The airport beyond the glass walls crawled with airplanes moving slowly toward the runways. Thousands of passengers sat patiently in their seats. Dobkin's attention went first to the traffic pouring down the final approach. He judged the inbound lights with a familiar mixture of confidence and concentration.
The inbounds first showed up on the tower's radar screen, where we watched the work of an unseen Approach controller who was pushing the flights closer together than the tower would be able to keep up with for any length of time. Dobkin asked the pilots for speed reductions, which worked at first but soon rippled backward. The tower supervisor telephoned Approach for better service, and was told irritably that Approach itself was being force-fed by the long-distance controllers over at New York Center, who in turn were squabbling with their counterparts in Cleveland. In the meantime, because of Teterboro and LaGuardia traffic, New York Approach could not swing the Newark inbounds any wider. Approach threatened to make space by freezing turboprop departures off runway 29, a restriction that would have crowded the turboprop departures over to the parallels.
There simply was not enough room for all the airplanes, not in the air and not on the ground—at least not without delays. So far Dobkin had avoided any wasteful go-arounds, but the airplanes now were barely clearing the runway before the traffic behind flared down across the threshold. After a Delta pilot checked in with a stately drawl, Dobkin knocked twenty knots off the Continental flight that followed. To me he said, "You learn to read the signs." Delta dawdled after landing. Off the radio Dobkin snapped, "Come on, dumb boy, clear the runway." Delta did, and Continental landed short, with company behind.
The unfortunate consequence of Dobkin's success was the speed with which it was filling up the airport. To make room for the new arrivals, ground control kept pushing loaded airplanes up the taxiways toward the departure runway, 22 Right. Dobkin cleared them for takeoff as aggressively as spacing within the outbound corridor allowed. With relief provided by runway 29, the tower had managed to avoid gridlock on the ground; nonetheless, departure delays were steadily growing longer. The reason had to do with aircraft performance: while descent angles and approach speeds can be matched for most inbound traffic, optimal climb angles and speeds vary widely among departing airplanes. Moreover, because the heaviest airplanes generate especially strong wakes immediately after lift-off, additional spacing behind them is required.
For Dobkin the result was an inevitable irregularity in takeoff timing, which translated into the inefficient use of 22 Right. Ground control worked to reduce the effect by bunching airplanes by type so that they could be launched in quick order. The success of this strategy then created another problem: having landed on 22 Left and pulled onto the taxiways between the runways, the arrivals could not cross the departure runway to get to the terminals. They accumulated between the runways until, by threatening to block the runway exits for landing traffic, they forced Dobkin to hold the takeoffs. Dobkin tried hard to avoid such holdups by exploiting the natural gaps in the departure flow. He said, "It's Traffic 101: you cross behind a heavy jet, a seven-two, a prop. You use every chance you've got. You don't forget any part of it. You keep this traffic moving." What he did not say was "You keep this traffic apart."
Not that the lives-hanging-by-a-thread idea was entirely absent. I asked Dobkin about the famous burdens of the work, the toll on controllers in shattered health, divorce, and drink. "Sure," he said uncomfortably. "It's hard sometimes—I've known guys who had to get out."
Another controller joined in. "Me, I always picture the children. Nightmares, you know."
I may have smiled.
Dobkin indicated me with a tilt of his head. "He's a pilot." He meant it as a caution. The other man turned away, saving his emotional pitch for someone else.
Earlier a controller had said to me, "Stressed out? If you're the type, sure, but then it's the freeway traffic when you're driving to work that will really do it to you."
PILOTS do not believe that air-traffic control is in the business of keeping them alive, or that it needs to be. This is a matter not of principle, or of bravado, but of simple observation. The surrounding sky is so large that even when another airplane passes nearby, it remains by comparison very small. I talked to a controller involved in research with radar simulations, who said, "You'd be amazed how hard it is to vector two airplanes into each other."
Air-traffic control does have an immediate technical presence in the cockpit, but moment by moment it has less to do with the safe operation of the airplane than with the forward progress of flight. From tower control to radar to tower again, a procession of voices accompanies each airplane across the map. Their presence is humanized by accents, moods, and informalities, and by a shared sense of accommodation and competence. Good controllers are neither automatons nor traffic cops. At the start of a trip they deliver a "clearance," assigning the airplane a computer-generated route to the destination. In an uncrowded sky such a clearance might stand alone as a guarantee of traffic-free flying, eliminating any further need for controllers. In the actual sky it serves instead as a fallback plan in the event of radio failure, and as an approximate prediction of the actual flight path.
The details fill in after takeoff. Controllers thread the departing airplanes through the first busy altitudes with headings and climb restrictions. Pilots are expected not to comply blindly but rather to judge and agree. They distinguish between the controllers' wrongs and rights. Often they can even predict their instructions.
Eventually the pilots are turned loose to proceed high and fast on course, either along the airways that zigzag across the grid of navigational stations on the ground or, more commonly, in airplanes equipped with independent long-range navigational devices, directly toward the destination. Over the continental United States the airplanes cruise under the surveillance of "centers"—en route radar facilities whose role, despite the increase in traffic, remains less interventionist than passengers might imagine: across a sky so deep and wide, "control" consists mostly of monitoring flights as they proceed by routes and altitudes that have been approved by the computers but that remain essentially the pilots' or airlines' choice. With controllers' routine approval, pilots cut corners, deviate around thunderstorms, ride good winds, slide above or below reported turbulence. Controllers intervene if they see a conflict developing, or if other controllers elsewhere ask for delays or route changes. The respective roles are clearly defined: controllers may separate airplanes, but pilots still navigate them.
Like other pilots who fly in crowded airspace, I have had close calls with traffic. But "close" can mean many things. Is it a crossing that merely surprises you, or one that requires an evasive maneuver, or one so tight and fast that no maneuver is possible? Or is it—most likely—merely the violation of an official standard that may to some extent be arbitrary? Again, collisions require extraordinarily bad luck. The sort of head-on encounter in which another airplane appears as a dot and within ten or twenty seconds fills your windshield is very rare. Neither pilots nor controllers need gunslinger reflexes. Airplanes sidle toward each other slowly.
Of course, you cannot just elbow your way through bad weather and into crowded airports. Still, let us imagine an instantaneous collapse, one day, of the nation's entire air-traffic-control system. Air transportation would of course eventually grind to a halt, and large parts of the economy would soon be paralyzed, but in the meantime safety probably would not be affected. Pilots in flight would certainly sit up and pay attention—and in some cases even grow worried and tense—but they would continue to fly and navigate normally. They would find the radio frequencies printed on the charts, and talk to their own airlines' dispatchers, and radio to one another as they do already at the many uncontrolled airports. If they were originally headed for a hub like Newark or O'Hare, they might turn and fly somewhere else. Some would have to revert to cumbersome arrival routes, and many would have to hold for a while. But few pilots would feel seriously threatened. This is all the more true in modern cockpits equipped with traffic displays. In the airplane I fly today, I often spot other airplanes electronically (not to mention by looking outside) before controllers mention them to me.
Certainly the air-traffic-control system has become an ill-planned patchwork with geographic overlaps, conflicting procedures, and chance redundancies that exist as remnants of earlier times. Airplanes move from one little zone of control to the next, are spoken to across overloaded voice-radio frequencies, are handed off from one controller to another, and are given the sort of customized service that often preempts the needs of the larger traffic flow. Individual control facilities function as parallel fiefdoms, each with its own traditions, procedures, and compromises, each speaking directly to (and squabbling with) its neighbors, without passing through any central command. If you were to design a system from scratch, you would never design this. Nonetheless, one consequence of the system's haphazard structure, of its decentralization and its very inefficiency, is to scatter whatever equipment failures there are, and to provide pilots and controllers with a rich weave of choices when something goes wrong—the radio quits, the radar quits, the computer quits. No air-traffic-control equipment failure has ever yet caused an accident.
My point about the hardware is not that it is perfect but that as an educated user I do not feel threatened by its imperfections. Within such a large and complex system we can assume that equipment will wear out or become outdated, and that the government will compound the problem by responding slowly or incompetently. It is, of course, absurd that the FAA has not yet replaced all the old unreliable IBM computers that contain routing information for flights. And it is annoying that the addition of new power supplies in several centers caused outages that in turn led to major delays. And it is disgraceful that the FAA wasted hundreds of millions of dollars between an overambitious attempt to consolidate control rooms and its poorly managed, ill-conceived, and now-abandoned "advanced automation system"—an attempt to automate a wide range of internal air-traffic-control transactions. But on what basis, exactly, do people care that a controller's radar display does not contain the processing power of a personal computer (which it doesn't need)? And why do we worry that backup flight information is still written out on strips of paper? And what was the point, technically, when a Secretary of Transportation a few years ago held up an old vacuum tube for ridicule? The controllers whose workplace he meant to improve are said to have jeered at his theatrics: even he must have known that vacuum tubes are not the problem.
THE real problem lies not in hardware but in human relations. To understand this one has to move beyond the public myth, and beyond even the operational reality of the business, into a murky inner world of pride and collective memory.
From its origins in the 1920s among the agencies responsible for the new and bewildering activity of human flight, the FAA developed an institutional personality that was raw, arrogant, and domineering. For generations most controllers came from the military, bringing with them a hierarchical view of organization that was further encouraged by the nature of the work. The managers were controllers who had worked their way up through the ranks, taking pride in each small step, savoring the distinctions that marked their rise. Those distinctions may have been subtle at first, but they grew and strengthened, and eventually came to define management's style.
By the 1970s, however, a younger generation of controllers had become restive. Faced with a belligerent rank and file, the FAA management commissioned a team led by a psychiatrist, Robert Rose, then of Boston University, to conduct a study of controllers' mental and physical health. The Rose report, published in 1978, confirmed the popular impression that controllers had stressful jobs (they suffered disproportionately from hypertension and certain psychological difficulties, including uncontrollable anger and antisocial behavior), but it concluded that the causes had less to do with the pressures of air-traffic control than with divisiveness within the FAA: "This finding of 'It's not so much what they are doing as the context in which they are doing it' holds definite implications for changes that might be considered in the work environment to reduce the risk for future morbidity." In short, much of the problem lay in the way the FAA was run.