Season 3 · Episode 12
Cleared for Upgrade
19 November 2026 · 1 hr 18 min
Show notes
Air Traffic Control seems particularly topical this week. A full ground stop by the US FAA and days later tech trouble grounded United. Not to mention Microsoft and the challenges that created for global infrastructure.
Little slips of paper don't seem all bad now do they? It wasn't the slips of paper that failed this last week. So, we decided to look at the history of Air Traffic Control upgrades.
Give it a listen while you're flying. Or sitting on the tarmac.
We'd love to hear from you. Click here to give us ideas on new episodes.
Join Renee and Marc as they discuss the history of technology, and what it teaches us about now.
email us at nostalgicnerdspodcast@gmail.com
Come visit us at https://www.nostalgicnerdspodcast.com/episodes or wherever you get your podcasts.
Merch is available on the site too. It helps cover the cost of the podcast.
Full transcript
Auto-generated and lightly edited. Marc & Renee.
Marc: Okay, Rene, we're going to talk about some war stories that you and I have lived through. Some of them at the same time. Yes. Wind the clock back. It was probably not 99. It was probably 2000. It was like after Y2K. Yeah, yeah, yeah. We got pretty good at upgrading from one thing to another. And i remember you know i think it was exchange like we went from you know a piddly little exchange server and then we we we basically planned it all out and had the new servers racked and running and everything switched and the dns moving and man that windows dns was crap by the way and you know yeah whatever but
Renee: That's why the incubator people used to laugh at us oh That's why we were made fun of, but go ahead.
Marc: Yeah. Well, I remember the blue haired kid. He used to say, oh, you guys are running, running lookout, you know, instead of outlook. Come on, dude. The blue haired kid. Anyways, but like we got that exchange server up and racked and running and, you know, switching and moved over, you know, in days. And that's, you know, that's a, that was, and I don't remember any issues that we had particularly with that. like email was flowing, email was moving.
Renee: They weren't always that good though. I mean, I remember doing one, this is before the days of Salesforce when your CRM platform might've been web enabled, but it was still running inside your organization, right? And it was like this, it was CRM, it was email CRM. Now remember this is 1990s, late nineties, early 2000s when email was still running. It wasn't, yeah, people still read it and stuff. It was a thing. It was a thing, right? And so you were going to try to manage all your customer information through, or customer correspondence through email. Like that was something a dot-com was doing back in the day that was pretty revolutionary. It probably did open the door to Salesforce and everything that came after it, was the idea that the dot-coms were like, no, our primary source of communication will not be an 800 number. It will be email. And if you can't do that, then don't shop here. kind of thing right like and so we were trying to upgrade that platform so okay okay okay we didn't try once or twice like six times but but the one i remembered was, going to the cio and saying listen here's the deal we need a cot, and i remember him looking at me like what do you need a cot for like you don't understand like six hours it will take us to just find out if this thing actually converted, this database. And then we got to do it again. And it'll be another four hours. So it'll take 10 hours total before we ever know whether or not that we pulled this off. And this is round three of this crap. So like, we need a cot. Someone has to be here for 10 hours making sure this thing doesn't blow up. And he just looked at me and he's like, Ikea, okay? Like he was going to go to Ikea to get it, like in his two-seater. Like an Ikea one, okay.
Marc: Yeah, yeah. Yeah. What was the name of our DBA that had to suffer through that process? I don't, I don't even remember who that, who that was.
Renee: Oh, he's Portuguese. Do you remember who?
Marc: Yeah, yeah, yeah.
Renee: Milk's a Deadly Poison. I think he gave me that book for my birthday.
Marc: Oh, no, that was Joe. But no, the other, the DBA guy that was like,
Renee: Yeah. From Afghanistan?
Marc: Yeah, maybe it was from, I don't know. No, that wasn't Brett. No, Brett.
Renee: No, it was the guy from Afghanistan who we would sit down and have lunch together and he would tell me horrible stories about like. Oh, yeah, yeah, yeah, yeah. About traveling in Afghanistan.
Marc: For Joe, that was, yeah, oh, I won't mention the name of the platform.
Renee: Joe's the one who did the upgrade with me. It's me and Joe sharing the time because it was just like, and at one point, we're just staring at each other like, why is it this bad? And I'm like, I don't know. I don't know. But we were two versions out of sync with the current version. And so something crazy probably happened. But at any rate, at some point, we figure out if we just ignore the instructions and do it our way, it'll work. And so we did, and then we wrote down the new instructions and gave it to the vendor and said, do it like this from now on because your way doesn't work.
Marc: I remember those instructions, man.
Renee: We did it with the cluster server for SQL, too. Remember, we tried to follow the build doc and it just didn't work. And we're like, fill the build doc out. Let's figure out how to do this. And then we figured out how to do it, me, you, and Jay building side by side each head. We're building a database, right? And then when we were done, we were like, oh, the documentation's wrong. Somebody get Microsoft on the phone. The documentation's wrong. So there, there you go, people. Like, you think you're up against something with your AI layering on on top of your COBOL? Yeah. Yeah. Okay. You don't even know. You don't even know what it was like just trying to get two versions of an upgrade done. You know who does know what that's like? This one guy I met when I was at Forrester as an analyst, he called me to say he was doing a, a customer like reference for sap, and he said it took me eight years but i have finally scripted my upgrades like eight years eight years you've been working on it and you had to tell him congratulations because i can't believe you got it done in just eight years right like you're like oh my god congratulations dude like i'm just really proud of him like oh that's wonderful like that's terrible that we should all be excited about that.
Marc: Yeah yeah no kidding i remember i I remember SAP, there was an SAP guy that I worked with at the gas company, and the guy literally only worked for six months. So he would do a project, and he made so much money on a project in six months that he would literally, he bought a place. Just take the rest of the year off. Yeah, he bought a place in Greece, and he lived in Greece for half the year, like vacation. Yeah, because SAP was so gnarly and hairy. The dude never needed it. Yes.
Renee: A whole cottage industries popped up around them. There was whole risk management industries around them. Yeah. Yeah. Upgrades. All kinds of companies that do nothing but API integrations for other stuff. Like, yeah, like a whole cottage industry popped around something, popped up around that. And just to get completely off track on what it takes to actually upgrade, everybody turned to the big four to say, well, then you manage this. And I don't think they did it because they were any good at it, because I think they fail the same way that all the rest of us do. Yeah. Right? But you did have somewhere to put the liability, and it wasn't on yourself or your team. It was on that third party. And I think that has stuck around for 30 years.
Marc: Yeah. I think we still do. Blame somebody else when they don't get it right.
Renee: Yeah. Yeah. I got to do an upgrade. Let me hire someone I can blame.
Marc: Yeah.
Renee: Okay. That seems crazy, but.
Marc: A great risk reduction.
Renee: So, okay, but imagine this, right? Imagine you're doing this kind of SAP crazy upgrade that's going to take you like eight years to get right, right? You're doing the same job to a system that decides which airplane goes first. Like, you're going to do this for air traffic control. You're going to modernize, right? Like, you're going to do this for planes. And you know what? There is never a good day to take this network down and you cannot have a Sunday off. So how are you going to do this? How are you going to do this? And I think that is the real question that I would like answered in this moment. Like, if you're going to modernize air traffic control, like, how do you do that when you can never take a day off?
Marc: Yeah. Yeah, I think that's the real constraint in these kinds of systems. The FAA has set it out, right? They have to keep the current system running the entire time they build the new one. This is worse than doing an upgrade because you have to run a complete production system while you're building a completely separate system and then figure out how to do it and cut over. You don't get a Sunday there. You get rolling migrations where the old system and the new system are both alive. And one patch of the country sky is on one side. Well, you know, the next patch might be on the other system.
Renee: Okay, so I'm going to be annoying about this because everybody says the technology is what makes it hard. But I don't buy that. Something like 40,000 flights a day. And the hard part, I think, is the people sitting at the scopes, right? A controller has trained for years on exactly one set of screens, exactly one way a data block looks, right? That's not reading. That's muscle. You move one button, you move the muscle. The muscle is the system. Like if you are going to modernize, it's almost like it starts in training. It doesn't start with technology. It definitely starts somewhere else. Technology isn't the thing that keeps this thing going.
Marc: Yeah, I'll give you the muscle part. Yeah. But I'll add that, you know, that that's that's the part that's just ugly engineering, though. Right. It's it's not one system. A flight goes from a tower to an approach facility to a center and then back down. And each one of those layers has its own computers. Right. Bought in different CD. We guess decades that have have to land the flight to the next one without dropping it. So when you place, you know, one piece, the new thing has to speak a dialect from the 1980s. So the old thing, you know, to the old thing that's sitting next door to it.
Renee: And you know what? They talked about it last year. Windows 95, floppy disks, paper stripes. There were strips, right? Strips, yeah. Right there in the conference, in the press conference. And I laughed. And then I thought, I fly a lot. Like, I shouldn't be laughing about this, right? Four hours at a terminal. And the text from the airline is said that there's an ATC issues, right? That was from Newark that spring. Like two words doing a tremendous amount of hiding. And what failed wasn't some exotic supercomputer or some AI upgrades. No, it was a copper wire carrying radar data to Philadelphia. A wire. A single point of failure in communication from one tower to another. That's what it was.
Marc: Well, yeah. Dubious. I'm skeptical of the, yeah.
Renee: Okay, so you and I have done our share of network engineering in this world, right? Like every telecommunications, that last hop, that last mile, that single point of failure between you and the central office, it's always the last mile, right? Yeah. And so like a lot of people think like it's so redundant. The cloud is so redundant, maybe, but there comes a point where you're going to run across. It's like that Simpsons episode. Remember when Homer types something in and you followed the internet, and it's going through all these tubes and all these crazy places and it ends up in this closet with this old man sitting in a rocking chair and then there's a Cisco box and it says Cisco on it and there's a bunch of flies around it because it's all old and there's a sign on it that says the internet. Like, yes, at some point you will hit that nasty fly server. You will at some point on the internet. It will happen to you. So, yeah, a wire. So something happened, right? And that was another grounding of the entire airspace because things were out of sync and no one had situational awareness of the flight.
Marc: So screens went dark for something like a minute and a half. And in a world where planes are flying at, you know, a few hundred miles an hour, a minute and a half is a very long time. The airplanes had their own collision avoidance. So they've got that system. So an official said they could still fly safely. But the controllers, the people that actually, you know, actually do the work there on the, you know, making sure they don't crash, they lost their picture.
Renee: And the plan is a brand new system by the end of 2028, which from today is 25 months. Oh, wait, no, it's not. They just went live with it everywhere in America three days ago. So, yeah, I guess it's done.
Marc: Well, clearly, there's some challenges.
Renee: And it's sort of like healthcare.gov, but we'll get to it. I'm going to say this country has been rebuilding air traffic control flights since 1929. And I want to talk about how it went every other time we tried it, because it's probably a pretty good yardstick about how it's going to go.
Marc: Yep, every upgrade in the history of the system happened while the system was running.
Renee: Even the wheelbarrow?
Marc: Yes, Renee, especially the wheelbarrow.
Renee: All right, here we go. Oh! Welcome to the Nostalgic Nerds Podcast. I'm Renee.
Marc: And I'm Marc.
Renee: Today is air traffic control. From the first controller to the floppy disks to my claim, my claim up front is that the most successful invisible technology in American life, I know we think it's the refrigerator, possibly the doorbell, but no, no.
Marc: Double entry bookie.
Renee: No, no. It might be air traffic control. You know, I change my mind every time, but I'm going to say it. It might be this.
Marc: The sewing machine.
Renee: It might, yeah, it might be secondary traffic control.
Marc: Okay. All right.
Renee: I mean, I like wearing clothes. There you go. But I could cut a hole in a sheet and still be covered. I cannot fly my own plane. Yeah, I can't fly your own plane. You know what I mean? Yeah. So when it works, nothing happens. It's all of IT, I think. Like, when it works, nothing happens, right? Something like 45,000 flights a day, and the entire product is that you need to read a paperback and land. Like, that's it. That's what they're selling you. That's the product. You just come here, step onto this chair, sit down, read your paperback, get off, go home, right? Or wherever you're supposed to go. And the very first version of that product was a man with two flags, Marc, and a wheelbarrow.
Marc: All right. Archie League. There's the man. He was hired on January 1st, 1929 by the city of St. Louis for Lambert Field, a licensed pilot and a licensed mechanic who barnstormed around Missouri and Illinois with a flying circus. I don't know if people even know what that barnstorming word means anymore, but, you know, they would basically fly these planes, you know, really low and, like, buzz barns. Yeah, through barns.
Renee: Or through barns.
Marc: Yeah. A red flag meant hold, and the checkered flag meant go. Like, you still see, you know, semaphore flags and stuff, especially in naval. But, like, yeah. Do you see semaphores on, you know, the LAX runway today? No. But our friend Archie here, he carries both of them out to the field in a wheelbarrow, along with a chair, an umbrella for the summer sun, a lunch pail, you know, because he has to eat, and a drinking jug.
Renee: A lunch pail. And here's why St. Louis. Two years earlier, Charles Lindbergh, whose home field was Lambert, crossed the Atlantic, the whole country went airplane crazy. Suddenly, every town wanted an airport, and every airport had a crowd of people who wanted to fly, and nobody had decided how any of that was going to work. So before League, the rule for the sky was basically that pilots would look out the window and just, I don't know, sort it out themselves. I mean, are you just signing to one another? Are you flipping each other off? Like, what are you doing? How is this working, shouting at each other. Amazing. Good.
Marc: Well, yeah, yeah. It's a perfectly workable system until it isn't, right? And think about how it ran. So the pilot has no radio to answer with, so it's a one-way conversation, a man with flags and a pilot with eyes. So the controller has to see the airplane and the pilot has to see the flag. So daylight, no problem, line of sight, you got decent weather. The interesting part is that in October of 29, League talked a biplane down through dense fog, which the record treats as the first time anybody was guided to a safe landing by air traffic control. That's also where the whole idea hits its limit, since a system built on eyesight has a problem with fog, or darkness.
Renee: Radio came to the tower in the early 1930s, and Leak became the airport's first radio controller, which I think is nice, because it means the first controller is also the first person to get his job upgraded. I bet he went home to his wife and was like, they're going to give me a radio. And she's like, oh my God, they're going to fire you. That radio is going to take your job, honey. Like, oh my, you're going to find another airport. We're going to have to move. Like, no, no, it made him better. But that's only the airport. For the airplanes, in between airports, we have to go to December of 1935, when a group of airlines opens the first air traffic control station at Newark.
Marc: And that was an airline job, not a government job. Airlines built that first station themselves.
Renee: I like that as a person who's blown commercially in this century, the airlines had the most to lose. So they built the shared referee themselves. And so when you hear people say, or industries say we want to self-regulate, like that's what they mean. Like that's what we think they mean. We think they mean this. So when you hear the AI industry say we'll self-regulate, this is not what they mean. They don't want to work this hard at it just so we're clear.
Marc: They don't want to build, you know, nationally viable infrastructure. No, no. Internationally mobile infrastructure. No. I wonder if there are other countries that, because the states, air traffic control is, you know, a government gig. I wonder if there's other countries where
Renee: Air traffic is. Is it UK like a half-half? Like it's part. Yeah. Like there's a part of it that's national security, so it leans toward the government, but then the commercial part of it leans toward private industry. Yeah. Isn't that how Nats works?
Marc: Yeah. It's split somehow. Now, I don't know. Andy knows all the ins and outs of that one. Yeah. So, all right, so walk the room with me here. There's no radio link to the pilots. Controllers talk by telephone to airline dispatchers and radio operators on the ground. And from those reports, they build a picture where every airplane is likely to be.
Renee: So the airplanes are doing 150 miles an hour over New Jersey, and the people responsible for them are on the phone, moving little strips of paper around on a rack. The sky was being managed like the front desk of a hotel.
Marc: Yeah, the strip, I love the strip. We were just talking about that before we started recording.
Renee: We were just talking about this.
Marc: Yeah. The strip is a bigger deal than it sounds. So just hold on for a second here and I'll explain it. So it's about six inches long and one inch wide. And it carries the flight's call sign, its route, its altitude, and what it's expected over the next fix, which is a navigation waypoint. The strips sit in a rack in time order, and the controller physically slides them around to keep the picture straight. The strip is the airplane, as far as the system knows. If it's in the wrong slot in the rack, you've lost the plane. And hold on to the strip because it's still live in 2026 and it's going to cause us some trouble.
Renee: And where does the strip get its information? A pilot radios a position report when they cross the checkpoint. A radio operator writes it down and phones it in, and somebody slides a strip into the rack. The freshest thing in the whole picture is a telephone call about where an airplane was like three minutes ago.
Marc: So controllers got very good at dead reckoning in their heads, working out from speed and heading where an airplane probably was right now. And remember, this is, you know, we're talking about this is before radar at scale. So every system that comes after this is an attempt to replace that guess with measurement.
Renee: And here's the part that I think is super important because it's the first piece of our theme. Within a year, the federal government takes over the airline run station and, doesn't throw them out and start over. Three of them become forerunners for the control centers we still have today. And the first modernization of air traffic control was a change of ownership. And the planes kept flying the whole time. So we take it out of the hands of the airlines. We put it in the hands of the government. The government creates the FAA. So this is your problem. and somehow we all kept flying.
Marc: Yeah. Can you imagine like, oh, you know what? We're going to change who's running this. And so, you know, we're just going to shut everything down.
Renee: Right. Like everybody stop, stop, stop, stop, stop, stop, stop, stop.
Marc: You can't fly for a week while we fix this.
Renee: Yeah, while we do the paperwork on the backside.
Marc: Hold on. We got to file that report. Yeah. So they had to do this, right? Because traffic was only going one direction. Is up. More traffic, more planes, you know, more passengers. By the war, that's the Second World War, the number of centers goes from 15 to 27 because everybody is flying everywhere and nobody has the option of pausing the system to renovate. Every addition gets bolted onto something that's already carrying passengers. Oh, doesn't that sound familiar? You know, you got something and then you bolt on an upgrade and you put on another upgrade.
Renee: And the next thing you know, AI will solve at all.
Marc: Stop. Yeah. Oh.
Renee: We're skipping most of the war on purpose because the thing that changes everything is what the Navy and the British have been doing with radio waves.
Marc: All right. We just did radar. This is awesome. I know. We did a microwave. Yeah. We did a microwave and then talk about, yeah, because it came from radar and now we're going back. Let's go. Out of the hood.
Renee: All right, let's go. My claim about radar, and I will defend it from any chair you put me in, radar did not give controllers airplanes. It gave them dots. Glowing, beautiful, completely anonymous dots.
Marc: I guess that's technically true, right? But it gets worse the more you know. A primary radar sends out a pulse and listens for whatever bounces back. So every dot is just something that reflected. A 747 is a dot. A flock of starlings is a dot. A thunderstorm is a smear of dots. The dot has no name. It doesn't have altitude, doesn't have an airline, only a position that's a few seconds old.
Renee: So the first identification system was a controller keying the mic and saying, turn left 30 degrees, and then staring at the screen to see which dot obeyed him. And then he would be like, oh, that must be that. Okay, I know who that is. Right? Like, I'll be fine now. That's the technology when the first demonstration radar tower goes up in 1946. And it's more or less the whole workflow. I mean, that's pretty much all it was. Eight turn left. You see it turn left. You're like, that's him. Like, wow. Wow. That's amazing.
Marc: Yeah. And they kept track of that by hand. So the controller set a little marker on the glass next to the dot. The things they, you know, nicknamed a strip boats and nudged it along as the dot crawled. That works for a handful of airplanes. It stops working somewhere around 50. I mean, can you imagine the little, like the little markers? Because I've seen the markers. And they stick them on and then they move them around. It's like a game board, you know, the little cheap pieces on the game board.
Renee: Funny you should say that because if you make it through the written test and then the psychological evaluation and they decide they want to train you to be an air traffic controller, you will spend the next two years coming into an office with a bunch of other soon-to-be air traffic controllers. And I'm not kidding you, two years. I think it's 24 months or something like that. Yeah. And you play a video game where you and all these people... Manage air traffic that's not real because you're being trained on how to so if they crash it's not a big deal because it's fake right and so i would.
Marc: Want to just play the game
Renee: Right like i'm just thinking right now do you think do you think it's in the app store like could we go find one i don't.
Marc: Know i have you seen those guys like i've you could watch them on youtube where i think it might be microsoft flight simulator it might be some other simulator game but like it's so accurate you know you've got they these guys get so like into all the atc calls and everything like like it has to be precise and
Renee: Exact yes and they have their own air traffic control and they have their own i didn't realize that these guys who build out cockpits in their garages yeah and they say i'm gonna go fly to hawaii they file a flight plan they talk to towers towers aren't there this flight's not real yeah exactly there's a whole aviation like hobbyist thing that.
Marc: Is sort of weird, but cool. It is very weird, but very cool at the same time.
Renee: Yeah. Hey, if anybody listening has built out a cockpit of a 747 anywhere in the house and is doing any of that, come on and talk to us about that. I would love to hear it. I would love to hear it. So would everybody else. We're all nerds here. And that's before. Okay. So it stops working somewhere around 50, right? And so that's before you even count the sweep. The antenna rotates. So a dot on the screen isn't where the airplane is. It's where the airplane was the last time the beam went past it, which makes the whole scope a rumor about the last, like, few seconds, right? That's what, that's, if you watch air disasters, they'll be like, I see them, oh. Oh, yeah, they're gone. Like, what? What? Like, how did that? And you just realize, well, they're plummeting somewhere, right? Like, they're just gone. You're just like, in a second. In a second.
Marc: Yeah, and that rumor, the thing, you know, that little dot, the rumor on the scope, it depends how old that thing is. It depends on the radar itself. So a terminal radar near an airport sweeps every five seconds or so. And the long range, you know, on route radars take about 12. An airliner at cruise covers a mile and a half in 12 seconds, so every dot on a center controller's scope is already behind its airplane, and the whole system has to be built around that lag.
Renee: So the fix is the one thing is in this entire history that I'd call elegant. Just make the plane talk.
Marc: Yeah. So it's called the transponder.
Renee: Oh, I love this transponder.
Marc: Yeah, I love this. This is cool stuff. And the transponder is a secondary radar. And the difference on this really matters. So instead of waiting for an echo, the ground antenna sends an interrogation, and the box in the airplane answers with a coded reply. Code is four digits, each zero through seven, so about 4,000 possibilities. And the controller assigns one. Later, a second node adds altitude. Now the dot has a tag that follows it across the screen. And you can see this in the TV shows and movies and stuff. You can see the little, you know, transponder codes. It's really cool.
Renee: And the code is called a squawk. And no human being has ever said that with a straight face. I don't think I did either. The 7700 is an emergency. 7600 means I've lost my radio. 7500 means someone's taken the airplane. I didn't know that. The pilot dials it in and the whole sky knows without anyone saying a word on the radio what's going on. That is the best piece of design in aviation. And I won't be taking any questions on that. It's just the best design.
Marc: All right.
Renee: You think, you know, what do they call them? Aerolons? You know, the things that keep the thing in the air, like, no, not as important as this.
Marc: The squawk. The squawk's more important. Yeah. Not fly-by-wire flight control systems. No.
Renee: Yeah. That seems crazy. No. No. So a spider lays some eggs in one of those pitot tubes. Next thing you know, you're crashing.
Marc: No. No.
Renee: No, none of that. No. This is important.
Marc: All right. Okay. So I'm going to ask you a question anyways. All right. So the airplane has to cooperate. Transponder only answers because somebody turns it on. And a lot of that comes, you know, later depends on that. So keep it in your pocket.
Renee: Pockets open. So now, December of 1967, a prototype IBM system goes live at the Jacksonville Center.
Marc: It's always IBM. It's always IBM. You know, right?
Renee: They get their hands in everything. And I want to note that this is the year before I was born, so that computer and I are roughly the same vintage. And one of us has been kept in much better shape, and it's not me. They've been taking good care of that IBM over there at Jackson Center. They've been taking good care of it.
Marc: It's the first piece of what the faa called nas nas on route stage a and the machine that rolled out from was the ibm 9020 which is a cluster of system 360 mainframes zed they're wired together so that okay okay we're talking about 1967 people and they're wired together so that when one fails, another took over. Okay.
Renee: What? Do you mean back in 1967, we understood physical failover? What?
Marc: Yeah. What? Yeah, I know. It takes, yeah. So the 90-20, it takes in radar and flight plans, tracks every flight it knows about, and does something new. It looks ahead. It projects two or three minutes into the future and flags two airplanes that are going to end up too close.
Renee: And listen to what it didn't do. It did not get rid of the strips. It printed them. The most important output of a mainframe in 1967 was a piece of paper with a slot to slide into a rack.
Marc: Yeah. All right. So this might be a little unpopular, but, you know, which is the piece of paper can't tell you two airplanes are converging.
Renee: No, but it also can't crash. It never needs an update. No one can hack it. It's not like those pieces of paper. I mean, like, third line of defense, I'd say that's a pretty good deal, right? The controller holds the strip, writes on it, hands it to the next controller, and that handoff is a commitment. Someone touched this airplane, and someone is now on the hook for it. People have been trying to replace that with software for decades, and controllers keep asking for the paperback, right? To me, that says the strip was doing the job the software wasn't doing. They didn't feel like it was safe enough just to trust the software. They were going to keep those strips of paper just in case.
Marc: I think that's a fair claim, right? I think it's right in kind of the narrow sense. The screen shows you where the airplane is. A strip shows you what you promised it, right, where it's going. Those are different jobs, and the screens only ever did the first one. The screens don't predict where that plane is.
Renee: Thank you. I'll put that on a mug. And in 1987, the centers got the host computers, IBM 3083s, and noticed what they replaced. The computers. The screens stayed. The strips stay. The sky stays open. Because the one sentence no one in that building could ever say was that, we're going to close down for a couple days to do this upgrade. Can't do it.
Marc: Hey, we're just going to shut down. All right. So let's wind forward a little bit. June 30th, 1956. A TWA Super Constellation. The Connie. And a United DC-7. Yeah.
Renee: I have a photo of me and a Connie because that Connie is at JFK now. Oh, yeah. At the JetBlue terminal. Yeah. That's good.
Marc: People should go back and listen to the commercial airline.
Renee: Oh, the airline one.
Marc: Yeah. Yeah, that episode. So, all right. So, TWA Super Constellation and a United DC-7 leave Los Angeles within minutes of each other, both heading east, and they meet over the Grand Canyon. 128 people die. And at that point, it's the worst airline disaster in American history.
Renee: And the detail that everybody skips is that nobody was controlling either one of them. They were up high in airspace with no controller at all. And the rule was see and avoid, which means you look out the window and hope. Oh, it's like plug and pray. In the five years before the crash, there had been 65, 6-5, midair collisions. 65.
Marc: That's nuts. In a five-year span.
Renee: Right? Like, they were 12 at a time falling out of the sky. One a month for years. And we thought that was the cost of flight? Like, that's crazy.
Marc: Yeah, yeah. And that crash, that 128, that's the crash that finally moves Congress. The Federal Aviation Act of 1958 creates the FAA. And the big idea is positive control, meaning above a certain altitude, you're talking to a controller whether you like it or not.
Renee: Don't say Congress moved fast, because it didn't. It took two years to pass the law, and the radar coverage to back it up didn't arrive until the middle of the 60s. And that's almost a decade of knowing exactly what the problem is and flying through it anyway. Yeah.
Marc: Yeah. I mean, you know, fair, right? I'll concede the decade. I'd add that this is the first time the system gets rebuilt in response to a body count, though, and it's a template for the next 50 years.
Renee: And meanwhile, the airplanes are getting faster. The 707 starts carrying passengers in 1958, the same year as the law. And sea and avoid was already a coin flip with the prop planes. And now things are going to travel at 350, 400 miles an hour. That's scary.
Marc: Yeah. So now you think about the math on this, right? So two jets head on could be closing at 1,000 miles an hour. So if they're both doing about 500 miles an hour, the closing speed is 1,000 miles an hour, which is nearly 1,500 feet a second. So if a pilot needs 10 seconds to spot another airplane, recognize it, decide and maneuver, that's close to three miles. And a small airplane is hard to see at three miles. Human eyes weren't scaling with the engines.
Renee: And it didn't end the problem. December 16th, 1960, a United jet and a TWA prop liner collide over New York and 134 people died with wreckage coming down on Staten Island and in a Brooklyn neighborhood. Radar existed by then. The problem had changed shape instead of going away.
Marc: Yeah, so that's the pattern, right? The system gets rebuilt around the most recent failure and the rebuild is always a little behind the next one.
Renee: I don't know if I would call this my favorite fix, but I'm going to skip ahead to my favorite fix in the whole history of all of this, because it costs nothing. It's a word. It's a word. In March 27, 1977, Tenerife and the Canary Islands, to 747 KLM and Pan Ams on a foggy runway where no one can see anybody. 583 people die. It's the worst crash in aviation history. History comes down to a conversation in which the crew heard what they expected to hear.
Marc: Yeah, the KLM crew had the route clearance, and the tower said something that sounded close enough to take off clearance. And the radio interference stepped on the end of the transmission that was meant to say, wait. So the fix was phrasing. The word takeoff is now reserved for the actual clearance and nothing else. And every other time the word is departure.
Renee: It's the cheapest safety upgrade ever shipped in the history of mankind, right? Remember that when someone pitches you a billion-dollar solution, be like, I don't know, should we just use a different word? It's actually a thing. And, you know, here's what I would also say about that one. This is one of those disasters where the KLM is going to take off, but the Pan Am is still on the runway. And so the Pan Am is coming in. The KLM wants to take off. The KLM gets just high enough that it eclipsed the top of the Pan Am. And the Pan Am's a jumbo jet. So it has that hump over the top of it where first class is. And that pilot survived. The pilot of the Pan Am flight survived. And he said, instinctively, he just ducked. He saw the plane coming. And he said, I just ducked. And when I sat back up, the whole deck above me was gone. He's like, it was fresh air. It was just gone. The whole top of the plane was gone. And so and he said all of first class was gone and and, and you and this was two people who just misunderstood because one thought he had clearance to cross the runway and the other one thought he had clearance to take off, right yeah crazy oh and that klm pilot was a decorated pilot with 40 some years experience he trained other pilots he did all their training videos like he was the guy at klm he was the guy.
Marc: So, that's the cheapest safety upgrade. But then let me pitch you one that did cost some money because this one's, you know, one of my favorites. So, September 25th, 1978, San Diego, which, okay, not a great airport, people. Not a great airport.
Renee: No, crazy airport. Crazy airport. Downtown, and, like, you almost have to, like, fall in a Harrier into it.
Marc: You have to, like, you have to fly through downtown. You look out the left, you see a building. You look out the right, there's, like, the San Diego Zoo. It's yeah no it's not yeah
Renee: It's crazy oh.
Marc: Yeah i used to fly in there all the time that
Renee: Is colorado springs dude colorado springs is terrifying too i'm just gonna put that out there.
Marc: Burbank you could do burbank on the the there's a parking garage across the street from the the landing strip and it's the tires clear it by like 12 feet or something anyways all right So San Diego, September 25th, 1978, a PSA 727 and a Cessna 172 collide. 144 people die, and the controller's conflict alert did fire, but it fired too late to matter. So the cockpit crew, they were relying on their own eyes to keep the Cessna in sight, and they lost it.
Renee: So the ground system saw it, but the airplane didn't.
Marc: Right. Yeah, right. And the response is to put a safety net in the airplane. That starts a beacon-based collision avoidance system and becomes TCAS, T-C-A-S. Each airplane listens to the transponders around it, works out whether anybody is on a collision course, and first gives the crew a traffic advisory somewhere around 40 seconds before a projected collision. If it gets worse, there's a resolution advisory roughly 25 seconds out, which is a voice telling one airplane to climb and the other to descend. The two systems negotiate with each other so they don't both pick the same direction.
Renee: That was my question. Like you both say, go up. And you're like, oh, no. We both went up. Yeah. Oh, damn it.
Marc: There's a priority thing. It's not like it's, yeah, it's not like it's, you know, like two chatbots talking to each other. Now it's a position in the air, and it's determined.
Renee: And it only works if the other airplane has its transponder on. So the safety of your airplane depends on the other airplanes having joined the group chat, which is something the Malaysia Airlines M380, like, that flight, like, they get up into cruising altitude. They talk to the last tower, and they turn off the transponder. And, like, at that point, it's gone. Like, nobody knows where it is. They can project where they thought it was going to go. They know what the flight plan was, but we still don't know where it is. Yeah. It's crazy. Is that sounds?
Marc: So, yeah, but that transponder system, you know, it has to be cooperative. And that system is brilliant, right? But it has that cooperation requirement that's built into it.
Renee: August 1981. I'm 13. It's a Monday and I'm in my mother's kitchen and the radio was always on. And my grandmother has a flight that Thursday and the radio says that 13,000 air traffic controllers have blocked off the job. You don't remember that? Do you remember that?
Marc: But I mean, I was like five, six, something like that, you know, seven years old, something like that. So I don't really remember it. But I do remember learning about it in history class. That's that's called that's called PATCO, the Comptroller's Union. They went out August 3rd demanding better pay and a shorter work week, along with better working conditions.
Renee: And the president gave them 48 hours to come back to work. And on Wednesday, August 5th, he fired the ones who hadn't. Something over 11,000 people, the union was decertified that October.
Marc: Yeah. Thank you, Ronald Reagan. Yeah.
Renee: Yeah, exactly.
Marc: And the planes did keep flying. Supervisors, military controllers, and the ones who'd crossed the picket line ran it. And the FAA cut the schedules and held traffic on the ground to keep the sky thin enough to manage it.
Renee: They're still doing that, by the way. You'll still get ground stays in the United States to account for the fact that there's not enough controllers or they all had to go on break at the same time. Like Palm Springs will get all kinds of like around 230 because we just don't have enough of them. We get ground stops, which is crazy. We have a marine base right up the road. These guys can't come down and get a job on the weekends or something like what? I don't understand. I don't get it. I think your government should let you work two jobs, like just like everybody else. I don't know. New controller takes years to certify. President Clinton lifted the rehiring ban in 1993. And by 2006, only about 850 of the fire controllers had actually come back. A new union, NATCA, N-A-T-C-A, formed in 1987. The pipeline is the thing that has never fully healed. I can tell you this really quick story, though. So I'm going to a conference, a security conference up in Santa Clara, and I drove. And anytime there's a big a black bear diner i have to stop at the black bear diner, and there's one on the way there's one on the way and it's like i'm at the black bear diner and behind me is this guy he just he gets out of his big truck big truck right and he he you know just he's standing behind me and she's like table. For one and i said yeah unless this guy wants to have lunch with me, it's because who would want to have lunch she puts me and then she comes back three seconds later and says he said i'll eat with you if you want to eat with him he does he like he would and so this poor guy he's probably like 30. He sits down across from me and the look on his face is abject horror. Like he doesn't know what he's going to say next. He doesn't know. And I look at him and I'm like, hi, I'm Renee. I'm on my way to Santa Clara for a conference. I'm just telling him like, who are you? And he tells me he was, you know, he's Jose and he was in the service so he could get his, you know, his citizenship. And he drives, he did long, long trucking. Now he does short trucking. And we're talking and he's like, you know what I really want to do with my life? Okay. I just met this guy. I just met him. I just met him at the Black Bear Diary. I said, what? He's like, I really want to be an air traffic controller. And I'm like, you said what now? And he's like, I really want, I really think I would like, he's like, the more stress there is, like, the better I am, the more honed I get. And I'm like, oh, you may have ADD. And I'm like, okay. So like, and I'm just staring at him. Like, do you know we're short like 20,000 of them? Like, just go to the FAA website. And it says, hire me as a air traffic controller and get it running. He's like, what does that mean? I'm like, you served. They've already done background checks on you. Like, that's where it starts. And I'm like, and then they just put you through the training. And he's like, well, don't I have to go to school for that? Because I got the GI bill. I'm like, no, they pay you to train. If you really, please, please, if you really want to do this, go do this. This is ridiculous. Like, you want a job that no one else wants, and you think you'd be really good at it. Why don't we go figure that out, right? And so, yeah. So when I think about, like, they lifted it, and they're still short hiring people and all that stuff. Yes. Yes. But I met a kid who, like, who was driving trucks for a living, who served this country, and is now a citizen, who sat down across from lunch to me as a stranger and said, I would love to be an air traffic controller. I wonder how I could do that.
Marc: Well, there you go. Yeah. Yeah. You can swap a computer in a weekend. You can't manufacture a controller who's been through the pipeline. It takes years. So hopefully, Jose, wherever you are, you know, you did that.
Renee: Dude, do it, man. Do it. Yeah.
Marc: It's, you know, it supplies whoever started training two years ago, three years ago, four, I mean, years.
Renee: So in case you're wondering, my grandmother who flew that Thursday was delayed three hours and talked about it for the rest of her life, which is the best review a system can get, really. Yeah.
Marc: I can see. Like, I never met your grandmother, but I can see your mom, you know, or you complaining about it for the, like, so that's, like, that's how I can see it, you know. When Renee's 80 years old and we're still doing this podcast, she'll be like, ah, that one time when.
Renee: I remember that one time that happened to me. This is why I still want to be an orca. Yes. Yes, yes.
Marc: Yeah, all right. So, you know, yeah, she probably wasn't wrong. That was a personal affront. For somebody on the plane, the entire product was a delay that ended.
Renee: That was too sweet.
Marc: So every time somebody has tried to modernize the system by building a new one next to the old one, then cutting over, it's gone badly. And a big example is the Advanced Automation System, AAS.
Renee: Oh, all right. Give me the numbers first. What happened here? I was going to push back on every time, but I want to hear the numbers first.
Marc: All right. Yeah. So in 1983, the FAA estimated AAS at two and a half billion dollars, finished by 1996. Oh, boy. 83 to 96, 13 years.
Renee: That's, again, it should take that long.
Marc: Go ahead. Yeah, it should, I guess. IBM wins the development contract in 88. So it took five years just to get somebody to win the contract after a four-year design competition.
Renee: Design. Oh, my gosh.
Marc: And by then, the estimate is $4.8 billion in 1998. Yeah. So by 1994, the estimate has climbed to as much as $7.6 billion, with the completion as late as 2003. Okay, that's 20 years after they started the process. And about $2.6 billion had already been spent. Oh my gosh. And the plan on paper was to fold more than 200 en route and terminal facilities into 23 locations.
Renee: Okay, so it's tripled. So the technology isn't what killed it. The ambition was new controller workstations, new computers, new tower systems, new buildings all at once in a system that can't say we're closed for renovations. You can't do it to a patient. OK, OK, you can't operate on a patient who's running a marathon. You can't build a plane that's trying to fly. Like, what makes you think you can do this?
Marc: Yeah, I know. I know. It's right there, right? New computers, new tower systems, new buildings, new everything all at once. Like, no. No. Like, no. Yeah, the government's own reviewers said pretty much the same thing. Cost, schedule, technical problems, and the FAA restructured the whole thing in 1994. What survived was the least glamorous slice of the whole process. The new controller display system. DSR. With Lockheed Martin as the prime contractor. Man, those guys can't. They can't deliver anything for under 100 billion bucks. I mean.
Renee: Oh, no. Why would they bother with that RFP?
Marc: Yeah, exactly. Yeah. And the terminal piece was canceled and replaced by a smaller program called STARS. IBM sold its federal systems business to Laurel that same year, which is, you know, kind of a hell of a footnote on that.
Renee: Meanwhile, the host computers from the late 80s just kept on working. The failed future sat in a filing cabinet and the old machines carried the country. I don't want to oversell that because old and reliable isn't the same as good. But you can see why the people on the floor distrust change. They just don't have any of it. Like, I wouldn't trust it either. If it has never changed, but it has always worked, don't change it.
Marc: But I mean, especially in something where if a plane falls out of the sky, people die. So I think distrust of the new is, you know, it is partly rational. If your old system has a known failure pattern and you know how to work around it, the new one has failures nobody has met yet. Like, yeah, distrust is a valid and rational thing.
Renee: Yeah, you know what? And I agree. And that's the best argument for using those strips on blocks in this whole episode. Let me tell you about the day when the old way mattered most, because I think about it more than any other date in here. And that's September 11th, 2001.
Marc: All right, lay it out.
Renee: A man named Ben Sliney was on his very first day at the FAA's National Operations Manager in the command center in Virginia. Mid-morning after the second tower was hit, he orders the first nationwide ground stop in history, and everything in the air is to land at the nearest safe airport. About 4,500 planes, 700 of them land within the first four minutes, 2,800 in the first hour, more than 4,500 inside of three hours.
Marc: Yeah. Wow. Yeah. Think about the mechanics there, right? The controllers had to identify airports for airplanes that had no business being near them, right? Sequence them, then vector them and talk to crews, you know, they're scared, while also keeping track of the ones that had stopped answering the radio. Military was scrambling aircraft at the same time. There was a plan for exactly this. It's called SCOTANA, the security control of air traffic and air navigation aids. And I bet most of the people in that room had never expected to use it. Like that's the type of thing you develop and it's like you're never going to do that.
Renee: And you might do tabletop exercises around it knowing that you're never going to. Yeah. Sliney said later that he wondered in the middle of it whether he even had the authority to ground the whole country and decided the question could wait. I actually really like that. I'm the kind of person that begs for forgiveness rather than ask for permission. So yeah, I get that. I love that. So nothing about that morning needed new technology. They needed people who knew how to land an airplane on short notice and one person willing to make the call before checking his authority. Canada did the same thing at the same time, Operation Yellow Ribbon, and put more than 200 airplanes down in places like Gander and Halifax, with 40 planes and about 8,000 passengers landing in Halifax alone.
Marc: Yeah. I need... Halifax.
Renee: Right? He's like, I bet that airport doesn't, it has like one bathroom in it. You have to share it with everybody. Like, I don't know what Halifax airport's like, but in my mind, like, I conjure up a tiny airport, like, you know, like Fairbanks or something. Yeah.
Marc: Okay. All right. So we come back to something that we talked about earlier, right? On several of the hijacked airplanes, the transponder was switched off or changed. So controllers were looking at the anonymous kind of dot again and had to work out, you know, who's, who that was, who's the dot. Everything that talks to the sky assumes the airplanes want to be found. Underneath that, the system could do what it did because it's simple in the right places. A controller talks to a pilot and tells them where to go. You can strip nearly everything else away, and that part of it still works.
Renee: And that's why I get nervous when somebody says brand new. I want brand new radar. I really do. I also want that conversation between a controller and a pilot to survive any upgrade that anybody ever did, ever. And, you know, we talked about this for a second, that we've, like, human in the loop. In this case, like, the human is the loop, and the technology should enhance it.
Marc: Yeah. You know, there's a, gosh, I can't remember the name of the system, but, you know, it's in the States in particular. They haven't done this, but there's a different type of, it's basically GPS that they could turn on. Because radar, you know, does have some level of lag. And with an active real-time GPS system, you could land more and take more off because the gap, you know, you could manage a smaller gap I'm kind of okay with a larger gap, I think.
Renee: They're trying to get them down to like one minute. They're landing a plane every minute.
Marc: Yeah, I mean, it's crazy.
Renee: I mean, right now it's something like every three minutes or every five minutes. Like right now it's really fast anyway, but they're trying to get it. It's really fast. They're trying to close that gap.
Marc: Anyways, all right. So let me give you the upgrade that actually worked here because, you know, it did. Nobody throws a parade about upgrades that work. ADS-B, Automatic Dependent Surveillance Broadcast. The airplane figures out where it is from GPS and broadcasts its position, speed, and identity a couple times a second. So this is what I was just talking about. The FAA's mandate took it into effect in January 1, 2020.
Renee: It's the same thing as a transponder with one big change. The transponder waits to be asked. The ADS-B just announces itself.
Marc: Yeah. Yeah. So the practical difference is pretty big here, right? So radar can't see over the Gulf or across big stretches of Alaska, but a GPS broadcast can be received by a cheap ground station. And where radar needs something like five miles of separation, a better position source can let the controllers pack the sky tighter and still be safe. We hope.
Renee: Okay. So here's my complaint about ADS-B. It's a modernization complaint that I have, okay? It has two radio links, big airplanes, and anything above 18,000 feet uses 1090 megahertz, which is the frequency the transponders and the TCAS already crowd. And smaller airplanes flying lower use a second link at 978. So the FAA built a system where two groups of airplanes speak different dialects that sounds, And I don't know, it sounds like a mistake, right? It sounds like they're not all talking to each other. Big planes are talking on one band, while the smaller planes that are at lower altitude are talking on another band. And it just doesn't seem like it's connected at all.
Marc: Well, it was deliberate, though. The 1090 band was getting crowded, and the 978 link was meant to relieve that for general aviation. Two groups can still see each other because the ground stations re-broadcast, taking what they hear on one link and repeating it on the other. There's a translator in the middle I'd say most of the realistic thing is in this whole episode when you can't make everybody switch on the same day you build a translator and you live with it
Renee: Okay, that's a good argument. I noticed that the ADS-B still sits on top of everything else. It didn't replace radar. It just joined the pile. We put ADS-B in and all the old systems stayed in because we can't take downtime ever in aviation.
Marc: Yeah, I mean, it does seem a little like, okay, if one band in 1090 was getting crowded and then you're just going to rebroadcast into 1090 everything that happens in 978 anyways, or everything that's in 1090, you're going to rebroadcast down to 9782.
Renee: Yeah, right? Right?
Marc: It's like, okay. Yeah, anyways. All right. So, yeah. So take us to Newark and floppy disks, Renee.
Renee: So in the spring of 2025, you guys have heard me complain about this. So just let me do the facts. Radar data from Newark's airspace travels to a facility in Philadelphia. And one of the connections in the middle is an old copper wire. An old copper wire. And it failed.
Marc: Does does copper wire actually fail like i just want
Renee: To i guess i guess if some.
Marc: Like a squirrel homeless people
Renee: Homeless people come in to take the copper on the building while you're in there i don't know i don't know like no copper wire doesn't fail copper pipes don't fail this doesn't actually happen could it have been backhoe sure could somebody have unplugged it yeah, oh no layer one of the osi is it plugged in like it doesn't fail for about a minute and a half give or take, i'll say this can it have latency because of oversaturation sure yeah sure it can have that right but that's not a failure and it's not the cable's fault, for about a minute and a half give or take controllers lost the screens and the contact the airplanes had their own collision avoidance and everyone landed, but nobody who works in that room is going to forget that minute.
Marc: Yeah. Yeah. Within weeks, the transportation department stood up and announced a plan for a brand new air traffic control system by the end of 2028. With the press conference mentioning floppy disks and Windows 95 as examples of what's out there. Okay. Okay. So, all right. Spring of 25, We've got some network layer issue of some sort. And then they come and do a press conference and say, three years later, we'll have something in place. When the previous system took over 20 years.
Renee: There you go. There you go. Okay. And don't laugh because this is the part I find interesting. Congress put $12.5 billion toward it in July 2025 law. Dumb bill and in and in december the faa named a company called patron as the prime paraton oh my god what can't people just name it like phil's stuff i'd be okay with that yeah, As the prime integrator whose job it is to make 50-some vendors work fit together. Oh, oh, I see. So we're going to take all these little apps that do things like move data from NOAA to a server so that the scope can understand what it's looking at when it's looking at weather dots, right? And it's, I mean, it's something like 200 data feeds, right? Like it's something like 200 data feeds. By January, there were radar contracts to RTX and Indra and about 600 new radars, and the old copper wiring is being swapped out. And officials say they'll need roughly $20 billion more to finish it, with the whole thing expected to be a total of about 30. Hey, I'm just wondering, what did it cost for Quest to put in all that fiber all over the country?
Marc: Yeah, not that.
Renee: Not that much. And this is just government agencies. And there's only 20-some of them. It can't be that much. It can't be that much. Are we running copper to the desktops? Are we running fiber to the desktops? I don't know. Who do you think you are, Sony? Okay, okay, okay, all right.
Marc: Okay, all right, so this, but, okay, let's be, you know, somewhat fair, right? The scale is real, right?
Renee: If you have to be.
Marc: I guess. I guess. All right, scale's real. Some of it's already underway. 435 towers were slated for a new information display system. 200 airports get new surface surveillance so controllers can see what's moving on the ground. Have you seen some of those? There's some really cool stuff that they are doing with that. And the wire work is kind of boring, you know, but it really matters. The contract is also reportedly performance-based with Peraton's profit tied to outcomes. What I'd keep an eye on there is maybe the integration report surfaced in August that there was tension between the FAA and the contractor. I don't know. who's right or wrong about that, but integration exactly where AIS fell apart.
Renee: And as Marguerite, too, and I'll say it carefully because I'm not predicting, the timeline of under three years to rebuild a thing that's been rebuilt in flight since 1929 is.
Marc: Yeah, you know, I'm going to be kind.
Renee: It's ambitious. It's ambitious. You know, some of the pieces will land on time and some won't. The question is whether anybody keeps the strips in the rack until they do. And you know what? And I say yes, please. Keep the strips as long as you feel happy about it. Air traffic controllers, keep them forever.
Marc: Last fall's shutdown read 43 days, and controllers worked without pay. The FAA ended up ordering flight cuts at 40 major airports, about 6%, and over 10,000 flights were canceled in the first week of it. At one point, the FAA said nearly half of its major facilities were short-staffed, and nearly 90% of controllers were out at the New York area facilities. The FAA's administrator told the Senate it lost four to five hundred controller trainees during that stretch.
Renee: And before the shutdown even started, the FAA was about thirty five hundred controllers short of its targets, with a lot of people already working mandatory overtime in six days a week. So already they're overworked and they're still short thirty six hundred people. So all of it comes back to the kitchen table in 1981. You can buy radar. You can buy fiber. You cannot buy somebody with five years of training, and the clock on that starts the day you hire them. Controllers have to retire at 56, so there's no waiting out a shortage. If you can hear my voice and you've always wanted to do that job, go. Go, go, go.
Marc: Unless you're 56.
Renee: Oh, yeah, unless you're my contemporary, then. Too late for you.
Marc: Too late.
Renee: A young person who can't find a job, you can find a job with your government.
Marc: Yeah. Yeah, I'm not sure, like, you know, my anxiety-ridden children would do so well with that one.
Renee: No, I wouldn't either. But I can't be an emergency room doctor. I can't be a veterinarian. There's all kinds of crap I can't do because I can't handle the stress.
Marc: Yeah, yeah. All right, so it's the PATCO lesson again, right? And I think this is the headline, right, of the whole modernization story. A new system plus too few people to run the old system while you install it is the hard. It's like, this is a huge part of this problem.
Renee: All right, so I'll tell you how to do it, and you can tell me if I'm wrong. You run the new thing beside the old thing in a shadow. Same inputs. You compare the answers until they agree for a long time. Then you move one little corner of the sky open, and you keep the old system warm behind it. I think that's how you do it. This isn't one of these. I mean, but let's go back to the idea that this project that was supposed to go until 2028 and was outcome based and was supposed to be done on time meeting milestones and bonuses, I guess, happened if milestones were met or were done early, just got done two years early. Something we've been fixing since 1929 only took two months to fix? No.
Marc: Yeah. Yeah, I don't think so.
Renee: That's not possible. No. It's not magic. It's software.
Marc: AI will do it for you, though.
Renee: Oh.
Marc: I wonder. Okay, so before I get back to this.
Renee: But that's my thing, right? You're going to run them together. You're going to audit them out, make sure they're doing what they're supposed to be. I mean, this goes back to the way we used to deploy old ERP systems.
Marc: Right? Yeah, of course. This is how we used to do this stuff. Yeah, that's the right thing to do.
Renee: This is not agile and iterative process improvement. It's air traffic control. There is no iterative process improvement.
Marc: There's no fail fast?
Renee: No, no, no, no. There's no go fast, break things. No, I meant the thing you're trying to break. Stop it.
Marc: Yeah, don't fail fast. I think like when I worked at one of the, that soft, the security software vendor, like we were talking about our, our terms of agreement or license agreement, you know, that you have to get. And if you read some software license agreements, sometimes you should be, you should be scared because sometimes they'll put things in there that, you know, like this software is not meant for running things like nuclear power facilities or air traffic control systems or, you know, whatever. Wow. Yeah, it's really interesting.
Renee: I've been targeting operations.
Marc: Yeah, yeah. So I wonder if, you know, my license agreement for Claude, you know, or ChatGPT says, this is not meant to produce systems for air traffic control or nuclear power facilities.
Renee: Yes.
Marc: Yes. I'm going to have to read the license agreement now. Anyways, but yeah, you're right. The textbook version, right? One little sector at a time, you know, with a way back, you can reverse it in seconds. And, you know, the hard part is that, you know, is the phrase agree for a long time. The old system has bugs the controllers have learned to work around. So when the new one doesn't have the bug, the answers disagree. And the new one is the one that looks wrong. You know, sometimes you end up reproducing the old quirks on purpose.
Renee: Yeah, that's the muscle again. So the training matters as much as the code, right? 113 towers are getting new simulation systems. And I'm glad because a controller who's practiced on the new screen for six months is a lot safer than one who sees it for the first time with a 747 on final approach, right? So maybe that's the answer. Maybe the new system is really just a training platform until it's not. Until every one of those people are trained enough on that new thing that when you say to them, next Friday, you guys, we're not going back. We're on the new platform. They're all like, all right, we're good, right? But that's what it would take. It would take a year of them using it as a game, and not wrecking anything and helping you fix it. But you need years of that. It's not days of that. It's not weeks of that. It's years of that before it would be ready to go. And to say that you did it in eight weeks so you get a bonus, I'm sorry. There is no one in corporate America IT who has been working for the last 20 years who would have thought any of that was a good idea and it would have turned out okay. Not one of us. Not one of us who do this for a living for real would have thought that was okay.
Marc: You also need rollback. It doesn't involve, you know, some sort of meeting, right? If a sector says that new screen is wrong, the old one has to come back in seconds. You know, it has to still be running, which is why this costs so much. You're paying for two systems that are running at the same time for years. And nobody likes that, you know, in the budget.
Renee: Well, because they want to treat, They want to treat modernization like some kind of like lifecycle upgrade or patch or something when really, you know, what it continues to be to this day, is a huge undertaking of organizational change, technology change. Like it's a lot of stuff, right? Like right now, people will tell me I don't want to adopt AI just yet because we are not good at change. SAS ruined that muscle memory for us. Right. SAS was easy. It looked like everything else. People updated it behind your back. You didn't do anything with it anymore. There is no more organizational change around IT. We don't do big rollouts anymore. And now all of a sudden, here's a big rollout. Here's one where you have to go back to like, this is SAP. Like, let's do this. It's going to take two years. Like, let's do it. Like, let's meet, like, let's put real milestones, POAMs and all that stuff. Like, let's do it. No. Like, the one chance that they would have got away with that to say, of course, we have to take our time. This is the air traffic control system is the one place they didn't bother doing it. That's crazy to me. And you know what? That's the price of not having a Sunday window, right? Like we can never take it down. And it's where we started. Real-time modernization with the sky running and a short bench. It's hard for that exact reason we said at the beginning. And I'm still not betting against it. I still don't think it's right. I still don't think it's, you know, I mean, they call it a done deal. They launched Smart three days ago. And three days later, everything seized up. Do we have an infrastructure in the FAA and in our aviation system that can handle agentic AI, 200 data streams times 22 centers? Do we have the infrastructure to support that? It does not sound like it. I was thinking about Archie League while we were talking. He lived long enough to see the thing he started as a guy with flags becoming something with computers and satellites. He retired in 1973 as an assistant administrator.
Marc: Wait, he started in what, 1929?
Renee: 29.
Marc: Yeah.
Renee: And he retired in 73 as an assistant administrator at the FAA. And he died in 1986. And the Controllers Union now gives out a medal with his name on it. But I like that the person they chose to honor is the one with the wheelbarrow and the lunch pail. Like, I like that. I like that.
Marc: I like that, too. I like that. I keep, you know, I keep coming back to the sort of Sunday window idea, right? Four hours, you know, you've got some, you know, limited capability. You know, there's always something that's going on that you got to take care of. You don't always get a window, right? Every upgrade in this story happens with somebody on some sort of frequency, right? There's a plane in the air, and you say, we're now going to switch to some other frequency. We're going to change this system for that plane. Yeah, that's tricky.
Renee: And I guess what we actually have is thousands of people handing an airplane from one to the next, and every handoff is someone saying, I've got it. Someone's accountable, someone is responsible, someone knows what to do with it, and it's a someone, right? Like, like, I don't, no machine will ever care about a human being. Again, that MIT stuff, like, taught me that I will run over an old person before I run over a baby any day of the week and twice on Sunday. Like, you don't want that kind of judgment from a machine. I don't want anyone to die. Of course I don't. And if I was the air traffic controller, I wouldn't be like, hey, put all the old people in the front just in case this goes bad, right? Like, I don't know. Like, if it was up to AI, would the Cessna go down or would the 707? Well, the Cessna had less people in it, so maybe it's expendable.
Marc: Yeah. I mean, this is the thing that's really interesting about that, the trolley problem, right, is that it's, like, for a human, it's an impossible decision. Because, of course, you don't want anybody to die. You don't want the cats to get hit. I don't want the kids to die. I don't want the trolley to crash. I don't want, like, whatever sort of situation it is. But you know with ai it's like well you have to make a choice you know where's the choice
Renee: Yes you know i need logic it needs resolved and we can't all live.
Marc: Apparently how do i resolve everybody
Renee: Can be okay.
Marc: Yeah because that's
Renee: Not an outcome we're willing to like like support.
Marc: Yeah it's just yeah anyways all right so so look underneath all this stuff right the host computers and the strips the fibers all of it exists to make the handoff reliable and if you want to know whether modernization is going to work, ask, you know, what it does to the handoff.
Renee: And I want to be fair to the people who sneer at floppy disks because I did. And then I thought about, you know, that time my grandmother flew. Nobody on the plane cared what the computer was. They cared, you know, what the voice, that the voice on the radio was right. Right. Like, like when you're talking air traffic control, like when you're the pilot and you're looking at all that stuff in front of you, who do you trust in that, in that, in that moment? Like air traffic control.
Marc: Yeah. Yeah.
Renee: Air traffic control.
Marc: You know, I think, you know, just given the issues that we've had this last week, it's easy to be cynical about FAA and air traffic control and the systems and all of that stuff. And I, you know, yeah, will there be more problems? I'm sure there will be, right? We've had issues with this, you know, the FAA and this administration. You know, some of the changes will be late and some of it will be fine. You know, but I don't think anybody wants to purposefully, you know, screw things up, right? You know, I just hope that whoever's in charge, they keep asking the controllers what they need on the glass. Because, you know, they were right about how the strips work. You know, they're the users. They're the ones that are keeping people safe. So they need to be the ones that are in the loop on this process.
Renee: And the next time you're on a plane that lands without anything happening, look for the tower on your way to the jetway. There's someone up there who is going to do it again tomorrow. They're going to do it for six days a week. They're going to do it 12 hours a day, and they're not going to make a mistake for years. So thank you to the air traffic controllers in the United States of America and around the world. Thank you for your work.
Marc: All right. All right.
Renee: That's our show. So just quick before we go, just the reason we did this whole show is because we had an incident in the American airspace where apparently something went down. And for the third time ever in the history of aviation in the United States, we've had a ground stoppage where everybody had to land. Only it lasted four hours. And then to get everybody to where they're going, my brother started trying to get to California on Friday the 13th. He's still not here yet. We think he might make it today, but we're not holding our breath. They keep telling him, maybe Wednesday? Like, for four-hour outage, people are being told they have to wait five days to fly for four hours of outage. That's why we wanted to talk about this. We wanted to talk about the technology. We wanted to talk about what's going on. We wanted to talk about modernization. That's what brought us here. So that's our show. I'm Renee.
Marc: I'm Marc.
Renee: And this has been the Nostalgic Nerds Podcast, where two old friends talk about tech history until somebody laughs too hard to finish the sentence. Did we get there that time? I think we got there. You laughed pretty hard at something. All right.
Marc: All right, folks, tell friends. Leave some rating. And the next time your flight lands on schedule, thank somebody you'll never meet.
Renee: And please, if the pilot says there's an ATC issue, be nice. There's a person and possibly a floppy disk involved. So, you know, just cut everybody a break. All right. Thanks, everybody. And, hey, don't forget, subscribe, like, leave a comment, and merch. Don't forget the merch shop, you guys. I always fund a merch shop. I never miss it. I try not to. All right. Thanks, Marc.
Marc: All right. Thanks, everybody.