Skip

"But, since you asked, buckle up."
March 15, 2012 1:56 PM   Subscribe


 
Man. For some reason, I read this link as "Quora: What do all the animals in the airplane cockpit do?"

I was mistaken, and now I'm bereft.
posted by bicyclefish at 2:00 PM on March 15, 2012 [8 favorites]


Along the lines of this post: Airbus A320 - From Cold and Dark to Ready for Taxiing. A useful video to watch if you want to feel prepared in case you ever need to steal an airliner in an emergency. They also have a video for the Boeing 737 Classic.
posted by zsazsa at 2:04 PM on March 15, 2012 [6 favorites]


I wish there were some sort of abbreviated, offhand way I could dismiss that very long answer. It was so long, in fact, that I really was unable to read all, or even most, of it.

I did locate the most important sentence, about one third of the way down the page:
"The rear knob has no function."

Now, when I glance into the cockpit as I'm boarding a plane, I'll always think to myself, "The rear knob has no function." That alone is enough for me.
posted by perhapses at 2:10 PM on March 15, 2012 [9 favorites]


In those pictures, the FMC looks like a TI-59.
Less so close up.
posted by MtDewd at 2:14 PM on March 15, 2012




I suppose simply writing "they help fly the plane, duh" would have been too dismissive.
posted by ShutterBun at 2:25 PM on March 15, 2012


Is Quora like the internet's version of Dora the Explorer?
posted by Fizz at 2:26 PM on March 15, 2012 [2 favorites]


tl;dr: they make you not fall down go boom
posted by idiopath at 2:26 PM on March 15, 2012 [2 favorites]


Thanks for this. I've printed it out and will carry it on every flight I board on the chance I'll be called into the cockpit to give advice, or on the even rarer occasion, am asked to fly the plane.
posted by Metro Gnome at 2:29 PM on March 15, 2012


Wow, I'm hoping James May catches wind of this.
posted by obscurator at 2:43 PM on March 15, 2012 [2 favorites]


Where is the inflatable auto-pilot?
posted by munchingzombie at 2:51 PM on March 15, 2012 [7 favorites]


You know those adverts that thrillingly promise you can speak a new language if you give it just five minutes a day?

Then you realize you have to concentrate really, really hard for those five minutes? And you think that, actually, maybe you won't bother learning a new language after all?

That's how this post made me feel. Then I read the disarmingly candid post title - a direct quote from the enthusiastic author of the piece -"But, since you ask, buckle up..." and I feel such a negative & lazy shit...

(Great post!)
posted by Jody Tresidder at 3:00 PM on March 15, 2012 [1 favorite]


I really enjoyed this thank you.
posted by Meatafoecure at 3:01 PM on March 15, 2012 [2 favorites]


Reading that article, I can't help but feel the gulf between airline cockpit design and consumer electronics design. That cockpit had so many user-unfriendly knobs and controls, I couldn't believe it. OTOH, I am glad to know that there's a backup backup way to do all the critical flying-the-plane functions.
posted by Phredward at 3:11 PM on March 15, 2012


I was surprised at how many of the knobs control screen brightness. How useful is that exactly?
posted by simonw at 3:15 PM on March 15, 2012


I was able to take a flight from Antalya to Manchester in the cockpit of an A-320 back in 1999, before they stopped allowing such things; I was flying jumpseat with my brother, who's a pilot. I was surprised first by the fact that there wasn't any yoke, just the joysticks. I was then surprised by how automated the whole thing was, and how little the pilots actually did during the flight after takeoff and reaching cruising altitude; just sitting around reading between radio contact to handoff between airspaces and twiddling a couple knobs to change course. The view from the cockpit was incredible (at least, compared to normal airline windows). Flying over the Taurus mountains and Anatolian plateau was an amazing show of color (the rainbow of brown!) and topographic texture. We left in the late afternoon and the sun was setting over the Aegean Sea before we crossed into Bulgaria. I was amazed at how much of that part of southern Europe was dark at night. We had to avoid then-Yugoslavia because of the Kosovo war, which apparently caused some annoying extra knob-twiddling and worries of military traffic.

It was one of those small-world experiences, not just because we traveled over so many familiar-sounding countries so quickly, but because as we sat in the cockpit and read, I saw a piece in the Sunday Times about the enslavement of an adopted girl in a small Tennessee town that I spent my summers in. She was kept for years trapped in a building I used to ride my bicycle by all the time, and this was happening back when I was riding by.

When we came into Manchester, there were 30-knot crosswinds. It was really bizarre seeing the landing strip coming up, and realizing just how much the plane rolled, pitched, and yawed as we came in. Everyone had shut up in the cockpit to do their job and I was thinking, man, we sure are moving a lot and we sure are coming in fast and then a bright red light flashed on above their heads and this weird Star-Trek phlegm-hymen in the throat voice started saying, "RETARD....RETARD....RETARD...."
and I was thinking WHAT THE FUCK DOES RETARD MEAN and the lights at the end of the runway were coming fast and I thought for sure we were undershooting the runway and my heart jumped and then we just bumped down and skittered a bit and no problem, we had landed. I asked my brother later what the fuck the red light was, and he said it was just telling the pilot when to throttle down. Meanwhile, I had very nearly shit myself. But still the best flight experience I've ever had.
posted by Red Loop at 3:32 PM on March 15, 2012 [25 favorites]


DON'T THINK! ... DON'T THINK!
posted by ShutterBun at 3:48 PM on March 15, 2012


Somebody's playing Falcon 3.0 on the Primary Flight Display. I'm no expert, but that's just asking for trouble.
posted by tumid dahlia at 3:49 PM on March 15, 2012


The article images came from Microsoft Flight Simulator X which has thousands of incredibly detailed 3rd-party models to buy.

If you're feeling adventurous, a 'freemium' successor was recently released called Microsoft Flight that lets you fly a couple of small planes for free. It's supposed to appeal more to casual players, which has obviously irritated stalwarts in the flightsim community.
posted by rh at 4:02 PM on March 15, 2012


A few months ago I started to teach myself to fly jet airliners by trial and error. Not the real ones, obviously! On a computer. I started out with a little handheld computer app, and worked my way up to X-Plane, the flight simulator for people who think regular flight simulators are insufficiently nerdy. The advantage of X-Plane is that its primary function is allowing pilots to practice flying their planes. So there's very few training wheels (other than the fact that the planes are stupidly resilient) and when you do something wrong, you really know about it.

Not knowing anything significant about aircraft, of course, it's been an interesting exercise. I read the manuals so I could learn what the basic instruments did, and I watched some online tutorials, but most of my 'learning' has been produced by just trying stuff and seeing what happened.

Here are a few of the things I've learned:

1. Flying is even safer than I would have thought. It's very, very hard to accidentally crash a big jet airliner, even if you're a complete neophyte. The engineering that allows these massive things to handle so well is amazing (especially when you compare it to flying a plane from sixty years ago), and when you figure out how the autopilots work, they will even land themselves. If I were ever in a 'Zero Hour' situation, landing a plane with the help of ground controllers, I'm quite confident I could land safely, even though I suck at flying planes.

2. That said, the interface design of several critical subsystems, including FMC's and autopilots, is terrible. I had always been told that, to a layman, those thousands of knobs and dials might be confusing, but that they're actually a miracle of ergonomics allowing for the instant recognition and diagnosis of problems. Turns out that this is only true in some cases. They also have these systems which were apparently designed in the dark ages, and they still haven't updated them for the days of computerisation.

The autopilot is the best example I've found. I always assumed that the autopilot would just fly where some kind of computer told it to fly. While it can do this, it's usually doing something more complicated and... in a way more simple at the same time. There are a series of individual things about the plane that the autopilot can control. For example, there's climb-rate, altitude, heading, bank angle... a whole bunch of things like that. So you'd think that the autopilot would have a series of lights and buttons saying which aspects of flight are being controlled. You might say 'fly East, and climb at 100 feet per minute until you get to 30 000 feet, then fly level', and the lights for 'heading', 'rate of climb' and 'altitude' would light up (altitude probably in a different colour because you're not holding at 30k ft yet). Oh no! That would be far too simple.

What actually happens is that the autopilot has a series of modes it can be set to, and each mode controls a different set of flight characteristics. When a mode button is lit up, you just have to remember which flight characteristics are being controlled, and if you want to do something complex, you have to get the right combination of behaviours by combining modes - by activating several mode buttons - some of of which are compatible with one another, and some of which are incompatible. This is made even harder by the fact that the autopilot controls are scattered around the cockpit. There's usually a main panel somewhere in the middle, but other, critical controls are often quite far away from it, snuck in among other controls that do something quite different.

When you have the autopilot set wrong, you get absolutely no indication that the thing isn't working until Bad Things start happening. It's also quite difficult to tell, with any rapidity, exactly what combinations of flight characteristics are being controlled. Just reading the thing involves doing a little logic puzzle in your head. And if you've set several incompatible settings, well, I guess they're just going to write 'pilot error' on the incident report.

Before I tried doing this myself, I was always a little astounded by incidents in which planes had accidents because nobody noticed that the flight control system was set wrong (example, example). Now, with the benefit of my trial and error education, I can easily imagine how such incidents would occur.

3. Oh, right, lesson three... apparently when you fly into a cloud during a snowstorm your plain can get covered in ice and crash. I'd always heard people talk about 'icing', but I had no idea what it actually entailed. Learned that one the hard way!
posted by Dreadnought at 4:08 PM on March 15, 2012 [6 favorites]


How useful is that exactly?

There's not really a way an airplane could use "headlights" like a car does. So you need the control to be dim enough for the crew to see outside at night without the help of lighting, and bright enough during the day to not wash-out.
posted by Monday, stony Monday at 4:09 PM on March 15, 2012 [3 favorites]


Q: What is the ideal cockpit crew?
A: A pilot and a dog...the pilot is there to feed the dog, and the dog is there to bite the pilot in case he tries to touch anything.
posted by autopilot at 4:45 PM on March 15, 2012 [10 favorites]


Is Quora like the internet's version of Dora the Explorer?

Well, they do have basically the same hairdo.
posted by Strange Interlude at 4:48 PM on March 15, 2012 [1 favorite]


I was surprised first by the fact that there wasn't any yoke, just the joysticks.

Airbus uses joysticks, while Boeing uses yokes.
posted by Harpocrates at 5:23 PM on March 15, 2012


That is one of the worst interface designs I have ever seen. It is as if somebody laid out the cockpit, then found a bucket of randomly labelled buttons which they scattered hither and yon.
If a software developed designed a program so poorly, they would be shot.
Hell, it makes Wordperfect for DOS look user friendly.
At least, try and group controls by function, and arrange back up systems either near the primaries, or together in a separate 'back-up' zone. As it is, sometimes the failover systems are here, sometimes there, and occasionally, both.
Surely this design fail is aggravating emergencies.
posted by bystander at 6:31 PM on March 15, 2012


Pretty epic post, for sure. I... uhhh... didn't read it... because... you know.... but still, it's nice to know it's out there.
posted by ph00dz at 6:42 PM on March 15, 2012


I started reading this thinking: I will need to know this in the event of some catastrophic failure where I find that I am immune to oxygen deprivation and must wrest my self Sully-like from 29B in the back of coach and take over the controls of the plane. I will have to push aside the passed out pilot, don the headset, and tell the air traffic control "Mayday, Mayday....passenger two-niner-bravo here....I have taken over the controls...let's land this bird..."

But then I started drifting off and yawning about two paragraphs in (so much for that oxygen deprivation thing). So, I started over. But then I lost my place when I tried to find the knob in the picture and then go back to the text. Then I thought, I probably wouldn't even know how to use the headset. Then, I wandered off looking at flight simulator games instead for about 10 minutes. Then, I came back to the article one last time, but zoned out again after trying to convert it to Readability mode for a minute or two.

I finally just saved it to Instapaper and moved on. I can always just pull that up again if I need it.
posted by This_Will_Be_Good at 6:43 PM on March 15, 2012


Red Loop,

I had the distinct pleasure of sitting in the cockpit of the 747-400 during take offs and landings many, many times. Always a thrill. I remember the first landing, sitting in the jump seat and as we were about to cross the threshold, a stark woman's voice loudly saying

DECIDE... DECIDE... DECIDE...

It freaked me out a little and I asked the pilots what that was all about after we'd parked. Basically, it was the final chance to do a go around. Beyond that the plane was going to be on the ground no matter what.

It's so sad that it's next to impossible to ever have that experience again.
posted by michswiss at 6:54 PM on March 15, 2012


Admittedly, it's not that bad once you understand what's up -- and it's much better than it once was. Back in the day, you had to read tons of dials, and it quickly became work-intensive. And I'd say that most stuff is grouped logically. You've got your basic airspeed/artifical horizon/altitude/vario on one display, and the navigation stuff on the other. Then all your radio stuff is mostly in one place, all the autopilot stuff is together, as is the hydraulics things.
posted by Monday, stony Monday at 7:03 PM on March 15, 2012


I've been told that a modern airline pilot has only three decision to make during a flight:

1 - take off or abort;
2 - land or go around;
3 - tea or coffee.
posted by Monday, stony Monday at 7:05 PM on March 15, 2012 [2 favorites]


I was amazed at how much of that part of southern Europe was dark at night.
posted by Red Loop


Well ... I'm not sure how you sophisticated Europeans do things, but here in Canada, all of the continent gets dark at night.
posted by PareidoliaticBoy at 7:23 PM on March 15, 2012 [6 favorites]


I suspect that the problem of cockpit controls — their arrangement and complexity and all that — is similar to the problem of the QWERTY keyboard.

We can argue for hours about how well-suited it is for typing and whether it's better or worse than other ways of getting text from a brain into a computer. But it's a system that was designed the way it was for reasons other than user-friendliness, and additions to the original system have been stuck wherever they'll fit.

And if we ever decided, en masse, to redesign the keyboard from the ground up, it'd be extremely disruptive to productivity and would provoke all kinds of backlash.

I suspect that few people in aviation would be interested in the disruptions that would be involved in a radical re-imagining of airplane cockpit controls. Instead, it'd be better to make incremental changes, such that each successive new model of plane would feature some small and uncontroversial relocation of a knob from here to there, and we'd end up with something much better a few decades hence.

(Except that probably won't work because planes don't get replaced very quickly — the 737-600, for example, was first produced in 1996 and Boeing is only now starting to talk about its replacement.)
posted by savetheclocktower at 8:18 PM on March 15, 2012


The field that I'm currently doing graduate work in, Human Factors, was somewhat born out of the design of cockpit interfaces and the aircraft manufacturers are still some of the primary employers. I'm not really into that aspect of the field, but from what I've heard the placement of buttons and levers and whatnot used to be a lot worse before they started thinking about this stuff. You would think that with such an effort and concentration in this realm, in addition to the stakes, the cockpit would be the most usable and safe interface design ever, but from what I hear savetheclocktower is right.

You can't completely re-arrange cockpit designs that pilots are used to precisely because the stakes are so high. Gradual improvements are all that they can do without substantial re-training, and airline companies don't like spending money on stuff unless they think the cost-benefit makes sense. The FAA is only going to push so much and unless the airlines are convinced that an improvement is going to substantially reduce the chances of a plane crashing and people dying (which is expensive to them in many ways) they don't want to bother.
posted by Defenestrator at 8:32 PM on March 15, 2012 [1 favorite]


Yeah, defenestrator, I agree. One thing, however, that doesn't seem to change- no matter how much effort goes into the human factors aspect- is just how damn difficult it can be to hit the right button or grab a knob when you're being bounced around in heavy turbulence!
posted by drhydro at 9:09 PM on March 15, 2012


And just forget about those &^%$%$#@# touchscreens.
posted by drhydro at 9:10 PM on March 15, 2012


The guys at Giant Bomb took about 40 minutes trying to take off in DCS: A-10C Warthog.

John Denver's fatal plane crash was a standard example of aircraft UI design failure twenty years ago, I assume that's still the case?
posted by vanar sena at 9:14 PM on March 15, 2012 [2 favorites]


A very large proportion of aircraft accidents could have been prevented with better UI and human factors related design, the recent Air France 447 crash being one of them. The sticks that determine altitude were apparently not easily visible to other pilots and one of the more inexperienced co-pilots had his stick pulled up the entire time the plane was stalling (a big no-no I gather). Not only were the other pilots unable to visually see him doing this, his inputs were averaged with the inputs of the other pilot instead of one overriding the other.
posted by Defenestrator at 10:02 PM on March 15, 2012


Speaking of flight sims my favorite was Falcon by Spectrum Holobyte on the Macintosh. You got to pick out your armament before the mission, the number of migs you fought and if you were doing a noise-dive a female voice would tell you to "pull-up" over and over. There was a nice cannon, long runways, a kick-ass compass, flaps, landing gear, throttle of course, etc... The Harrier flight simulator for the Mac was also quite nice that came before that. If you were followed by a missile just release some flares with the space bar. Nothing, nothing beats vertical take-offs!
posted by Meatafoecure at 10:33 PM on March 15, 2012


Q: What is the ideal cockpit crew?
A: A pilot and a dog...the pilot is there to feed the dog, and the dog is there to bite the pilot in case he tries to touch anything.
posted by autopilot at 6:45 PM on March 15


ICWYDT.
posted by hypersloth at 10:36 PM on March 15, 2012 [2 favorites]


I suspect that few people in aviation would be interested in the disruptions that would be involved in a radical re-imagining of airplane cockpit controls.

I'd also like to imagine that commercial air travel's relatively stellar safety record (even when adjusted for miles traveled vs. total number of trips) is a pretty good indication of things not being broken.

QWERTY keyboards may have been initially designed to intentionally slow people down (or at least force them to alternate hands semi-regularly) but cockpit design seems to have evolved more organically. And human brains are remarkably capable of re-learning (although yes, time and money do come into play)

The Apollo astronauts had to learn to fly the LEM on two separate axes, depending on their attitude, and this proved no problem. Drivers from foreign countries (despite anecdotes to the contrary) can typically adjust to driving on the opposite side of the road pretty quickly. Typists on QWERTY keyboards can easily adjust to Dvorak keyboards. Even people equipped with "upside-down glasses" will begin to see things right-side up within a few days.

So, this is all suggests that a radical redesign of the cockpit could be both beneficial, and hassle-free. (introduce it as a prototype, train a handful of pilots, etc.) The question then becomes: is it necessary at all? The John Denver scenario seems like a complete outlier (experimental aircraft, not built to specs, single pilot, etc.) so it's probably not fair to use that example at all. Certainly there have been situations where a "not easily accessible / identifiable control" has resulted in a crash or mission failure ( SCE to AUX strikes me as one example) but given the organic nature of cockpit evolution over the years, I'd be kinda surprised if there were a whole lot of "obvious improvements" to be made in one fell swoop. The shit just WORKS.

I definitely get the impression that pilots (both current and former) are very involved in cockpit design innovations, so it would be surprising to me if there were a lot of them calling for a total redesign, as opposed to a few tweaks here and there.
posted by ShutterBun at 11:06 PM on March 15, 2012 [1 favorite]


According to Matt Damon's pilot character on "30 Rock," the only buttons pilots use are 'take off,' 'autopilot' and 'land.'
posted by Gilbert at 11:39 PM on March 15, 2012


Thanks for posting this, zarq. My 9 year old is going to love it- when he isn't drawing planes he's playing XPlane on my iPod, reading Jane's Airline Recognition Guide, begging me to take him to LAX or Bob Hope Int'l to planespot or, worst of all, making me watch endless episodes of Air Crash Investigation* on YouTube.

*This episode about British Airways flight 5390 is one of the most harrowing things I've ever seen. It just reinforced my kid's desire to be a pilot, so I guess he's chosen his course..
posted by biddeford at 12:03 AM on March 16, 2012 [1 favorite]


According to Matt Damon's pilot character on "30 Rock," the only buttons pilots use are 'take off,' 'autopilot' and 'land.'

"You know what a great pilot would have done? Not hit the birds. That's what I do every day: not hit birds. Where's my ticket to the Grammys?"
posted by savetheclocktower at 1:10 AM on March 16, 2012


Where's the "eject snakes" button?
posted by arcticseal at 2:02 AM on March 16, 2012 [2 favorites]


If I were ever in a 'Zero Hour' situation, landing a plane with the help of ground controllers, I'm quite confident I could land safely, even though I suck at flying planes.

I marvel at the audacity of that statement. I've logged time flying both powered aircraft (small fry) and gliders and I would seriously worry if I were called on to land an airliner - and I don't suck at flying.
posted by oxidizer at 2:54 AM on March 16, 2012 [1 favorite]


I marvel at the audacity of that statement. I've logged time flying both powered aircraft (small fry) and gliders and I would seriously worry if I were called on to land an airliner - and I don't suck at flying.

To be fair, the statement did include "with the help of ground controllers". While I would not want to be in that situation, it's not that far-fetched for it to work. (Mythbusters actually did a segment on this, where Adam and Jamie went into a flight sim for a commercial airliner having never flown one, and were talked through a landing by an experienced pilot. They managed to do it just fine.).
posted by tocts at 5:05 AM on March 16, 2012 [1 favorite]


Back when my brother was working for an airline (not as a pilot) he arranged a little outing down to the simulator facility (Alteon, I think), courtesy of some friends he'd made amongst the pilots.

This was quite a big thing. Airline pilots spend a lot of time in simulators, getting the required ratings to take off and land and airports around the world. Modern simulators are very accurate, and also very expensive - pilots spend thousands and thousands of dollars on this part of their training. We were going a couple of days after Christmas when no-one uses the facility, but it was still some impressive string-pulling on my brother's part.

The simulator itself looks like a big white cube supported by hydraulic legs. Inside it's a faithful recreation of the cockpit of a specific aircraft - in my case a 737-800NG. The hydraulics provide all the physical sense of pitch, roll, acceleration and so on. All the cockpit windows are monitors, of course. Your senses are totally fooled.

So you sit in the pilot seat in this simulator capsule, and the trainee pilot is beside you in the copilot seat, and the trainer is behind you setting up the scenarios and talking you through them. Takeoff. Landing. Landing at night. Landing at night in bad weather (seeing the airstrip at Melbourne Airport materialize frighteningly close out of nowhere was a surreal experience).

At one point I he was talking me through banking, just doing a lazy pass around the Melbourne CBD (and by the way, the yoke in those Boeings is surprisingly heavy) when I saw another aircraft below and to the left, and I mentioned it out loud.

The copilot craned his head to look. "I don't like the bearing."

The displays all flickered and suddenly we were on the runway again.

"What happened?"

"I quit the scenario" replied the trainer. "I didn't want to end up doing a close approach and risk a collision."

"Why, what happens if we 'collide'?"

"It's, uh, bad" said the trainer. The trainee pilot nodded emphatically. And that's all they would say.
posted by Ritchie at 5:27 AM on March 16, 2012 [1 favorite]


I like the little buttons next to the important warning lights that you press during the pre-flight check to make sure those lights aren't burnt out.
posted by smackfu at 5:43 AM on March 16, 2012


That is one of the worst interface designs I have ever seen. It is as if somebody laid out the cockpit, then found a bucket of randomly labelled buttons which they scattered hither and yon.

This has been the Week of Things I can Expound At Length On:

WHY YOUR AIRPLANE IS NOT LIKE YOUR CAR (OR YOUR IPHONE)

There are a lot of competing factors that go into the design of a cockpit, but the primary few are:
-Easy access to information
-Easy access to controls
-As little confusion as possible

There was an incident a few months ago where the copilot went to unlock the cockpit door to let the pilot back in, but that switch was conveniently located next to the "autopilot freakout" switch or whatever he actually pressed, because the plane made a sudden dive and threw a bunch of people around instead of actually opening the door. Probably a bad design choice to have those two buttons next to each other.

Your car, compared to an airliner, is a simple machine that requires very few interactions to operate it. Gas, brake, steering wheel, switch for the lights, stalk for the turn signals, stalk for the wiper blades. That's pretty much it, everything else is gravy. Imagine if you needed a switch for every individual light in your car (left headlight, right headlight, left taillight...) and you're starting to get close to where aircraft complexity is. Lots more functionality, not much more space to put it than your average sedan.

Now put a second steering wheel in your car for the front passenger. He or she needs all the same controls you do and all the same information you have. Now you have a second set of instruments, and the available space to put controls just went down. In addition to this, all of your controls need to be able to be pressed, turned, or whatever when you're bouncing around in heavy turbulence. Each pilot needs to reach every control in the cockpit without getting of their seat or moving around too much, so that further limits where you can put things.

So how do you deal with something like this? Controls get bigger - buttons get big so that you can hold on to them. They have strong resistance to motion - knobs that turn take a bit of force and provide a very satisfying click when moved to the next position. They need to be spaced far apart so that you can fit a gloved hand around each one individually without activating others by mistake.

I can't help but feel the gulf between airline cockpit design and consumer electronics design.

The reason for this is that your iPhone can break and no one will die. There are layers of requirements and regulation that govern what things can look like in a cockpit, how the react when you interact with them, and how the underlying code is designed. You would never, ever get an iPhone or iPad certified to be permanently installed in an airplane, and here's why:

UI is very important, and several people have touched on this already. Your iPhone looks very pretty, but it does not convey information very clearly. Let's take... display of weather data as an example. I was on a regulatory committee that was designing new recommendations for how to display weather data. Easy enough, right? You see bands of color on your TV every day with the morning weather report. What problems do you have simply dropping that on to a screen in a cockpit?

First off, colors. There is cultural significance attached to colors. Green generally means everything's good, but on a weather display it means there's precipitation. Red is Very Significant in cockpit design - anywhere else it means PAY ATTENTION TO ME NOW OR YOU WILL DIE. Red on a weather display means heavy precipitation, but not the heaviest precip! Magenta is even worse! Now we have a conflict where on one display red means "bad, but not too bad" and everywhere else it means "HOLY FUCK WE'RE ALL DOOMED!" Green is also used to color the terrain map - how do you distinguish green terrain from green weather?

You have no idea how many hours I spent on the phone talking about the color red.

Second, clutter. You have a lot of information on some of these displays - terrain, traffic, waypoints, airports, flight path, and finally your weather data. What's most important? Do you just overlay weather on top of terrain? What if the terrain is more important right now because you're about to crash into a mountain and wouldn't you like to know about that except you've got happy green clouds covering that information? What about icons that indicate wind velocities, icing conditions, or other hazards? Do those take priority over traffic icons? Does wind take priority over icing? Should you be allowed to turn all those off?

Third, staleness. Data comes in over a datalink connection, and this data is gathered from weather stations across the globe. How old is this data? How do you indicate to the pilot that new data has come in? What about if no new data comes, do you wipe his screen so he can't use old data? Have a warning pop up? What color should the warning be (back to the red/not red again)? Should be be allowed to clear that warning, or should it persist until new data arrives? How old is too old for weather data? And on, and on...

How do you react with all of this information? There are buttons on all the displays, what do they do? How many levels of menu structure should be allowed? How much control does the pilot have? What if there's a conflict between what the pilot sets and what the copilot sets? You want functionality to live at the same button for every different display option, but what if you run out of buttons? Can you shuffle them?

Finally, software. Computing equipment in the cockpit of an airliner must be certified to DO-178B, level A. Level A in this document means that every piece of functionality must be tested, all failure modes must be tested, and every single line of code must be run during testing. More than anything, this is why you will never see an iPad permanently installed in an airplane. I worked for company that certified their equipment to a lower level, and we only had to hit every branch in every decision tree; it would have been impossible to hit every line of code. To make a certifiable system, the code needs to be simple and so the functionality must be simple. Sometimes that means taking the long way around, so to speak, from a UI perspective in order to make a product that can actually go in the aircraft.

This was all kind of long and ranty, but this is the sort of thing that makes engineers grind their teeth when "designers" complain something doesn't look pretty enough.
posted by backseatpilot at 5:48 AM on March 16, 2012 [27 favorites]


Your car, compared to an airliner, is a simple machine that requires very few interactions to operate it. Gas, brake, steering wheel, switch for the lights, stalk for the turn signals, stalk for the wiper blades.

Not to mention that people would probably complain about a car's obscure controls if they just saw a picture of it. "You twist the stalk to do one thing, and push it to do something else entirely, and pull it for another function, and this is all labelled with cryptic symbols?? How can anyone drive a car?"
posted by smackfu at 6:22 AM on March 16, 2012 [2 favorites]


Why? This is all you need to rely on.
posted by stormpooper at 6:32 AM on March 16, 2012


"I have no training or experience in your field, but having casually looked at your solution for a few moments I'm going to explain why all you so-called experts, who do this for a living, are doing it wrong."
posted by LastOfHisKind at 7:46 AM on March 16, 2012 [7 favorites]


I marvel at the audacity of that statement. I've logged time flying both powered aircraft (small fry) and gliders and I would seriously worry if I were called on to land an airliner

But that's exactly my point. Small aircraft are much more difficult to land in routine situations (I'm quite confident airliners get hard in complex, non-routine situations). The 'Zero Hour' landing procedure for an airliner goes like this: fly the bearing the guy tells you to (using autopilot), descend to altitude and bearing the guy tells you to (using autopilot), tune Nav 1 (as per instructions), activate LOC mode, put down the flaps, put down the wheels, pull the brakes. No VOR bearings, no visual approach, no monitoring of speed and altitude... pretty much nothing other than twiddling some knobs to the right numbers when the guy on the radio tells you to.

In fact, given that you have actual flying time, I might even be safer landing the plane than you, because I would be less likely to second-guess the ground controller when things looked different from the way they do in a small plane. My very ignorance would protect me.
posted by Dreadnought at 8:22 AM on March 16, 2012


Flashbacks from my first real job. Working at NASA on 757 simulators. I wasn't an aero engineer, so I couldn't program the flight and engine modules. Thus, I got the grunt work of testing each and every damn one of those @#$%@#$% buttons, switches, dials, inputs, lights, whatever.

While I may curse that my days were spent in the "hangar" (think sheet-metal walls and a roof put over a former parking lot), it was a fun job. For some controls it was a matter of making sure the button pressed registered in the code (and in the required amount of time). For others, I had to have the simulator "running" (ie to pull yokes/throttle/flaps and what not to drive certain indicators). This was before they put the graphics in, and wires were still exposed everything. So I have a small claim to "I can fly a 757 blind" (well, I used to..)

And I did learn that auto-thrust/auto-pilot can both take off and land the plane.
posted by k5.user at 9:29 AM on March 16, 2012


That's an amazing skill to be drunk and remember what all these controls do!
posted by slogger at 11:04 AM on March 16, 2012


Metafilter: I have no training or experience in your field, but having casually looked at your solution for a few moments I'm going to explain why all you so-called experts, who do this for a living, are doing it wrong.
posted by yoink at 11:15 AM on March 16, 2012 [2 favorites]


To be fair, the statement did include "with the help of ground controllers"

Ground controllers or not, I don't think your average person will manage to manually land an airliner. I don't think that Mythbusters episode really proves anything to the contrary. It's one thing to do this on a sim, quite another to come barreling in on a runway with 200+ thousand pounds of aircraft under your butt and perhaps a dim notion of how to fly a go-around - assuming they get to that topic during the talk down.

But that's exactly my point.

Sorry, it wasn't clear to me that you were referring to an automated landing.
posted by oxidizer at 11:44 AM on March 16, 2012


I stand chastened and corrected when it comes to criticizing aircraft user-interface design. It's absolutely true that I come to the subject with a fully amateur eye and I very much bow to the professionals in this thread.

That said, I'm still very interested in the subject, and many of the decisions made just don't make intuitive sense to me as an outsider. I would be fascinated to learn why these controls are laid out the way that they are. If any of your aircraft interface designers could explain to me, I would be both grateful and enthralled.
posted by Dreadnought at 11:48 AM on March 16, 2012


Sorry, it wasn't clear to me that you were referring to an automated landing.

Oh I see. I thought that it was implicit given my previous comment about autopilots landing the plane, but I guess I wasn't clear enough.

Having said that, it's worth nothing that in these big planes, everything seems to happen much more slowly than in a light aircraft. While they're actually much faster, the scale of the runways and the plane itself is just greater. Things take much longer to actually happen, which makes the whole flying experience much more forgiving.

Then there's all the artificial stabilisation and the safety systems built into the controls. A little plane gets buffeted around by turbulence and the mass of the aircraft is small enough that little twichy movements will radically change the way the plane is pointed. Modern jet airliners are stabilised by a computer and pretty much keep flying in the direction you point them. Very often they won't even let you fly into the ground. I'm not sure exactly how it works, but I notice that, when flying at low altitude, if you push forward on the controls, the plane will actually fight you as you try to put the nose down. Many times I've managed to crunch a (simulated) Cessna into the ground short of the runway, but I've never done this in an airliner. The controls start pushing back, and you get a lot of 'PULL UP' audible warnings letting you know you're screwing up and have to cancel the landing.
posted by Dreadnought at 11:57 AM on March 16, 2012


All those buttons and not a single way to set the pre-sets for the radio. It reminds me of an older office phone system we used to have. I swear there was some way to launch a rocket or order tea with a press of a button, but God help you if you simply wanted to redial the last number.
posted by dgran at 12:10 PM on March 16, 2012


yoink: "I have no training or experience in your field, but having casually looked at your solution for a few moments I'm going to explain why all you so-called experts, who do this for a living, are doing it wrong."

To be fair, this is one of the best places on the web for avoiding that kind of thing. In that it's one of the only places where I can read a thread like this without thinking "I have no training or experience in this field either, but the flaws in ExplainToTheSoCalledExperts's takedown are so glaringly obvious I can't read any further.". Not to mention the fact that LookedAtYourSolutionFor1Second, NoTrainingOrExperience and YouAreDoingItWrong have joined with ExplainToTheSoCalledExperts in monopolising the thread.
posted by ambrosen at 12:22 PM on March 16, 2012 [1 favorite]


Q: What is the ideal cockpit crew?
A: A pilot and a dog...the pilot is there to feed the dog, and the dog is there to bite the pilot in case he tries to touch anything.
posted by autopilot


I thought I had not heard this joke before you posted it, and then was just watching some very old QI and Stephen tells it!
posted by lazaruslong at 3:01 PM on March 16, 2012


That said, I'm still very interested in the subject, and many of the decisions made just don't make intuitive sense to me as an outsider.

This is why we get paid the big bucks! Or, you know, the small-to-medium-sized bucks.

As was mentioned upthread, a lot of this stuff started out because back in the 1920s someone decided they needed a way to tell how high off the ground they were and the design just stuck around for awhile. Why should an altimeter look like a clock or a tape? I feel like it's very similar to how file systems for computers were dreamed up - well, it works for my filofax so why not just transition that over to this new block of metal sitting on my desk.

There are standards - lots and lots of standards - that govern everything from what radio frequencies you use for what purposes down to how big the font size on the dial should be. They're called Technical Standard Orders and you can search for them on the FAA website. Most of these TSOs will reference an industry standard for guidance (RTCA and SAE are the two big ones, I think), so be prepared to spend a few hundred dollars per document if you really want to read them.

For something like an electronic display (TSO-C113), the FAA defines what everything must look like in order to be granted the TSO. That particular one is nice because it actually has a table of color coding information right in the TSO - red for warnings, yellow for caution, etc. How did these colors get decided on? Again, it's mostly cultural history and for other stuff with less of a "normal" meaning attached to it (like the color blue, for example), metaphors are decided on. Sky is blue, so the artificial horizon shall be marked by blue above and brown below. If you're overlaying data on top of that, you don't want to use a color that's close to blue lest it be obscured or hard to see. Text needs to be a certain height to be readable. A lot of it is common sense stuff.

"Why doesn't this look like my iPhone?" is a little bit more difficult to answer. When I was working on this stuff, we had a full-time graphic designer on staff just to visualize graphics and layouts, so it's not like there's no thought that goes in to this stuff. However, the regulations usually don't allow you enough wiggle room to have fancy things like drop shadows and beveled corners. Using fancy fonts might look nice, but could impact readability. There are tradeoffs.

As far as things like buttons and knobs, again there are regulations - size of knob, distance between knobs, force required to press button, tactile feedback of button press. Turn the volume knob on your car's radio and you'll notice a little bit of play before it "catches" and starts adjusting the sound; that does not happen in an aircraft. I can go fly and grab a knob in my fist if I have to.

Finally, a lot of decisions are made on availability of parts and cost. There are not a lot of glass displays certified to operate from -40 to 150 Fahrenheit, and if the only one you can find has a native resolution of 640x480, then so be it and now all your graphics look like blocks. Maybe adding an extra multi-purpose button adds a thousand dollars to the cost of the equipment - is that worth it? Is your processor certified? Is your operating system certified? We were using an embedded version of Windows NT because nothing else had the proper certs to go on an airplane!

These decisions really don't happen in a vacuum and given the choice of usability or gloss, the answer is obvious. That being said, you can find some pretty slick looking equippage, especially in the general aviation sector - Garmin and Avidyne are the big players, but there are others.

The unasked question, I feel, is more along the lines of, "Why can't I sit down in a cockpit and immediately know what everything does like I can with my smartphone?" And the answer I will give you is that there is a lot more going on in a cockpit than you will ever get your iPhone to do. These things require training. I just recently started flying the G1000 and I'm still floundering looking for buttons that I always have at hand when I'm flying other cockpits. It's just not going to get to be like automobiles where you can sit in front of any steering wheel and go for a spin.
posted by backseatpilot at 3:03 PM on March 16, 2012 [4 favorites]


The key thing about the airplane cockpit is that it exposes control over all components of the aircraft to the pilot. You'll notice there is a lot of control over fuel supply, electric supply, etc. This lets the pilot adjust the operation of the plane in an emergency so it keeps flying. It also lets the ground maintainence crew and pilot test and verify everything prior to takeoff.

Imagine if in your car you could switch the fuel pumps, water pump, off if neccesary, and you also had backup pumps, alternators, etc. in case one of them broke, and restarting the engine while going 80 down the highway needed to be a very straightforward procedure despite all this and despite the automation your car does have (auto. transmission, cruise control, whatever).

If you watch the video linked to by zsasa at the top of the thread you'll notice that to some degree, the overhead panel (hydraulic and electrical switches) in particular of the planes are sort of layed out based on startup procedure, or at least his startup procedure fits with the layout of the switch panels in a straightforward manner, so there is some rhyme and reason. (Of course there are certain controls that need to be closer to your hands than others as well.)
posted by thefool at 3:45 PM on March 16, 2012


Thanks very much for some really insightful answers, backseatpilot.

What puzzles me is not the failure of a complex jet airliner to look like an iPhone. I'm more intrigued by the apparently strange decisions made to lay out systems in ways that seem to defy easy, rapid access and understanding on the part of the users.

I'll return to my example of the autopilot from above. Let's describe the layout of a hypothetical airliner, the Dreadnought 9000. I base the design choices, in this description, on a real airliner design with which you might be familiar.

The D-9000 is a large, wide-body intercontinental airliner. Like all such modern planes, it needs an autopilot to do three things:

1. to conduct preprogrammed emergency maneuvers (eg. TOGA)

2. to accept input from automatic navigation systems (FMC, VOR, LOC, etc) and essentially fly the plane by wire.

3. to automate specific flight tasks as the pilot instructs (eg. hold altitude or change flight level)

For simplicity's sake, let's say that the main panel which controls the autopilot is right in the middle of the control console of the cockpit, at the top, right between the two pilots.

Job 1 is the easiest to design: they're only going to be needed in an emergency, and they're going to be needed quickly, so the designers of the D-9000 have put special buttons on the control column which the pilot can trigger at a moment's notice.

Job 2 is rather more difficult. The autopilot has to be able to accept inputs from several different navigational systems located in different parts of the cockpit. Firstly there's the VOR navigation system which picks up radio beacons on the ground. The VOR system is controlled by a set of radios located by the pilots hip, with the numbers oriented to the rear of the plane so that the pilot has to read them sideways. The readout for the VOR system is on the console in front of the pilot, and is read by observing several 'bugs', or ditto-marks on the edge of the HSI readout (like ships and submarines, the aircraft has numerous compass readouts but, unlike in ships and submarines, each one has a different name and behaves non-uniformly). Then there's the FMC (the computer which controls preprogrammed flight paths). The FMC control is located down to the right, just behind the knob that controls the flaps, so that it's difficult to reach without accidentally jostling this critical flight control. Then there's the LOC system which automatically lands the plane. This doesn't have its own control interface; it just has a button on the main autopilot control panel.

In order to choose between these inputs to the autopilot, the designers have provided a selector knob. This is not located next to the autopilot or the VOR radios or the FMC; it's on the pilot's left, behind the control column, in a row of other knobs which look identical to it, but do completely different things. But this selector knob has no way of engaging the LOC automatic landing system. That system is engaged by a completely separate control maybe a meter to the right.

I've already talked about the oddities of job 3 in my comment above, so I won't bother you with them again, except to say that the addition of these additional modes and jobs make them even more difficult to interpret.

So that's the kind of thing I'm interested in: why are the controls spread out? Why is it that controls which do radically different things are designed to look the same and placed right next to one another? Why are critical systems placed behind other critical systems such that they're difficult to reach in flight? Why are they set up in such a way that it's difficult to tell exactly how the system is set at a glance?

I'm not surprised that a plane is not set up like an iphone, but I am surprised that its controls are not grouped by function and designed to give unambiguous feedback.
posted by Dreadnought at 4:28 PM on March 16, 2012


forgive me if I missed this up-thread.

A tremendous amount of the complexity in cockpits has to do with upgrade cycles. The 737 discussed on Quora originally had a flight engineer and a navigator for over-ocean flights. That's twice the flying crew with more than twice the instruments. Now all of that instrumentation (much of it albeit further automated) has been jammed between the captain and the first officer. Clean sheet designs, like the 787 or even the Airbus when it first came out) significantly reduce clutter and at least try to improve the interaction modalities.

Another major challenge has been the migration from mostly analog instrumentation to most digital instrumentation. It has happened frequently in small steps with concomitant disorganization. I've been flying for only 20 years and have seen massive change. My speculation, along the lines of some what backseatpilot implies, is that we will see cleaner design/HCI in the future. As long as the FAA can play well with the manufacturers.
posted by lomcovak at 4:32 PM on March 16, 2012 [1 favorite]


Why is it that controls which do radically different things are designed to look the same and placed right next to one another?

Gotta love Kirk's armchair controls on Star Trek: 1. Yellow Alert 2. Red Alert 3. Eject Pod
posted by ShutterBun at 7:51 PM on March 16, 2012


Along the lines of this post: Airbus A320 - From Cold and Dark to Ready for Taxiing yt . A useful video to watch if you want to feel prepared in case you ever need to steal an airliner in an emergency. They also have a video for the Boeing 737 Classic yt .

I've just finished watching both of those videos, and now I'd like to see a qualified 737-NG pilot do a "holy shit we are repossessing an airplane from a bankrupt airline which is located at an airport run by a sketchy third world junta and we need to start it up ABSOLUTELY AS FAST AS POSSIBLE" run on a simulator. How long would that take?
posted by thewalrus at 3:49 AM on March 17, 2012


A tremendous amount of the complexity in cockpits has to do with upgrade cycles. The 737 discussed on Quora originally had a flight engineer and a navigator for over-ocean flights.

Old 737s were not certified for ETOPS and rarely made long over-ocean flights. 707s and such, yes, but not 737-400 and earlier.
posted by thewalrus at 3:50 AM on March 17, 2012


@thewalrus:

my father was a USAF navigator involved with taking very early versions of the 737 from Cal - Hawaii. That was in the early 70s. I think ETOPS is relatively recent? like 2007? Your point is valid for commercial stuff for certain.

kinda irrelevant anyway; we've been taking meatware out of airplanes and replacing it with technology for a long time with somewhat predictable, and ongoing, challenge.
posted by lomcovak at 12:22 PM on March 19, 2012


« Older Big Brother at the Petrol Station   |   Skywalker Sound Newer »


This thread has been archived and is closed to new comments



Post