Larry Masinter and Frank Halasz on Xerox PARC, Interlisp, NoteCards, and why “residential programming” still matters
On the March 10, 2025 episode ofDo You Speak Tech?, I sat down with two names woven into the early fabric of modern computing:Larry MasinterandFrank Halasz. Their careers span the era when Xerox PARC wasn’t just a research lab … it was a kind of technological weather system, generating ideas that would later condense into today’s interfaces, networks, and web culture.
Masinter is widely known for his role in establishing standards for the World Wide Web. Halasz is a key figure in hypertext history, best known as one of the principal developers ofNoteCards, an early hypertext system built at PARC. Together, they’ve also been central voices in theMedley/Interlisp revival, an effort to preserve (and make usable again) one of the most influential programming environments ever created.
What follows is an edited, article-style story drawn from that conversation: a tour through PARC’s “golden age,” the philosophy behind Interlisp, the cost of being “too far ahead,” and why the past might still contain tools for the future.
“Utter freedom”… with one condition: it had to be good
Ask people what made PARC legendary and you’ll often get the same answer:culture. But culture is vague until you hear how it operated day to day.
Frank Halasz doesn’t hesitate: working at PARC felt like“utter freedom.”You arrived in the morning and worked on what you believed mattered. Projects were often self-directed, shaped by curiosity, and refined through constant peer interaction. The lab’s density of talent meant help (and strong opinions) came whether you asked for it or not.
Larry Masinter frames it slightly differently: yes, there was freedom, but it came with accountability. No one told you what to do, but each year you had to explain what you’d done, andit better be good. The twist?No one really defined what “good” was.That ambiguity could be stressful, but it also protected exploration.
In the interview, Masinter draws a line that still feels relevant today: the difference betweenresearchandengineeringisn’t so much what you do, but how success is judged. If a product fails to work, engineering failed. If a research prototype fails but you understandwhy, research may have succeeded. PARC lived in that experimental zone … sometimes building systems that ran, sometimes building theories, often doing both at once.
Supporting a living system, not just prototyping an idea
One reason Interlisp evolved so differently from many research projects is that it wasn’t born as a “prototype-first” experiment. Masinter describes a group that began bysupporting and extending a running systemused by other teams, something closer to infrastructure than a lab demo.
That mattered, because it meant Interlisp wasn’t only an idea: it was a lived environment. It had users. It had workflows. It had expectations. It was tested not just by evaluation papers, but by daily dependence.
Over time, it “morphed” into a more conventional research effort, but it never stopped being practical. That hybrid identity (part research platform, part production environment) helped form what both guests repeatedly circle back to: a philosophy of programming that feels strangely modern again.
Did they know they were shaping the future?
With inventions like Ethernet, laser printing, and windowed interfaces tied to PARC’s story, I asked the question everyone asks:did it feel historic while it was happening?
Masinter points to a quote attributed to Alan Kay:the best way to predict the future is to invent it.There was a sense they were running ahead and that the hardware itself was a time machine. Researchers were using machines far too expensive for ordinary office work, effectively “buying access” to capabilities the mass market wouldn’t see for years.
Halasz adds a useful nuance: by the time he arrived, PARC’s reputation was already established so much that his first year overlaps with the famousSteve Jobs visit, when Jobs saw PARC demos and pushed his own team toward similar ideas. But inside PARC, that external awe could fade into routine. You didn’t walk around thinking you were changing history; you were building what you needed, because it seemed like the right next step.
And sometimes, beingthatfar ahead became its own trap.
When you assume the world has Ethernet… and it doesn’t
One of the most vivid moments in the conversation is Halasz’s story about how PARC’s assumptions collided with reality.
At PARC, networked computing wasn’t a feature, it was the baseline. Workstations assumed a local network, shared printers, file servers, and services like time synchronization. In that world, a machine that needed a network time server to boot didn’t feel strange.
But outside PARC, the broader market wasn’t there yet. The result was a commercial mismatch: systems designed for a fully networked future, sold into a present that didn’t understand the prerequisites.
In other words: PARC didn’t just invent the future. Sometimes itoverfits.
Interlisp’s big idea: treat the program like a database
When the conversation turns from PARC to Interlisp itself, the tone shifts from nostalgia to philosophy.
Masinter describes Interlisp’s orientation as fundamentally different from the “compile files and run” model that came to dominate mainstream programming. Interlisp treated theprogram as a database. You didn’t primarily “edit files” … you workedinside a living environment, modifying running code, evolving systems from within, and saving out when necessary.
Halasz names this explicitly:“residential programming.”The phrase captures the feeling of inhabiting your software while you build it. Source code across the environment was accessible. You could patch behavior on the fly. When something broke, you didn’t always restart the world … youfixed the world you were currently inside.
They argue that something important was lost when the industry standardized around file-based compilation workflows. Interlisp enabled fast invention. The tradeoff was engineering discipline.
The dangerous superpower: you could change anything
If Interlisp made experimentation easy, it also made long-term maintainability hard.
Halasz gives a candid example from his own work onNoteCards, the hypertext system he built entirely in Interlisp. In the PARC environment, it was normal to reach into another component, for example the text editor, and tweak internals to get your program working immediately. That freedom got results fast, but it violated modularity and clean interfaces.
Later, when Halasz moved into managing larger product-scale software projects, he revisited NoteCards with different eyes. To make it sustainable, he needed “public interfaces” instead of internal hacks, meaning collaboration, negotiation, and design discipline. PARC’s invention culture could produce astonishing prototypes; turning them into products demanded a different muscle.
Is there an Interlisp echo in today’s development culture?
Both guests see partial parallels today especially inweb developmentand the world of rapid iteration.
Masinter compares modern microservices, scripts, APIs, and “assemble-it-as-you-go” systems to the Interlisp style: you don’t build one monolith; you build a constellation. Halasz points to JavaScript-in-the-browser and even Python environments as carrying a faint resemblance to residential programming, fast feedback, interactive loops, and experimentation baked into the workflow.
But they also suggest modern tools still haven’t fully recaptured Interlisp’s power: the ability to patch a running system with deep visibility into every layer, without the long compile-load-run cycle.
If you’ve ever wished your IDE felt more like a living workshop than a pipeline, this is the problem Interlisp solved decades ago.
Lessons from the UI that “got it wrong”
One of the most entertaining, and genuinely insightful, parts of the interview is Halasz describing what it’s like to use Medley today.
You notice immediately that it’sdifferent. The scroll bar appears on the left. Interfaces assume three mouse buttons. Title bars behave unlike modern window systems. And crucially, Interlisp lacked a concept that modern GUI programming treats as foundational:the event loop.
Halasz doesn’t present this as failure. He presents it as a learning opportunity: using Medley now reminds you that many “obvious” interface conventions are historical accidents. We evolved into today’s patterns; we didn’t discover them like physics. Seeing alternate paths helps developers recognize what is essential versus what is convention.
“AI doesn’t know anything”: a sharp warning from the past
At one point of the interview, we couldn’t avoid the big modern topic: AI.
Masinter, who lived through earlier AI cycles, offers a blunt caution: today’s machine learning systems are powerful, but people confuse impressive outputs with understanding. His formulation is memorable:today’s AI is “all form and no substance.”He worries about mythology, the singularity talk, the certainty that machines are “thinking,” and calls much of it “hogwash.”
Halasz distinguishes between the “AI” of PARC’s era, largelysymbolic AI, and today’s machine-learning-heavy landscape. In his view, Medley isn’t particularly useful for modern ML workflows, which require efficiency and tooling ecosystems already well served by today’s languages and environments. But Medley may still matter for symbolic traditions, linguistics research, and, probably most importantly, software preservation.
From museum artifact to living system: the Medley revival
A core theme of the interview is that software history shouldn’t be a static display.
Masinter and Halasz want people torunthese systems, not just read about them. To really understand old software, you need to inhabit its assumptions, experience its interactions, and see how it shapes thought.
That’s the spirit behind the Medley/Interlisp revival and the ecosystem aroundInterlisp.org, which the guests describe as having three major areas:
- History: bibliography, glossary, papers, talks, videos, and documentation
- Project: who’s involved, what’s been done, annual reports
- Software: ways to try Medley today including running it online and installing it on modern operating systems
They’re also motivated by access. Interlisp once ran on rare and expensive machines. Now it can be made available widely, letting people explore not only the environment, but the programs built inside it, many of which had limited distribution at the time.
The surprising size of the software preservation movement
Halasz shares a point that caught him off guard: software preservation is no longer niche. Major universities increasingly have dedicated library staff focused on preserving and reviving software. Networks exist, graduate programs exist, and volunteer communities are restoring everything from classic games to foundational early programs (even projects likeELIZA).
In that context, the Medley revival isn’t just nostalgia. It’s part of a growing effort to keep computing historyusable, so future technologists can learn from it and perhaps steal a few good ideas.
A final takeaway: timing, translation, and product reality
Late in the conversation, we return to a question that always haunts PARC stories:how did Xerox invent so much… and still not become Apple or Microsoft?
The answer is complicated, but both guests underline two factors:
- Cost and readiness: The technology was often too expensive, too network-dependent, or too early for mass adoption.
- Timing and packaging: You don’t just need the invention. You need the moment, the market, and a product that meets people where they are.
Halasz tells a cautionary tale of a PARC spin-out: a brilliant technology that embedded data into the grayscale of photocopies (robust even through repeated copying). It worked wonderfully at what it did. But as a fax product, it didn’t excel at the main thing people bought fax software for so it failed commercially. A lesson: invention must be layered onto what users already value, not offered as a replacement for it.
Where to start today
Masinter’s advice is practical: if you want to try Medley Interlisp now, start atinterlisp.org, explore the history section for context, then jump into the software section to run it online or install it locally. Be prepared for a learning curve especially keyboard shortcuts and interface conventions that don’t match modern expectations. The reward, they suggest, is a new perspective on what programming environments can be.
Why this matters now
This interview isn’t only about celebrating PARC. It’s about recovering a kind oftechnological imaginationand the willingness to rethink fundamentals.
Interlisp reminds us that environments shape how we think, and that many “modern” workflows are not inevitable. It also reminds us of a tension we still live with: the tradeoff betweenspeed of inventionanddiscipline of engineering.
And maybe the deepest lesson is Halasz’s: you rarely recognize a special environment while you’re inside it. You recognize it later, when it’s gone.
If you care about software history, innovation culture, or how developer tools shape the future, spending time with Medley isn’t just an exercise in retro-computing.
It’s a way of asking a living question:What did we forget that we still need?