Human factors has reinvented itself every decade since Bell Labs, the oldest lessons are the ones pointing the way forward.
User Experience did not begin with Don Norman’s job title at Apple. The work is older than the name, and most of it is older than anyone teaching it.
You can earn a degree in this field and never hear about John Karlin shaping telephones at Bell Labs in the 1940s, the principles and laws we follow in UX Guidelines, or about the decade the web spent learning to organize itself. School hands you tools, methods and critique; it rarely hands you the history that explains why the tools exist.
Years of princples and laws, going back decades.That gap matters, because the field is not a straight march of progress.
It is a stack of eras, each one solving the last and quietly inventing its own blind spot.
There are two famous lines about history, and UX earns both. George Santayana wrote that those who cannot remember the past are condemned to repeat it. The version usually credited to Mark Twain is gentler: history doesn’t repeat itself, but it often rhymes. I like that one better.
This field does the rhyming kind. Each era makes a version of the same mistake in a new accent — it keeps the disposable habit and forgets the durable lesson that arrived with it. One thread runs through all of them, and it is worth naming up front. Economists refer to Jevons paradox: when something gets cheaper to produce, you don’t use less of it, you use far more.
Each time computing got cheaper — the personal computer, the browser, the smartphone, the design system — the work did not contract.
It expanded.
Truth in point — An Apple II in 2026 dollars would have set you back $14,300 ($2,638 original price). You can now pick up a MacBook Air for $1,099, at over a few hundred thousands times faster of the speed.
We went from computers that took up whole rooms to those we can fit in a pocket. It got more accessible, more people did it, and the industry grew.
Hold onto that, because the AI era is the same move at a larger scale, and the panic mistakes it for the opposite.
One idea outlasts every era, and it sets the bar. In a 2003 New York Times profile of the iPod, Steve Jobs pushed back on the idea that design is painting the house: design, he said, “is how it works,” not how it looks.
That is the durable lesson sitting under all five eras. Each one either remembers it or quietly trades it back for the surface.
So here‘s user experience in five eras, with rough dates, what each got right, and what each left you holding that you should put down. This is not nostalgia. Other histories slice the same decades differently — by interface paradigm, by platform, by tool — and those cuts are just as valid; this one is organized around the lesson each era kept and the habit it should have dropped.
The era you are in rewards the people who can tell the durable lessons from the expired ones.
Bell Labs designed around the human hand before anyone called it experience.The Human Factors Era (1940s–1988)
Before the screen, there was the system. Karlin’s group at Bell Labs shaped the touchtone keypad layout you still use today, working from how human hands and memory behave rather than from what was easy to build.
The three by three grid — What won and still lives on today.R. L. Deininger’s 1960 study tested circles, arcs, and two-row arrangements before the three-by-three grid won — it tested fastest, and people read the digits most naturally running left to right, top to bottom.
The founding insight of the whole field: you design around the person, not the machine, always.
The era is full of small decisions that quietly outlived it.
Yes. There’s a hamburger menu in there.The hamburger menu is one — Norm Cox drew those three stacked lines in 1981 for the Xerox Star, and four decades later we’re still arguing about them in bars.
The language caught up slowly. Brenda Laurel used the phrase “user experience” in a 1986 essay, well before it was anyone’s title.
Don Norman coined it for his group at Apple in 1993 — the first person to carry the term in a job title — and he meant something wide: industrial design, graphics, the interface, the physical interaction, even the manual. The whole experience. Not the screen.
Norman’s word was never about screens. We made it about screens because screens were the part we could control.
What It Got Right
Experience is the entire system a person touches, not the rectangle they tap. Karlin’s team designed around hands and memory; Norman drew the term wide enough to hold the hardware, the interface, the interaction, even the manual.
The founding definition was expansive on purpose, and at its best the field kept that whole picture in view instead of retreating to the part that was easiest to control.
What It Got Wrong
Over the following decades we narrowed that definition down to flows and pixels, because pixels were the part designers owned. Now the rectangle is dissolving into voice, agents, and ambient interaction, and the narrow version leaves you designing the one part of the experience that is disappearing.
Today you can read the cost off the org chart: design owns the screen, while the model’s behavior, the data, and the policy get decided in rooms it was never invited to.
Pioneers
- John E. Karlin, the first psychologist hired at Bell Labs to design how people use systems
- Douglas Engelbart, who demonstrated the mouse and interactive computing in 1968
- Jef Raskin, who started the Macintosh project around the human interface
- Bruce Tognazzini, Apple employee #66, who wrote the first Apple Human Interface Guidelines in 1978
- Don Norman, who wrote The Design of Everyday Things (1988) and became the first person to hold a “user experience” job title, ever
- Brenda Laurel, who used the phrase “user experience” in 1986
Key books
- The Psychology of Human-Computer Interaction — Stuart Card, Thomas Moran, and Allen Newell (1983)
- Designing the User Interface — Ben Shneiderman (1987)
- The Design of Everyday Things — Don Norman (1988)
- Tog on Interface — Bruce Tognazzini (1991)
Typical devices
- Switchboards, rotary phones, and the touchtone keypad
- Aircraft cockpit instrument panels and early radar consoles
- Mainframe terminals and teletype machines
- Xerox Alto and Star workstations
- The original Macintosh (1984)
The Web Era (1994–2007)
The web arrived with no map. Early sites were filing cabinets dumped on the floor, and the discipline that rescued them was information architecture. Card sorts, taxonomies, navigation schemes, Jakob Nielsen’s usability heuristics. For roughly a decade, the central question of the field was not how a page looked but whether a person could find anything on it.
The browser also made publishing nearly free, and the predictable thing happened: everyone made a website, and suddenly the world needed people who could organize them faster than it could train them. The job grew faster than anyone could fill it.
This is also when the work got a process — Kelly Goto’s Web Redesign: Workflow That Works gave a chaotic young field a repeatable way to plan, budget, and ship a site, and taught a generation to treat redesign as a discipline rather than a scramble.
Information architecture didn’t die. It went underground and waited for the machines to need it.
What It Got Right
This era taught designers that structure is the product. How information is organized decides whether anything is usable at all, before a single visual choice gets made. A beautiful site with a broken hierarchy is a broken site.
We argued about labels and parent-child relationships because those arguments were load-bearing — get the structure wrong, and no amount of polish on top of it saves the experience.
What It Got Wrong
Then we mostly stopped. As apps and design systems took over, IA got demoted from a discipline to a deliverable, then to a phase nobody had time for, then to a sitemap made once and never reopened. Structure never stopped mattering; it just stopped being visible.
Today the bill arrives: teams wiring up AI agents find no legible structure for a model to reason over, and rebuild, under deadline, the architecture they deleted a decade ago.
Pioneers
- Jakob Nielsen, who codified usability heuristics and discount usability testing
- Steve Krug, who distilled it all into Don’t Make Me Think (2000)
- Jesse James Garrett, who mapped The Elements of User Experience and later named Ajax
- Peter Morville and Louis Rosenfeld, whose polar bear book defined information architecture
- Alan Cooper, who introduced personas and interaction design through About Face
- L. Jeffrey Zeldman, who championed most of the web standards we live by today.
- Kelly Goto, who turned web redesign into a repeatable workflow and championed design ethnography
Key books
- Information Architecture for the World Wide Web — Peter Morville and Louis Rosenfeld (1998)
- Don’t Make Me Think — Steve Krug (2000)
- Web Redesign: Workflow That Works — Kelly Goto and Emily Cotler (2001)
- The Elements of User Experience — Jesse James Garrett (2002)
Typical devices
- Desktop PCs with CRT monitors, driven by keyboard and mouse
- Dial-up modems giving way to early broadband
- Web browsers: Mosaic, Netscape, Internet Explorer
- Palm Pilot and the first PDAs
- Public information kiosks
The Mobile Era (2007–2014)
The iPhone shipped in 2007 and handed the field a constraint it had never had: a screen the size of a playing card and a person using it one-handed on a moving train.
You could not fit the desktop in there. You had to choose.
It also put a computer in nearly everyone’s pocket, which is where the thread from the intro shows up again. Computing got cheaper and smaller, so far more people used it, and the demand for designers who understood touch and mobile exploded.
The latest incarnation of mobile in a foldable form factor.The constraint even resurrected old ideas: the hamburger menu, dormant since the Xerox Star in the 1980s, came roaring back, because three stacked lines were the most economical way to hide a menu on a tiny screen. A decision made in 1981 became the default for a billion phones — not because it was beloved, but because the new scarcity made it useful again.
The phone made you choose. Take the phone away and most designers stop choosing.
What It Got Right
This was the most clarifying decade the field has had. Constraint forced focus. One primary action per screen. Progressive disclosure. Cut the secondary, hide the tertiary, kill the rest.
Designers learned to defend a single decision against a roomful of stakeholders who all wanted their feature above the fold, because there was no fold left to fight over. Scarcity did the editing, and the work got sharper for it.
What It Got Wrong
Then the constraints relaxed. Screens got bigger, scroll went infinite, compute got cheap, and the discipline the small screen imposed quietly evaporated.
We went back to cramming.
Constraint was never supposed to come from the device; it was supposed to come from the designer. AI removes the last hard one, the cost of producing the work, so nothing external forces the cut anymore. Today you see the result in interfaces bloated with everything the model could generate, because nothing in the tooling says no.
Pioneers
- Scott Forstall, who led the software that became iOS
- Jony Ive, whose industrial design set the hardware language
- Luke Wroblewski, who argued the case for Mobile First (2011)
- Josh Clark, who wrote Tapworthy and pushed designing for touch
Key books
- Tapworthy: Designing Great iPhone Apps — Josh Clark (2010)
- Mobile First — Luke Wroblewski (2011)
- Responsive Web Design — Ethan Marcotte (2011)
- The Lean Startup — Eric Ries (2011)
- Hooked: How to Build Habit-Forming Products — Nir Eyal (2014)
Typical devices
- The iPhone and the capacitive multitouch screen
- Android phones and the app-store model
- The iPad and the first wave of tablets
- Early fitness wearables such as Fitbit
The Design Systems Era (2014–2022)
Figma, tokens, component libraries, design ops as a job.
This era industrialized the parts of the work that repeat. The idea was older than the tooling, though: erin malone founded the Yahoo! Pattern Library (2004–2008), the first corporate pattern library published publicly, and the 2014 wave largely gave her team’s approach better tools and a budget line.
Instead of redrawing the same button forty times, you defined it once and propagated it everywhere.
The gains were real and measurable, if mostly self-reported: Figma’s own write-up credits Headspace with 20 to 30 percent time savings on routine tasks and Swiggy with halving feature rollout — vendor numbers, so read them as such. Independent work points the same way: in a controlled Sparkbox study, developers timed building the same form from scratch and then with IBM’s Carbon system, and the system produced faster, more consistent code, though the team was candid about the small sample.
The thread holds here too. Making design cheaper to produce did not shrink design teams; it let companies ship more product across more platforms, which meant more screens, more surfaces, and more designers to tend them.
Cheaper production bought more product, not less work. The design system was the factory, and the factory ran hot — right up until the next cost collapse arrived and changed what the factory was even for.
Process was insurance against the cost of getting it wrong. When the cost of getting it wrong drops, the insurance becomes overhead.
What It Got Right
Systematize the repeatable so you can spend your judgment on the parts that actually need it. Define the button once, propagate it everywhere, and stop relitigating settled decisions.
The gains were measurable — faster builds, fewer inconsistencies, less redundant work — and the discipline of naming and reusing components was a real advance. Components are good. Consistency at scale is good. None of that is in dispute.
What It Got Wrong
What is in dispute is everything that grew up around the components — the ceremony, the handoff rituals, the choreography of approving a screen. It made sense when execution was slow and expensive.
But execution is collapsing toward free in the world of AI, and the ritual that once guarded against costly mistakes is now just a tax on every decision.
Today the cost shows in the gap between teams: some ship in days, others take weeks to approve an undisputed button.
Pioneers
- erin malone, who founded the Yahoo! Pattern Library (2004–2008), the first corporate pattern library published publicly
- Brad Frost, who formalized Atomic Design
- Nathan Curtis, who wrote much of the working theory of design systems
- Alla Kholmatova, whose Design Systems (2017) gave teams a shared playbook
- Matias Duarte, who led Material Design (2014)
- Dylan Field and Evan Wallace, who built Figma into the era’s default canvas
Key books
- Org Design for Design Orgs — Peter Merholz and Kristin Skinner (2016)
- Atomic Design — Brad Frost (2016)
- Design Systems — Alla Kholmatova (2017)
- Expressive Design Systems — Yesenia Perez-Cruz (2019)
Typical devices
- Responsive web spanning phone, tablet, and desktop screens
- Standardized laptops and large external monitors as default kit
- Smartwatches (Apple Watch, 2015)
- Smart speakers and voice assistants (Amazon Echo, 2014)
- Smart TVs and connected-TV interfaces
The AI Era (2022–Present)
Generative tools went mainstream fast. By 2025, generative AI adoption among US adults reached 54.6 percent, up ten points in a year, according to the St. Louis Fed. Inside companies the picture is more measured: the US Census Bureau’s Business Trends and Outlook Survey put firm-level AI use at roughly one in five through early 2026, and near 37 percent among firms with 250 or more employees.
That sits well under the louder headlines — self-selected commercial surveys tend to run two to four times higher, because they ask a broader question of a self-selecting crowd.
The cost of producing the work is falling toward zero, and the field is feeling it.
The UXPA salary survey found the worst net staffing reduction on its record, worse than the 2009 financial-crisis trough, with 37 percent of organizations reporting layoffs. The boom-time hiring of 2021 and 2022 has, by most readings, reverted to the mean.
Read that as a signal about the work, not a verdict on you. Here is the test of the thread: every earlier cost collapse grew the field, but this one so far looks like contraction. Jevons paradox says cheaper production should pull in far more work, and it may yet — but only for the part of the job AI can’t do, the way desktop publishing kept the editors and absorbed the paste-up artists. Drawing the screen is no longer where your value sits.
You are no longer the person who draws the screen. You are the person who decides what the screen is allowed to do.
What It Got Right
Here is where the whole history converges. The lesson the AI era is teaching, early and unevenly, is that the leverage moved upstream — to the rules, the structure, and the information the system reasons over.
That is the Norman lesson, design the system and not the rectangle, and the IA lesson, that structure decides whether any of it works, both returning at once. For the first time the two oldest insights in the field are also the most immediately useful.
And we almost forgot it.
What It Got Wrong
The first instinct was to bolt a chat box onto everything, then point the new tools at producing the same disposable screens faster — the oldest habit in a new coat.
Today you can spot it in products where the AI is a chat bubble stuck beside the old interface instead of rethinking it, and in teams measured by mockups shipped while the decisions that shape the experience happen in the prompt and the data. Stop counting mockups; start measuring the clarity of the structures the machine executes against. That was always the actual job.
Pioneers
- Don Norman, still widening human-centered design to meet the moment
- John Maeda, tracking wh