What Forty Years of Displacement Looked Like Before It Was for Everyone

Every company in America is wrestling with the same directive right now. Adopt AI, roll it out safely, train the staff, and don’t expose intellectual property. Corporate legal sits in the middle of it, watching for the moment a careless prompt sends a trade secret into someone else’s training data. The pressure is real and the guardrails are necessary.

I watched that mandate land in my own work. Sanctioned tools and approved training were coming, but slowly. Forty years in this marketplace has shaped how I respond to a moment like this. I don’t wait well.

This wasn’t my first encounter with a tool I needed to understand.

1985: PostScript

In 1985 I was the camera operator at the Stat Store in downtown Manhattan, eventually the floor manager. We served the graphic artists, designers, photographers, and models who passed through with materials needing reproduction. The work was analog — film, chemistry, paste-up, halftones.

Then a new paradigm arrived alongside the existing business. The Stat Store added a service called PostScript Design Services, anchored by a high-resolution Linotype PostScript printer. Agencies could now send their desktop-publishing files to us and walk out with high-end printed output. PostScript itself was the radical part. It described type and graphics mathematically rather than as fixed images, which meant a single file could print at any resolution on any compatible device. For an industry built on physical paste-up and photographic reproduction, this was a category change. Steve Jobs thought so too. He went to Adobe and commissioned a screen-rendering version of PostScript called Display PostScript, then made it the imaging engine of his next operating system, NeXTSTEP. PostScript was supposed to be a printer language. He repositioned it as the foundation of a graphical desktop.

One of the partners would not let me near the imaging side. I learned anyway — found the time, found the access, worked it out.

A former Stat Store associate had moved on to design work at Elle Decor. When the magazine needed a desktop publishing analyst, she remembered what I’d been doing on the side, and Hachette Filipacchi hired me. The job confirmed what I’d already taught myself. That was the pattern in its first form: the thing I wanted to learn was kept from me, and I learned it sideways. The learning didn’t just give me a skill — it gave me the next door. Looking back, I had more moxie than I realized at the time.

Forty years on, what started as a necessity-is-the-mother-of-invention drive had matured into a discipline.

The AI mandate was real, but the path through it wasn’t laid out by my company. So I found one. A decade-old Jekyll blog I’d been meaning to modernize sat there waiting. I pointed AI at it and worked through the problems with my own material, on my own time, where the only thing exposed was my own learning curve — two weeks of evenings compressed into two winter-break days. Then I migrated my wife’s design portfolio from Squarespace to Astro the same way. Real projects with real outputs, and no risk of IP exposure for my employer.

Same pattern as 1985. The path I needed wasn’t being offered, so I made one out of what I had access to. The difference this time isn’t the impulse. It’s what’s on the other side of the learning.

The Second Time: Elle Decor

By the second time, I wasn’t sneaking up on a tool. I was running production on it.

Elle Decor was a proof of concept. Hachette Filipacchi wanted to know whether a high-end magazine could be produced on desktop computers, and they were willing to find out by publishing one. I joined as the desktop publishing analyst — the bridge between the editorial and art teams.

We used QuarkXPress on the Mac, and the launch consumed everyone. Nobody was thinking about who this would replace. We were thinking about whether it would work at all.

It worked. Elle Decor overperformed on expectations, and Hachette Filipacchi promoted me to oversee desktop production across the magazine portfolio. That’s when the displacement became visible. The typeset galleys I’d worked alongside — long strips of phototype, hand-corrected, pasted into mechanicals — were being replaced by output from QuarkXPress on a Mac. The trade that had defined commercial publishing for a generation was being absorbed into a workstation.

Some people learned the new tools and kept working. Others didn’t, and they faded.

I hired a former typesetter named Laurie. She was brilliant. She knew what good typesetting looked like, and QuarkXPress was nowhere near as sophisticated as the Atex system she’d come from. She made it work anyway. She used her typesetting craft to push QuarkXPress closer to what it was replacing, and her work showed the difference.

The agent who placed her revealed the darker side. The agent only knew the old paradigm, and she was distraught — emotional about a market she was clearly unprepared for. Two women, same trade, different outcomes.

The mandate I’m living with isn’t bounded the way the desktop publishing wave was. I’ve been thinking about that against a memory from those years. I was old enough at Elle Decor to understand I had a skill set, and the question was how preciously to treat it. Watching Laurie and watching her agent, the answer was clear enough. Not too preciously.

That lesson is what’s serving me now. Desktop publishing displaced typesetting, but it didn’t displace clinical researchers or auto designers. AI isn’t a trade-shaped wave. The boundary lines that used to define which trades were at risk and which weren’t are softening. The path forward looks similar — find the work, learn by doing, don’t get sentimental about what made you valuable yesterday. But the scope is wider now.

The Third Time: The Web

The scope had been widening for a while. The third time I noticed it was the web.

We moved to North Carolina in 1994. Hachette Filipacchi held a director-level promotion in front of me as a reason to stay, but I left anyway. Family came first. Print was still strong that year, and most people would have called it a peak. But Hachette had a proof of concept running on AOL’s NaviSoft web publishing tools — internally, I remember it as Rainmaker — an attempt to publish to both print and web from the same source. The promise was that desktop publishing would evolve into multi-media repurposing, with the same files driving every output. HTML in 1994 wasn’t ready for sophisticated media. Neither were DTP files, which had been engineered for one destination — the printed page. The PoC was premature in one direction and structurally constrained in another. The signal that the web was coming was real. The path everyone predicted to get there wasn’t.

I took a prepress job at a North Carolina print shop. Prepress prepares agency art for printing. Graphic arts, all over again. I held that role for several years and the work was familiar and steady.

Print’s decline came faster than I expected. Small jobs went first — the kind that didn’t justify the overhead of a print shop when a web page could carry the same message. Then larger ones followed. Every print shop was losing customers to the web, not just ours, and the trade I’d just rejoined was contracting around me.

The click moment was quiet. A few of us in prepress built the shop’s online presence together as a side project, and that’s when I perked up at something other. The side project was more fun than our prepress day jobs. That’s the whole insight — not a strategy, not a forecast, just attention going where attention wanted to go.

I got a Master of Science in Information Technology from RIT in 2001, and my print employer paid for part of it. The pattern was the same as Hachette Filipacchi: they had offered a director-level title to hold me in magazine print, while the print shop was funding the credential that pointed past print. The technical horizon had shifted. My goals had shifted with it. Neither employer could reposition fast enough to meet either change. That’s how it goes when external forces move faster than an organization can adapt.

What Made Each Wave Local — and What Changed

What I had seen was niche to my industry. Typesetters fading at Hachette Filipacchi. Print shops contracting around me in North Carolina. Each wave was real but local — bounded by the trade it was reshaping. You could read about it in the trade press if you cared, and most people didn’t have to.

The future is never a straight-line projection. PostScript was a printer language until Steve Jobs commissioned Display PostScript and made it the screen-rendering engine of NeXTSTEP. Desktop publishing was supposed to evolve into multi-media repurposing, and the predicted path stalled — repurposing arrived later through a different stack entirely. The web didn’t follow any line at all. It expanded by answering questions nobody had thought to ask, and human and market creativity bent the technology to fit problems its inventors never anticipated. That’s why it became universal. AI is on that trajectory. It’s becoming all things for all people, but the path it takes there will keep surprising us.

Most current AI guidance — both the practical kind that tells you which tools to learn, and the philosophical kind that tells you what AI means for human work — leans on straight-line projections. It’s not just unhelpful. It teaches you to plan for a future that won’t arrive.

Now it’s in every sector at once. Anthropic’s most powerful model, Mythos, isn’t being released to the public. It’s going to a select group of companies under a security program called Project Glasswing, with meetings between the administration, tech CEOs, and bank executives about what models like this can do. That’s the inflection point. The frontier has gotten capable enough that even releasing it requires a controlled rollout and government-level conversations.

In 2022 the worry was that AI would take our jobs. By 2026 the worry has flipped. We’re worried we can’t scale up the workforce fast enough to meet what AI is asking of us. Companies are mandating adoption while trying to find the people who can actually do the work. The technical horizon is moving faster than any organization can reposition. I watched two of my employers run into that wall years apart, in two different decades. Now every company in the country is running into it at the same time.

AI doesn’t care what you think of AI. Your criticisms might be time-stamped and no longer relevant. AI is scaling. What will you choose to do?

The Same Pattern, Wider Scale

I’ve used the methodology I refined over forty years to bring myself forward. Find the work that engages you. Use your own material. Learn by doing. Don’t get sentimental about what made you valuable yesterday. The Jekyll modernization and Elena’s site migration are the latest applications of an instinct I first acted on in 1985.

The displacement I watched at Elle Decor and at the print shop is happening everywhere now, not as a trade-shaped wave but as something that crosses every trade. Laurie’s adaptation and the agent’s paralysis are the same two outcomes available to anyone with a skill set right now. The choice is the same. The scale isn’t.

As for me, I’ll keep using what I’ve learned to adapt to another technology shift, knowing my cohort is far broader and more diverse than before. At sixty-three, I presume this is the last major adaptation I’ll work through. I also know that’s the kind of straight-line thinking I just spent this post pushing back on. So I’ll hold the presumption loosely, but don’t hold me to it.