We Called It Progress

I'm on the job market right now. Which means I've been doing what everybody on the job market does: refreshing LinkedIn like it owes me money.

It's a lot. The layoffs keep coming in waves, and every wave is a fresh page of "open to work" banners. Some of them are people a year out of school who did everything they were told: learned the stack, built the portfolio, and walked in right as the door closed. Some have been at one company longer than I've been in this industry, and are now writing a résumé for the first time since it was a Word doc. A lot of them are people I know. Brilliant people. People I'd hire in a heartbeat, asking if anyone knows anyone who's hiring.

Every listing says "Over 100 applicants" before I've finished reading the title, which is a fun number to see when you're applicant 101. My résumé gets read by software before any human sees it. (Somebody built that, by the way. Probably somebody like me.)

Then I switch over to X, and it's wall-to-wall people shipping. Daily. Designs, design systems, whole tools, one person, a few days. And they're beautiful. Not "good for AI." Genuinely beautiful. Things that a few years ago would've been an entire company, with a seed round and a team, now show up in my feed on a Tuesday, and there's another one by Wednesday.

And everybody's angry about it. The posts practically write themselves. It's trained on our work without asking. It's coming for the junior roles. The executives buying it don't understand what they're replacing. And it's not even that good! Engineers say it has no craft. Writers say it has no voice. Designers say it has no taste, usually in a reply right underneath one of those beautiful designs.

I get it. I'm on the same job boards, and some of that is even true. I've written some of it myself. And I'm not pretending AI caused all of this. Plenty of these layoffs are interest rates and years of overhiring finally coming due. But somewhere around my fortieth application, those posts started to bug me, because I remembered what we'd all spent our careers building.

Somewhere in your work is a person's job

Not a whole one, usually. A slice. The cron job that reconciles invoices so nobody spends Friday afternoon doing it. The onboarding flow you designed so customers never have to talk to anyone. The PRD for the chatbot, with "deflect 40% of support tickets" under success metrics. You never met the people on the other end. You ship the thing, the number goes up, and somewhere a team gets reorganized into a shorter team.

It's not even a secret. It's on the pricing page. "Do the work of a team of five." Every ROI calculator you've ever seen is a headcount calculator in a friendlier font, and somebody in our industry picked that font, wrote that copy, and A/B tested which headcount number converted best.

That's what bugs me about the posts. The pain is real, and I'm in it too. But we've been on the other end of this for thirty years, and the whole time we called it progress.

The medallion

For most of the last century, you couldn't drive a yellow cab in New York without a medallion on the hood. There were a fixed number, the city barely issued new ones, and so they did what scarce things do: they got expensive. By the early 2010s one traded for over a million dollars. Drivers, a lot of them immigrants, borrowed against their futures to own one. It was a retirement plan you bolted to a car.

Then we built an app.

Not the royal we. The whole building. Researchers rode along with drivers. Product wrote the specs. Design made hailing a car feel like nothing, a pin and a button. Engineering built dispatch and payments. Data science tuned surge. Growth handed out promo codes until switching was a no-brainer. Every one of those was a good job, done well.

Within a few years the medallion was worth a fraction of what people owed on it. Hundreds of owners went bankrupt. In 2018 several drivers killed themselves.

To be fair to the app, the medallion was already a bubble. The Times won a Pulitzer showing how lenders and insiders pumped up the price and steered drivers into loans they could never repay. Then our disruption came along and helped pop it. Nobody cheered for what came after. We cheered for the disruption, and we took the rides, because they were cheaper and honestly the cabs kind of sucked.

Here's the part I can't stop thinking about: I have a medallion too. I'm just now finding out what it's worth.

A job in tech has been a scarce license for my entire career. There were never enough people who could build software, design it, or figure out what it should even be, and the market priced that in. The salaries, the equity, the "we'll relocate you, name the city." Some of that was skill. A lot of it was scarcity. And if you were around for the free-money years, you already know some of it was a bubble too.

Scarcity doesn't care how good you are. It only cares how many other people, or things, can do it. The medallion drivers were excellent at driving. My friends on LinkedIn are excellent at what they do. That isn't the variable.

Layers all the way down

Uber wasn't built from scratch. It was built on top of everything else.

At the bottom, GPS. Built by the military, opened to civilians after the Soviets shot down a stray Korean airliner in 1983, and deliberately fuzzed for civilians until May 2000, when the government flipped a switch and consumer GPS got roughly ten times more accurate overnight. Your taxes paid for the satellites. Nobody sends you a royalty check.

On top of that, phones with GPS chips. Then an app store. Then Google Maps. Then Twilio for the texts, somebody's payments, somebody's cloud. When Uber went public, it disclosed paying Google about $58 million for maps over three years. The company that ate the taxi industry couldn't find its own riders without renting somebody else's map.

That's not a knock on Uber. That's the whole game. You stand on a stack of other people's work, add one layer, and the people one layer down have a very bad year. Sometimes the new layer wipes out a profession. Sometimes it makes it wildly more productive. Sometimes it just changes who collects the check. The people living through it rarely get to choose which.

Google Maps had its own victims. In October 2009 Google made turn-by-turn navigation free on Android, and Garmin fell 17% that day. TomTom fell more than 20%. Nokia had agreed to pay $8.1 billion for map data two years earlier. Then Google gave it all away, because to Google, the thing those companies sold was a feature.

Somewhere there's still a TomTom suction-cupped to a windshield, holding on for dear life. Pour one out for it, and for the glovebox full of gas station maps.

Everybody gets eaten, especially us

Being the disruptor doesn't make you safe. It just makes you next.

Kodak invented the digital camera in 1975: eight pounds, a hundred by a hundred pixels, saved to cassette tape. Kodak filed for bankruptcy in January 2012. Three months later Facebook bought Instagram for a billion dollars. Instagram had thirteen employees.

And we've been doing it to ourselves the whole time. "Computer" used to be a job title held by a person. Assembly programmers got compilers. Sysadmins got AWS, then DevOps, then "platform engineering," which is DevOps with a better logo. Some of those jobs vanished. Some turned into bigger ones. None of them stayed what they were.

None of it was the weather. It was pitch decks. Somebody wrote "ripe for disruption" on a slide, a room full of people like us nodded, and every discipline in the building went and made it real. "Software is eating the world" wasn't a warning. It was a recruiting pitch, and we signed up.

And a lot of it was wonderful

And honestly? Most of that list isn't a tragedy. A lot of it made ordinary life dramatically better.

You can video call your grandmother on another continent for free, when that used to cost a phone card and a prayer. A kid with a laptop has better access to knowledge than a university did when I was born. Somebody making candles in their garage can sell to the whole planet by Tuesday. People who can't see get their screens read to them.

And it compounds. Free GPS made cheap maps. Cheap maps made rideshare and delivery and a thousand logistics companies. Each layer made the next one cheaper to build, which is how thirteen people built Instagram, and how one person on X ships what used to be a whole company.

Some of those rungs cost somebody a job. The map cost the mapmakers. The candle shop online cost a shop on Main Street. Others created jobs nobody had imagined yet. We decided, correctly I think, that the trade was worth making.

That's exactly the argument now being made about us. Same math. We just used to be the ones holding the calculator.

We thought we were special

Every automation forecast I read for a decade said the same thing. Truck drivers, factories, and cashiers go first. Creative and knowledge work goes last, if ever, because it takes taste and judgment and that ineffable human something. We nodded along, because "creative knowledge worker" was exactly how we saw ourselves.

It went the other way. The machine learned to write, draw, design, and code before it learned to drive a truck across the country. Copywriters got hit before truckers. Illustrators before plumbers. Turns out a surprising amount of what we called uniquely human could be written down as text, and we'd spent twenty years putting all of it online. For free. On purpose.

And no, it doesn't do the whole job. It doesn't have to. Uber didn't make cab drivers worse at driving. It changed what driving was worth. A model doesn't need to be a perfect engineer, or designer, or PM, to change what companies will pay for one. Ask anyone in my LinkedIn feed.

When it was everyone else's turn, we had advice. Retrain. Adapt or die. We told laid-off journalists to "learn to code," and plenty of people meant it as a punchline.

"It can't do what I do"

In the early 1950s Grace Hopper built one of the first compilers, a program that wrote programs. "I had a running compiler and nobody would touch it," she said later. "They carefully told me, computers could only do arithmetic; they could not do programs."

They were wrong about the machine, and even more wrong about what it meant. The compiler didn't end programming. It turned programming into the entire industry we're sitting in, and the people who picked it up early got to build it.

I've written twice this year about where these tools fall short, and I still believe every word. I've also watched them get better while I was busy explaining why they weren't good enough. The limitations matter today. Betting a career on them lasting forever is something else entirely.

The gaps are real, but they're not a moat. They're a to-do list. Nobody switched to Uber because the drivers were better. Nobody ditched their Garmin because Google Maps was better, at first. It was good enough, and it was right there.

So, me

I didn't turn down the project that cut the manual work. I built tools that replaced entire teams. I took the ticket, every time, and I was proud of the numbers. So I don't get to stand on LinkedIn acting like this is the first time anyone's job got cheaper. (I've caught myself wanting to. More than once. This month.) Principles that only switch on when your own salary is in the blast radius aren't principles. They're a negotiating position.

That doesn't mean shut up and be grateful. If your fight is about who captures the gains when work gets cheaper, that's a real fight and I'll stand next to you for it. Just make it for everyone, including the cab drivers we disrupted on the way here.

As for the work changing, I'm doing what we always told everyone else to do, this time without the smirk. Learning the new tools. I'm building at a pace that, for almost my entire career, I'd have called impossible. Not writing code. Building. That isn't giving up on the craft. We program in English now, and the job is making sure it compiles to the right code. Grace Hopper would recognize it. Right now the most valuable part of my job isn't typing. It's knowing what should exist, what shouldn't, and which answers to trust. Maybe that changes too. But it's where I can do the most good today, so that's where my effort's going.

Every layer that eats a job also becomes the floor for something new. The compiler did. So did the phone, and the map, and the cloud. Whatever gets built on this one needs people who understand it well enough to build on it. That could be us, if we stop insisting the stack ends here.

We built this

I love this industry. I love that we take slow, expensive things and make them fast and cheap, and I love who that's helped. I've spent my whole career stacking layers, and I'm not going to pretend I didn't enjoy it.

The machine was always going to get around to us. That was never a secret. It was the roadmap. We wrote it, designed it, prioritized it, and shipped it.

So I'll keep refreshing LinkedIn, sure. But I'm not going to stand in front of this insisting it isn't real, or that we're the one exception in the history of the stack. I'm going to learn it, build on it, and argue for the people it lands on. Us, and everybody it landed on first.

Learn to code, we said.

Well. It did.