How to Understand the Reagan Airport Crash

It took the social-media site X about an hour to reach consensus on the cause of the midair collision over the Potomac River on Wednesday evening: the helicopter was trying to take down the plane, to kill someone on board. There is lots of other speculation, including, famously, by President Trump, who blamed the Federal Aviation Administration for supposedly hiring incompetents, under a diversity-equity-and-inclusion policy. Trump is unique in many ways; he is the only U.S. President to have owned an airline, the Trump Shuttle, which ran from LaGuardia to Logan and Reagan, then known as National Airport, from 1989 to 1992. If he picked up much aviation knowledge from the experience, it hasn’t shown yet.

It is too soon to know all the important factors that brought a Canadair Regional Jet and an Army Black Hawk helicopter simultaneously to precisely the same spot, near the approach end of Runway 33 at DCA. There were sixty passengers and four crew on the American Airlines flight, and three soldiers on the helicopter; all died. But even before the National Transportation Safety Board completes its investigation—a neurotically thorough process that will probably take about a year, and thus meshes poorly with the news cycle and the conspiracy-theorist cycle—it is easy to see some of the lines of inquiry that crash detectives will pursue.

The N.T.S.B. operates on the “Swiss-cheese” principle of accidents, which holds that they hardly ever have a single cause. Lots can go wrong, but nothing bad happens unless all the holes line up in successive layers of Swiss cheese. In this case, there are a lot of slices. Two obvious questions are whether procedures and equipment functioned as intended, and whether the design of the airspace set up a potential hazard in a busy spot.

Information already in the public domain suggests that large parts of the system were working. All aircraft in a controlled airspace such as Washington’s carry a radio, called a transponder, that reports the aircraft’s identity and altitude each time a rotating radar “interrogates” it. The maps of each aircraft’s path that we see in the news, from a commercial service that eavesdrops on F.A.A. radars, are evidence that the system was working, although one detail to check is whether the helicopter was correctly measuring and reporting its altitude, and was at the altitude to which it was assigned. The Black Hawk was supposed to stay under the approach path for Runway 33, the one used by the CRJ that night.

Helicopter pilots like flying in tight spaces. I saw this in November, 2005, when the F.A.A., attempting to show off its precautions against incursions into Washington’s restricted air-defense-identification zone, organized helicopter tours of the area for reporters. I flew on a Coast Guard helicopter that slipped down the National Mall, skimming the rooftops of the Smithsonian museums. The crew enthusiastically continued over the hills to the east, and on to the Beltway, and then followed that highway just above the treetops. They had few flight hours to train in, and turned this into an opportunity to practice low-level flying, literally under the radar.

The aircraft in the collision were not invisible to radar. Another component of the system that seems to have worked was when the air-traffic controller in the tower at DCA, apparently seeing the conflict on his radar screen, pointed it out to the helicopter. (The airplane, operating as American Airlines Flight 5342, did not have much flexibility at that stage of the flight; it was in a stabilized approach, a straight line at a fixed angle, aimed for the touchdown zone on Runway 33.) Broadcasts on the air-traffic-control frequency show that the controller twice called out to the helicopter. He told the helicopter to look for “traffic,” the first time about forty seconds before the collision. The helicopter responded that the crew had the traffic in sight. The crew requested “visual separation,” meaning that someone on board (there were three people, and investigators haven’t said who was at the controls and who was working the radio) acknowledged seeing the other aircraft, and the helicopter crew accepted responsibility for maintaining separation. But the helicopter did not change course. In a second transmission, the controller told the helicopter to “pass behind” the jet. Recordings of the air-traffic frequency don’t register a response from the helicopter to the second warning. And, although the helicopter acknowledged the traffic after the tower’s first warning, its reply leaves open the question of what the crew actually saw and which lights they identified as the plane. The sky was clear but dark, and at the center of a substantial metropolitan area, crowded with lights.

Defense Secretary Pete Hegseth said that the crew had night-vision goggles. But these can limit peripheral vision and are sometimes hard to use in environments where there are some lights. Investigators, with the cockpit voice recorder and other evidence, will want to establish whether the helicopter crew was using the goggles.

Investigators have recovered the cockpit voice recorder and the flight-data recorder from the American Airlines plane; these will provide a record of what the two pilots heard on the radio, what they said to each other about the traffic, and what alarms went off in their cockpit. (The Pentagon has not said what kind of recording devices the Black Hawk was carrying.) The CRJ had a collision-alarm system, but its function is deliberately inhibited at low altitude; the investigators will have to determine if it should have gone off, and, from the cockpit voice recorder, if it did, when it did so. Those anti-collision systems have averted many midair catastrophes, often by telling one plane to dive and the other to climb. In crowded environments, conflict alarms are common and not always treated as urgent, the same way a car driver can hear an alarm while parallel parking, and consider it normal, but the controller spotted this one.

N.T.S.B. investigators will also want to know whether the helicopter received the second transmission from the tower warning of the converging flight paths. In cases of collisions on the ground, some radio transmissions have not been received because they were “stepped on” by someone else keying a microphone at the same moment.

The New York Times reported that control-tower staffing was “not normal” at the time of the accident, with one controller doing tasks usually done by two. But it is common for positions to be combined during slow periods, so another question is whether the traffic was at a level that could be safely handled by a single controller. The investigators will have to evaluate what connection this had to the collision, if any.

The F.A.A.’s stewardship of the air-traffic-control system is another likely area of scrutiny, first by the N.T.S.B. and later by the aviation community and the political world. The adequacy of the system’s staffing is a perennial issue, and, intermittently, so is the government’s role. Running the air-traffic system is one of the few nonmilitary functions of the federal government which involve day-to-day operational responsibilities, and at times the F.A.A. has had difficulty modernizing its equipment, notably its computer systems. Its nationwide complex reflects the communications capabilities of the sixties and seventies; the F.A.A. could close and consolidate many of its radar-control centers. But politics keeps the system as it is: congresspeople don’t want to lose clusters of high-paying jobs in their districts, and the controllers’ union likes things the way they are. Other countries, notably Canada, have spun off their systems to private or quasi-public operators, with some productivity gains.

The last crash of a scheduled airliner in the United States which killed everyone on board was that of a turboprop on approach to Buffalo, New York, in February, 2009. There were forty-nine passengers and crew; the accident also killed one person on the ground. In 2013, a Boeing 777 operated by Asiana Airlines came in short at San Francisco International Airport, killing three of the three hundred and seven people on board and seriously injuring forty-nine.

The system, in other words, is quite safe, with more than ten million passenger flights per year, almost all of them unremarkable. Better technology, notably for maintaining airplanes and for avoiding crashes into mountains, combined with better training and greater vigilance by government inspectors and the airlines themselves, has produced an accident rate so low that it was almost undreamed of in decades past. Crash detectives, at least the ones who do airliners, haven’t been busy.

But the risk of collisions on the ground or near it is obvious, and, as the airline industry has evolved to concentrate traffic at major hubs, it has loomed larger. Trump’s nominee for Army Secretary, Daniel Driscoll, testified at his confirmation hearing that it might be time to end training flights in the congested area where the collision occurred. On Friday, the F.A.A. restricted some helicopter movements there. (The safety record for helicopters is not as good as it is for fixed-wing planes. There is a joke in the aviation world that helicopters don’t actually fly; they are so ugly that the ground repels them. They commonly run into power lines, radio towers, or other obstacles, because they fly low. Still, they rarely collide with fixed-wing aircraft.)

The Safety Board will not find blame in the DCA crash. Arguing that prosecutorial investigations of air crashes only discourage the openness needed to develop a full understanding of these events, and to make changes necessary to avoid more deaths, it makes findings only of probable cause. In doing so, it employs a “party system,” with representation by the unions that cover the pilots and the air-traffic controllers, plus the airline, the Army, the F.A.A., and, in this case, the government of Canada, where the jet was manufactured. These parties add expertise in the arcane details of operations, but they also get the opportunity to deflect responsibility from the entities they represent. The board, an entity that is independent of the F.A.A. and the Department of Transportation, gets the final say on probable cause, but it has no regulatory authority. Its recommendations are hard for regulators to ignore, though. In many crashes, the courts get a say, too, after the investigation is done; they apportion liability.

Transportation accidents retain the ability to surprise. In February, 1996, two New Jersey Transit trains collided near Secaucus, killing three and seriously injuring sixty-nine. An engineer had run a red signal. It quickly emerged that the man was working a twelve-hour overnight shift, including a five-hour break in which engineers usually tried to nap on the commuter train’s seats. Perhaps the culprit was fatigue?

Thirteen months later, the board reported that it had found diabetes medicine in the dead engineer’s medicine cabinet. The disease had made him color-blind, unable to reliably read the signal lights. The railroad was shocked that he couldn’t see. The engineer’s personal doctor was shocked that his problem was serious enough to cause a crash. ♦