officers knew Brooks had taken amphetamines they might have treated the encounter very differently.
Several news reports remark on the entre-acte between the officers and Brooks, in the moments between the arrival of the DUI-certified officer and Brooks’ struggle to avoid arrest, as cordial, respectful, and calm. But there is a subtext here, part visible, part invisible, neither articulated in the open. First, the DUI-certified officer is clearly skeptical of Brooks’ evasive and rambling answers. The officer knows before administering the breath test for intoxication that Brooks is intoxicated. As Brooks continues to dodge and weave, the officer’s clear suspicion colors the entire interaction.
The second aspect of this encounter is completely unspoken but, if mutually understood, might have led to a completely different outcome. This is the fact, known to both of the officers and to Brooks, that he is on probation and if found guilty of a DUI would almost certainly go to prison. This knowledge could explain in part his evasiveness and misdirection. The officers know this, too, which makes them increasingly suspicious while raising tensions in the encounter. (Examining the sequence of events also illuminates something that is often an aggravating factor in police encounters: Brooks carried no weapon. He denied having a firearm to the first officer and was searched by the second. The officers therefore had no reasonable fear of Brooks using a firearm.) If the officers had informed Brooks of this knowledge, specifically articulated his arrest, and prepared physically to take Brooks into custody, it is possible the final, fatal encounter would never have begun.
This short, non-professional inquiry models the kind of faultless, independent investigation inquest in the mode of the NTSB or the Chemical Safety and Hazard Investigation Board. Logging every contributing factor to an accident, issuing a finding, and making recommendations will publicize risk factors in common with other cases, determine a probable cause without assigning criminal liability, and issues recommendations (for both police and the public) in policy, practice, and training.
While a federal investigative board comes to mind, policing is done primarily at the local level so a state or even a large city independent investigative board could be convened. This would shield police from liability but it also will give a clear picture of incidents while recommending improvements to police practice that, aside from damages and a prison sentence, is the best possible outcome for the families of the deceased. This is the case as much with aviation disasters in the Potomac as it is with fatal police encounters in Atlanta. There are no winners. There is only horror and loss.