Something I’ve been occasionally hinting at over the last half dozen years or so has been the apparent collapse of basic flying skills in some pilots.
Although most pilots will continue to proudly trot out the party line that being a pilot requires consummate skill and years of training, in my opinion most pilots spend most of their time ‘flying a computer’ and being administrators, rather than actually flying a plane as you and I would understand it.
Sure, ‘flying a computer’ – ie, programming the plane’s auto-pilot, telling it where to go, etc, is a complicated task in itself, and shuffling through the tons of paperwork that accompany any commercial flight also requires skills and training. So today’s pilots aren’t entirely unskilled. But their skills have migrated away from what you might know if you’ve ever played with a Flight Simulator program on your computer.
Even when they have a ‘hands-on’ experience flying their plane, it isn’t true flying. Many modern planes ‘interpret’ the pilot’s commands and translate them into valid actions, a bit like how a car with auto traction control, auto stability, and antilock braking will interpret and override non-optimum driving actions, a bit like how adaptive cruise control will not only drive your car at a constant speed, but will now vary it based on other traffic on the road, and a bit like how a car with auto transmission will change gears on behalf of the driver.
In the case of flying a plane, not only does the plane filter the pilot’s instructions, but most of the time, even in ‘manual’ mode, the pilot still has ‘auto-rudder’ and ‘auto-throttle’ engaged. The former means that to change the plane’s heading he just needs to turn the control column like a regular car driver turns the steering wheel – no longer does he have to coordinate the turn by applying rudder correction with his feet on the rudder controls at the same time as he turns the ailerons with the control column – one wonders if modern pilots even know what a slip indicator is. The latter means that if he pulls back on the control column to climb, the plane automatically adjusts its throttle settings to maintain a constant speed, and the same, if he pushes forward to descend, the plane automatically backs off the power.
There are other automatic things too that are halfway between a full autopilot and what passes for manual flying, like simply inputting a bearing into the flight director and allowing the plane to turn to that course automatically.
None of this is bad, but it isn’t ‘flying’ as we imagine it to be. The pilot no longer controls the plane. He controls the controls – indeed, on some planes these days the main control is now nothing more than a computer joystick, much like you’d get at Best Buy to use when playing Flight Simulator. It is the controls, not the pilot, which in turn then control the plane.
This only becomes important when there is a discontinuity in the normal chain/sequence of events. When the controls stop behaving properly, then today’s pilots are woefully unfamiliar with manually flying the plane the old-fashioned way – as we used to say, ‘by the seat of your pants’ (ie actually feeling and sensing how the plane was moving).
There’s another factor at work as well. I know that some airlines – possibly all – restrict the type of training they allow their pilots to do in the flight simulators, too. Dangerous maneuvers are forbidden. I asked an airline training captain, after the US Airways A320 landing in the Hudson River, if they trained for such events. No – apparently the thought of a water landing was scary and verboten; better to pretend that such a thing would never need to happen.
I asked another airline training captain if I could manually fly their simulator under a landmark bridge. They said they don’t allow that. Okay, so there might be little real-world application for a virtual flight under the Brooklyn Bridge, but it points to an attitude where pilots and the planes they fly are cossetted carefully in cotton wool. Remove all their flying aids, and what do you have left?
The reason for this lengthy introduction? One of the automated landing aids is out at SFO at present, and the FAA has noted a significant increase in the number of aborted landings, with flights breaking off their approach and going around to repeat a second time, in particular on flights from foreign carriers.
So they are now recommending that foreign airlines and their pilots use added GPS assistance (rather than doing a simple visual landing) on their approaches to SFO until such time as all the other landing aids are restored.
And then, they added a second thing – this time a requirement rather than a recommendation. Foreign flights are no longer allowed to simultaneously approach the two side by side runways. US flights can still do this, but foreign flights now have to keep more separation between their planes.
The thing about flying is that taking off is optional, but landing is mandatory. We’re creating a generation of pilots who don’t know how to land their planes. They can program a landing into their plane’s computer, but they can’t do it themselves.
If you read the pilot commentary in this newsletter (heading ‘Some damning details of Korean pilots and their training’, near the end the pilot writing the commentary says that modern-day pilots typically get only a very few minutes of actual flying experience for every five or ten hours of accumulated flight time. So your pilot with 2,000 hours of apparent flying experience might actually have less than 20 hours, spread over the last three years. And, while he doesn’t say so, his observations about the difference between logged ‘flying’ time and actual ‘hands on the controls’ flying time are as true of US pilots as they are about foreign pilots.
The situation is a bit like a person who has only ever driven auto-transmission cars suddenly finding themselves in a stick shift. But at least, that person can simply walk away from the car. 100 – 500 lives aren’t at stake.
Is automation making our pilots more dangerous, rather than more safe?
great article, i am an x B727 pilot and I totally agree with you.
David, Aviation is structured to work when things go well. It is not structured to work when things don’t go as planned. You hit on one fundamental issue, I’d like to raise a couple of other core issues as well.
One core problem is that a landing is almost an engineered drop, IOW an engineered falling out of the sky if you will. There should never be such situations where the pilot (or the computer) does not have active control of an aircraft. Yet landings are: Often the plane has to be on idle throttle which means it is not in active control. You are riding on momentum and air. Once you pass below a certain speed and height you can never take off again. This works when things go as planned. But we all know that many times things don’t go as planned. Right now when that happens there is no real solution. And with millions of takeoff and landings it is a mathematical certainty that things will not always go as planned.
Changing this style of landings and takeoff means a fundamental change in aircraft design, engine design, etc. so that landings and take-offs are always in active control (or at least to the last second). If you always need to “fall out of the sky” on a slope, you have a problem. This is why a vast majority of the accidents happen during landing. And vice-versa for takeoff.
While it will take some geniuses solve the above fundamental problem and build aircraft that can abort landings and fly around till almost the last second, there are some interim improvements that need to be made. Aviation is a few decades behind in technology. One could probably write an iPhone app that would know that the rate of descent is too fast / speed is too slow. Why are aviation institutions so far behind in technology? The irony is that the Asiana jet was landing in silicon valley, where such an app could be written and commercialized for less than a million dollars. What are we waiting for?
As for air traffic control, there are now 3D projection systems that can show all the planes approaching visually in an area the size of a dining table (or an area projected in to your smart glasses). Similarly, planes should also have this visual available to them so there are always aware of the traffic around them at all times. It’s easily possible with today’s technologies. Google’s automatic car knows which cars are around it but planes don’t know which planes are around it. Why?
As your article shows, aviation institutions have never thought about what happens when things don’t work as planned. Why are we expecting that nothing will go wrong, ever? When are we going to fix these fundamental issues of preparing the pilots for emergencies, of the aircraft not being in active control during the riskiest times such as take off and landing and not using technology for the most basic information?