One of the things Boeing said it would do in response to the Boeing Max 8 MCAS crashes (again, user interface failures, caused by the leaky abstraction problem which is, again, a whole other topic) was to add a declutter mode with arrows telling the pilot which way to point the stick. What. The. Fuck?
\n\n\n\nAgain, even in Adrian Park's article, he spends all this time pointing out how screwed up the alarms and declutter mode were but then doesn't seem to realize that this was the central problem.
\n\n\n\nI won't even go into the infamous Qantas Flight 32 incident where an A380's engine literally exploded in mid air and shrapnel destroyed a host of different systems. Nowhere in the recommendations did anyone suggest that burying the copilot in hundreds of checklists while providing no kind of sensible overview of the system state might not be a great idea. Make no mistake: putting checklists on the screen vs. in giant folders full of paper that pilots need to search through in a crisis is a Good Thing, but again this shouldn't replace the standard UI in a crisis.
\n\n\n\nI'm hoping that the checklists were prioritized in some way but that might be giving the Airbus software folks too much credit. Any time Airbus's \"glass cockpit\" and \"game-changing automation\" comes up in Air Crash Investigation it's a pretty safe bet that a plane is going to crash because, at least in part, highly trained pilots are confused by its software.
\n\n\n\nWhy 33,000 feet? The aviation industry uses a truly bewildering collection of units including knots, feet, meters, miles, kilometers, pounds (of fuel) and so forth. I've actually seen people argue that this is actually a \"good thing\" when it comes to altitude (feet) vs. distance along the ground (meters) but—and hear me out—no it fucking isn't. Having binged several seasons of Air Crash Investigations and listened to all those cockpit re-enactments never once have I thought \"wow, the fact that altitude is measured in feet sure simplifies things\".
\n\n\n\nI should also note that West uses virtual horizons with a fixed plane overlay representing the aircraft's orientation whereas the Russians at least used to use a fixed horizon with a virtual aircraft. Leaving aside that in this case the Russians are just wrong, consider what happens in an emergency when a Russian military pilot is in control of a 737 which has two engines that deliver different amounts of thrust at a given throttle setting… If you guessed that he crashes the plane into the ground and everyone dies, you win a shiny new Captain Obvious Merit Badge.
\n\n\n\nThe worst air disaster in history (if you discount 9/11, of course) led to more than 500 people dying and is a case study in poor usability. At an airport frequently covered in fog, signage on the tarmac was unclear and did not match the maps used by Air Traffic Control. When taxiing pilots correctly reported their position as a location the controller did not recognize because it wasn't on his map, no alert was sounded, and two 747s collided.
\n\n\n\nThe fact that the controller wasn't alarmed by the fact a taxiing 747 had given a position he did not recognize suggests that this happened often enough that he was habituated to not really knowing where planes were on the taxiways.
\n\n\n\nWhen you consider the cost of the measures recommended after air crashes (e.g. replacing the assemblies of every front-flag assembly on every DC-9 in service) the cost of painting visible markings on runways, making the signs indicating taxi-way and runway identities, and standardizing the airport maps used by ATC and pilots seems pretty minuscule, yet a runway collision in Detroit resulted from nearly identical issues (also an at airport frequently plagued by fog) over a decade later.
\n\n\n\nAgain, no-one seemed terribly surprised that ATC wasn't freaked the fuck out that pilots were lost on their tarmac and reporting their positions incorrectly.
\n\n\n\nAfter the runway collision in Detroit, the US started standardizing runway and taxiway nomenclature (I think—usability gets so little focus in these shows that all I heard was half a sentence). Again, planes are big hulking beasts full of jet fuel and spinning blades with shitty visibility that steer far worse than cows and are frequently operated by pilots who are tired and unfamiliar with the airport. Why would we want good, consistent signage and nomenclature?
\n\n\n\nModern aviation is incredibly safe. There's no denying that the NTSB's approach to finding the problems in the system that lead to mishaps and then incrementally improving the system to eliminate as many causes of failure is a beacon for how we should address many of the world's problems. But that doesn't mean they don't have blind spots.
\n\n\n\nIt is so much easier to fix software than it is to fix hardware. When everything was paper forms and procedures, with a sprinkling of mainframe software with spectacularly bad user interfaces that no-one knew how to fix, the common wisdom was: training is cheaper than fixing processes, and fixing processes is cheaper than fixing software.
\n\n\n\nApple's influence (in large part indirectly, through Microsoft Windows (which was a Mac clone), Word, Excel, and Powerpoint (all originating on the Mac, if you discount Word for DOS which was replaced by Word for Windows), and the Web (strongly influenced by HyperCard and built on the NeXT computer) showed that making software that didn't suck wasn't that expensive, actually reduced training costs, and allowed you to encapsulate and iterate on processes via intuitive user interfaces.
\n\n\n\n(And yes, Xerox PARC and Douglas Englebart and all the rest were channeled through Apple but—seriously—go try using a Xerox Star and compare it to the Mac or Lisa.)
\n\n\n\nThe Aviation Industry seems to be locked into the 1970s \"command line\" mentality, where people are invested in the sunk cost of learning terrible systems and think that hard-won knowledge is intrinsically valuable because it was hard-won. Cars have had automatic chokes and synchromesh for over fifty years but pilots still need to deploy flaps for takeoff or which autopilot modes can cause stalls during icy conditions. If your answer to most questions is \"it's a training problem\" you're asking the wrong questions.
\n","$updatedAt":"2024-06-05T09:10:29.725+00:00",path:"air-crash-investigation-when-usability-failures-kill-people",_created:"2024-07-09T20:28:42.370Z",id:"7438",_modified:"2024-07-09T20:28:42.370Z","$id":"7438",_path:"post/path=air-crash-investigation-when-usability-failures-kill-people"},"page/path=blog":{path:"blog",css:"",imageUrl:"",prefetch:[{regexp:"^\\/(([\\w\\d]+\\/)*)([\\w-]+)\\/?$",path:"post/path=[3]"}],tags:["public"],source:"