Part IThe Story
Two Minutes and Thirty Seconds
The document was a tax form. Not a sonnet, not a typeface specimen, not one of the illuminated manuscript pages he would later hang in his living room in Los Altos. It was a government form, with all the dashed lines, ruled boxes, and tiny labeled fields that make tax forms the most quietly hostile objects in American life. In late 1984, with Apple preparing to unveil the LaserWriter at the January 1985 show, John Warnock sat down and coded the whole thing by hand in PostScript, the page description language his eighteen-month-old company had staked its existence on. He wrote subroutines for the dashed lines. He wrote subroutines for what he would later call the "zillions of little fields and things like that." Then he sent the file to the printer and waited.
Two minutes and thirty seconds.
Steve Jobs loved the form. He wanted it on stage, coming out of the LaserWriter in front of the audience as proof that a desktop machine could now produce a page indistinguishable from a print shop's. He also wanted it faster. "John," Jobs told him, as Warnock recounted to Knowledge at Wharton nearly three decades later, "I can talk — but I can't talk for two minutes and 30 seconds waiting for something to print."
So Warnock did what he had always done with a problem that was too big: he broke it into smaller pieces until each piece was trivial. He wrote a little program he called Graph Bind, which caught each graphics operator as it executed and wrote it out, flattening the elaborate procedural language of PostScript into a simple, linear list of marks on a page. The form printed fast enough for Jobs to talk over. The demo worked. The LaserWriter shipped. And somewhere in that hack, written to spare a showman an awkward silence, sat the germ of the file format that would eventually carry court filings, tax returns, academic papers, mortgage documents, consumer manuals, and the rest of the bureaucratic sediment of the paperless age. The PDF.
It is a strange inheritance for a man who described himself, without apparent irony, as someone who had "never separated my visual world and the world of mathematics" — who drew, painted, photographed, collected first editions of Copernicus and Galileo, and could not abide software he found ugly. Warnock's life was organized around beauty and around exactness, around the conviction that a page should look on a machine the way a page had looked for five hundred years in the hands of the best printers. His most widely used invention is the format in which you download your insurance statement.
He would not have seen a contradiction. The tax form was the point. The hardest problems were always the ordinary ones, and the measure of a solution was whether anyone, anywhere, on any machine, could stop waiting.
By the Numbers
The Warnock Ledger
32Pages in his 1969 doctoral thesis, the shortest in University of Utah history
$5MSteve Jobs's 1982 offer to buy Adobe outright, refused
19%Stake Apple bought instead, at five times Adobe's valuation
400+Third-party programs supporting PostScript by 1987, with 19 printer licensees
$1.02BAdobe's 1999 sales, his last full year as CEO
20Patents held
200,000Adobe shares donated in 2003 to build the University of Utah's engineering building
The Cheese Wrapper
The counselor at Olympus High School had the test results in front of him. This was the tenth grade, in Holladay, a suburb of Salt Lake City, sometime in the mid-1950s, and the boy across the desk had already failed ninth-grade algebra. Asked what he hoped to do, John Warnock said he might like to go to college, maybe study engineering. The counselor told him he had no aptitude for anything technical. The test did indicate he worked well with his hands. Perhaps, the counselor suggested, he could become a cheese wrapper.
Warnock told this story to the University of Utah's graduating class of 2020, in a commencement address delivered to an empty Huntsman Center and a scattered audience watching on screens during the first spring of the pandemic. He was seventy-nine. He told it without bitterness, which is perhaps the most telling thing about it. The point of the story was not the counselor. The point was the next man.
George Barton taught mathematics at Olympus, and his method, as Warnock described it, was "really quite simple." Pick up a college-level algebra textbook. Solve every problem in it. Then move on to trigonometry and do the same. Then analytic geometry. Barton would walk into class and ask how many students could solve all the problems in the book, and a week later, Warnock recalled to the Computer History Museum, "a guy would come in with a stack of paper and have solved all the problems in the book." The approach was brute force in the service of fluency, the opposite of the way most mathematics is taught, which is to say gingerly. Warnock went from C's to A's and B's — in math and, mysteriously, in everything else. Forty percent of Barton's students, by Warnock's count, went on to advanced degrees. One classmate, Joe Taylor, finished a PhD four years after leaving high school and eventually ran the University of Utah's mathematics department.
"I fell in love with mathematics," Warnock said. "It changed me from being a non-student to a good student."
He came from a family that drew and painted. His father, Clarence, was a prominent Salt Lake attorney; his mother, Dorothy, kept the house; there was an older brother and an older sister. In the summers he worked as a stock boy in a photography shop, a job that fed and funded what became a lifelong habit of the camera. He graduated in 1958 and went to the University of Utah because, as he put it, "almost everybody at that time who grew up in Utah went to the University. My dad went to the University, and my mom went there for a while." He finished a bachelor's degree in mathematics and philosophy in three years and described himself, throughout, as "mediocre." He thought he might teach. He thought he might be a photographer. "I really didn't think about it much."
Then, in graduate school, he got "swept up" in a problem.
The Jacobson radical for row-finite matrices had been posed in 1956 by the algebraist Nathan Jacobson, in a monograph on the structure of rings, and had sat unsolved for eight years. Warnock worked on it for a year and a half and solved it. The result became his master's thesis in 1964 and, five years later, a paper co-authored with N. E. Sexauer in the Transactions of the American Mathematical Society. It was the first real evidence that the boy who had failed algebra possessed an unusual mind, and it was also, in its way, a dead end. Abstract algebra paid in prestige. It did not pay in money.
In the summer of 1963 he took a job recapping tires at a Firestone shop in Salt Lake City: grind off the old tread, apply adhesive, align the new tread, seal it, repeat. "Hot, dirty and incredibly noisy," he remembered. He lasted three weeks. "I remember thinking, 'This is crazy. I've almost got a master's degree in mathematics. I need to get a good job.'" He sent a résumé to IBM, which hired him and shipped him off to computer schools around the country. It was his first contact with the machines. He learned FORTRAN without difficulty — "with mathematics behind you," it was "fairly straightforward" — but the real education came on customer visits, solving actual problems, where he learned how operating systems worked because he had to.
He would later count the lateness of this introduction as luck. "My saving grace in life," he told Susan Lammers for her 1986 collection Programmers at Work, "is that I was not introduced to computers at an early age." He had gotten "a good, solid liberal education" first — mathematics, English, the basic sciences, philosophy — and only then turned to the machine. "Being acculturated to the rest of the society and then going into computers is a much more reasonable way to approach the problem."
In 1965, back on the Utah campus, he met a young woman named Marva Mullins. He married her five weeks later.
Divide Until It Is Simple
The University of Utah in the late 1960s was, improbably, one of the half-dozen most important places in the world to be doing computer science, and it was so because of money and one man.
The money came from the Advanced Research Projects Agency, ARPA, the Pentagon outfit born of the national panic that followed Sputnik. ARPA funded research centers at roughly fifteen universities and labs — MIT, Harvard, Berkeley, Stanford, UCLA — and gave each a specialty. Utah's was what the grant language called man-machine interaction, which meant, in practice, computer graphics. The man was David C. Evans, who ran the computer science department, recruited the interactive-graphics pioneer Ivan Sutherland (later a Turing Award winner), and with him founded Evans & Sutherland, among the first commercial computer graphics companies. Evans became Warnock's thesis advisor. Sutherland sat on his committee. Among Warnock's fellow students were Ed Catmull, who would make a short film converting his own left hand into a computer model and go on to run Pixar, and Jim Clark, who would found Silicon Graphics. It was a scrappy, improbable little cohort in a state best known to the rest of the country for its mountains and its church.
In 1968 the principal investigators of every ARPA center gathered at Rustlers Lodge in Alta, up Little Cottonwood Canyon. Warnock, a graduate student, was invited. "At that meeting," he recalled at a Silicon Slopes event in 2019, "they approved the communication network that was going to connect all of these centers together and that was the ARPANET … which is now the internet." The following year Utah became the network's fourth node, the first outside California.
The problem that occupied Warnock was older and more stubborn than networking. Computers at the time could draw lines — wireframes, the skeletal outlines of objects — but they could not reliably draw solids, because they could not decide which surfaces should be visible and which should be hidden behind others. A cube rendered as twelve lines is an ambiguous object; a cube rendered as three shaded faces is a thing you could pick up. Solving the hidden-surface problem was the precondition for realistic computer images, and it was computationally ferocious given the machines of 1968.
Warnock's insight began not with the computer but with the eye. "Suppose I examine a picture of a table with pencils on it," his thesis opens its account of the method. "I quickly determine that large areas on top of the table are open" — and the argument proceeds from there. Most of any picture is boring. The eye does not labor over empty tabletop; it spends its attention where the pencils cross and overlap. A computer should do the same.
So the algorithm asks a single question of a region of the screen: is this simple enough to draw? If the region contains nothing, or a single polygon, or has shrunk to the size of a single pixel, the answer is yes, and the computer draws it. If not, the region is divided into four equal quadrants, and the same question is put to each. Repeat. Recursive subdivision — divide and conquer, in the jargon — until every piece is trivial. As the Oxford Dictionary of Computing summarizes it, the method "solves the problem of rendering a complicated image by avoiding the problem."
That phrase deserves to be read twice. Avoiding the problem. Not by evasion but by refusal to confront the whole of it at once, by breaking it down until the hard part simply evaporates into a large number of easy parts. It is a temperament as much as a technique, and it would recur, with variations, in nearly every important thing Warnock did.
He wrote the thesis in words and pictures. It contains no code. From introduction to bibliography it runs twenty-six pages, thirty-two all told, and it earned him, as he liked to say, "the dubious distinction of having written the shortest doctoral thesis in University of Utah history." In June 1970 the algorithm appeared on the cover of Scientific American, accompanying an article by Sutherland. It was among the first methods to let a computer render a solid object, and it remains a fixture of graphics textbooks, a half-century of undergraduates learning to draw a quadtree on a whiteboard.
There is a second story from these years that Warnock's dean, Richard Brown, liked to tell, and it says as much about the man as the algorithm does. While working on his doctorate, Warnock and a colleague wrote the University of Utah's first computer-based registration program. Registration opened in January. They spent winter break in the Merrill Engineering Building, working, as Brown put it, "for days almost around the clock," sleeping a few hours at a time on the premises. "As he told me, the program had to work. The university was planning to do registration on the computer and there was no back up plan."
No backup plan. It worked. Utah became one of the first universities in the country to register students by computer. The elegance of the thesis and the brute endurance of the registration program came from the same person in the same building in the same few years, and it would be a mistake to remember only the first.
A Harbor Made of Text
After the doctorate Warnock drifted through the early commercial computing world — Computer Sciences Corporation, a stint at a start-up, and then Evans & Sutherland, his advisors' company — and it was there, by his own account, that the idea of PostScript first took shape. It came from New York Harbor.
E&S had a contract to build a ship simulator for the Maritime Academy. The simulator would project the view from a ship's bridge as it moved through the harbor, which meant someone had to build a digital model of the harbor itself: "1,500 buildings and tank farms," Warnock told Lammers, "and all the bridges and buoys and everything — all the landscape." Full color, three dimensions. They had a year.
The problem was that nobody yet knew what the simulator hardware would look like. And so the team made a decision that Warnock described with characteristic bluntness: "We decided the most stupid thing we could do was to design this database in a form that would be used directly by the simulator. In other words, to bind it up too tightly." Instead they wrote the harbor in text — a human-readable description of every building and buoy — and built a compiler to translate that text into whatever form the simulator eventually required.
Do not bind the description to the machine. Describe the thing itself, abstractly, and let each machine interpret the description as best it can. With a colleague named John Gaffney, Warnock built an interpretive language at E&S called the Design System, which ran on Digital Equipment PDP-11s and had, he said, the same syntax and structure that PostScript would later have. The harbor of text was the ancestor of the page of text. The idea that a printed page could be a program — a set of instructions that any printer, of any resolution, from any manufacturer, could execute to produce the best version of that page it was capable of — was already there, sitting in a compiler written so that a ship simulator could be built before anyone knew what kind of ship simulator it would be.
Then E&S offered him a promotion. The promotion meant moving back to Salt Lake City. He and Marva talked it over and decided they did not want to go. It is a small domestic decision, the kind families make every week, and it altered the history of computing. A friend, William Newman, worked at Xerox's Palo Alto Research Center. Newman asked around, and the name that came up belonged to a man about to start a new lab.
The World's Best Sandbox
Charles Matthew Geschke was a Jesuit-educated classicist from Cleveland who had come to computers sideways, the way Warnock had come to them late. He earned an AB in classics from Xavier University in 1962 and a master's in mathematics the following year, taught math at John Carroll University for five years, and then went to Carnegie Mellon for a PhD in computer science under William Wulf, finishing in 1972 and co-authoring Wulf's book on optimizing compilers. His father and his paternal grandfather had both been letterpress photoengravers — men whose craft was the faithful reproduction of images on paper, measured with a loupe. Geschke was devout, gregarious, an organizer, the kind of man who would later sit on the boards of the San Francisco Symphony and the National Leadership Roundtable on Church Management. If Warnock was the inventor who happened to run things, Geschke was the manager who happened to invent.
Geschke had arrived at PARC in October 1972 and spent years on programming languages and the tools used to build Xerox's Alto workstations and its Star office system. Not long before he met Warnock, he had helped stage an elaborate demonstration of the future office for Xerox's senior management in Boca Raton, Florida — a scripted production, built with help from Hollywood. He remembered, years later, the body language of the executives watching it. In an interview with Knowledge at Wharton, he folded his arms across his chest and tensed up to show what he had seen. "It was pretty clear to me that we were fighting a tough battle," he said, "but I was committed."
Back in Palo Alto he was given a new lab, the Imaging Sciences Laboratory, focused on graphics, optics, and electronic printing. He needed a chief scientist. In 1978 he interviewed John Warnock.
The charter Warnock received at PARC was, in effect, the harbor problem restated for paper. Xerox had pioneered the laser printer, but its printing protocols were rigid — fonts stored as fixed bitmaps, fine for letters and reports and useless for sophisticated graphics. Warnock was asked to build "the device-independent data representation for dealing with graphics," for the screen and for the printed page alike. In 1979, with a colleague, he built a language called JaM, a direct descendant of the E&S Design System. JaM was folded into Xerox's existing printing work and became Interpress, a page description language that could describe shapes as complex as typefaces. The Marconi Society, decades later, would characterize Interpress as "somewhat compromised by the many protocols of Xerox" — a polite way of saying it had been designed by a large corporation.
Even so, Xerox would not ship it to the world. The company adopted Interpress internally but showed no interest in licensing it to the printer and computer makers who were, at that very moment, beginning to flood the market with cheap machines and dreadful dot-matrix output. The only alternative to a dot-matrix page was a typesetting machine that could cost upward of $150,000.
"We spent two years of our lives going to meetings," Warnock told the Los Angeles Times in 1996. "Chuck and I were totally frustrated. One day, I went to Chuck's office and said, 'We can't continue to live in the world's best sandbox.'"
It is worth pausing on the phrase. Not the world's best lab, or the world's best research center. Sandbox: a place where children build things that are not meant to last, where nothing leaves the enclosure, where the tide — or the janitor — eventually comes. PARC had already invented the graphical user interface, Ethernet, the laser printer, and the modern personal computer, and Xerox had famously turned most of them into somebody else's fortune. Warnock and Geschke could see where Interpress was going. "There was a huge amount of technology inside Xerox PARC," Warnock told Forbes in 2000, "and it wasn't clear that Xerox was going to capitalize on it."
In 1982 they agreed to leave. "That was the beginning of Adobe," Warnock would write, nearly four decades later, in the hours after Geschke died.
Mudbrick
The creek ran behind both their houses in Los Altos. It was called Adobe Creek, after the clay found along its banks — adobe being the Spanish word for mudbrick, the sun-dried building block of the American Southwest, made from dirt and straw and water and patience. They named the company after it. The company started in Warnock's garage, was incorporated in December 1982, and acquired a logo — a stylized A — designed by Marva Warnock, a graphic designer who would later become a partner at Marsh Design in Palo Alto. From PARC they brought along four people: two computer scientists, Bill Paxton and Doug Brotz, and two electronics designers, Tom Boynton and Dan Putnam.
They did not, at first, know what business they were in. The original plans included a copy-service business and a turnkey office printing system — hardware, software, and printers sold as a bundle — an ambitious scheme for a garage. The venture capitalists of Silicon Valley were largely unmoved. Help came instead from David Evans, Warnock's old advisor in Utah, and from financial advisors who pushed the founders toward the simpler, cleaner model of selling software alone. They would write a new page description language from scratch, unencumbered by Xerox, and license it to the manufacturers.
"Because Warnock had briefly worked at a start-up and Geschke had spent his entire career in research," the Times reported in 1996, the question of titles resolved itself in a single exchange. "I said, 'This might not work long term, but since I've done it before, I should probably be chief executive,'" Warnock recalled, "and we've never talked about it again." Warnock was president for the first two years and CEO for the next sixteen. Geschke became chief operating officer in December 1986 and president in April 1989. They never, by any account, fought about it.
Before the year was out, Steve Jobs came calling. Jobs had seen what the new language could do and offered to buy the company outright for $5 million. Warnock and Geschke said no. Their investors urged them to find some arrangement, and so they did: Apple bought a 19 percent stake, at five times the company's valuation, and paid a five-year license fee for PostScript in advance. The deal made Adobe, by the company's own telling, the first company in Silicon Valley history to turn a profit in its first year. It is a perfect small piece of financial judo — refusing to be bought by the most famous man in the industry and getting him to fund you instead — and it was transacted by two former research scientists in their forties who had never run a company of any size.
PostScript itself was a radical bet on generality. Rather than store each letter as a fixed grid of dots, PostScript described letters, lines, and images mathematically — curves and outlines that could be scaled to any size and rendered by any printer at the highest resolution it could manage. A cheap 300-dot-per-inch laser printer and a $50,000 imagesetter could run the same file; the imagesetter would simply do it better. It was device independence, the lesson of the harbor database, made commercial. And it needed credibility with the typesetting trade, which meant fonts. Compugraphic declined to work with them. Linotype agreed, licensing Helvetica and Times Roman. When Jobs held up an early page, he reportedly called the result "God's own Helvetica!"
— Steve Jobs, on the Apple LaserWriterWhen that first page came out of the LaserWriter, I was blown away...No one had seen anything like this before. I held this page up in my hand and said, 'Who will not want that?' I knew then, as did John, that this was going to have a profound impact.
The LaserWriter shipped in 1985 with PostScript inside. And then — this is the part that gets elided in the triumphant version — sales flattened. "When the LaserWriter first came out, there was a lot of talk about it, a lot of pent-up demand," Geschke recalled in a joint interview with the Seattle Times, "but after a few months, the sales [to big businesses] began to flatten out and really slowed down." The product was brilliant and nobody quite knew what it was for. "We actually didn't figure it out [immediately]," Warnock admitted.
What saved it was a coalition. Adobe, John Sculley at Apple, and Paul Brainerd at Aldus, whose PageMaker layout program shipped in July 1985, got together and named the category: desktop publishing. Apple repositioned the Macintosh and the LaserWriter not as office equipment but as a printing press. "The customers just loved it," Warnock said. "They could buy a $10,000 platform and do publishing. It changed the economics of the publishing business."
"So we're responsible," Geschke added, "for all that stuff you lug in from the mailbox every day."
Brainerd remembered his first meeting with Warnock, in 1984. Warnock pointed at an illustration of the Gutenberg Bible hanging on the wall and said, "This is what it's all about — the aesthetics of publishing." It is a remark that would have sounded faintly grandiose coming from almost anyone else in the computer industry of the mid-1980s, an industry whose aesthetic peak was green phosphor. From Warnock it was simply a description of the job.
Adobe went public on Nasdaq in August 1986. By 1987 PostScript was the industry's standard printer language, supported by more than 400 third-party software programs and licensed by 19 printer manufacturers. In 1988 it added kanji, becoming in effect the first international standard for computer printing. That year the company was an $83 million business. In 1989 PC Magazine named Warnock its Person of the Year, praising PostScript for having "given new meaning to the term 'device independence'" and "fulfilled the laser printer's potential in ways that its investor never planned."
There is a revealing anecdote from Owen Densmore, an Apple engineer temporarily transferred to Adobe during the LaserWriter work. At lunch one day Densmore asked Warnock what he planned to do after the LaserWriter. "PostScript is a linguistic 'mother board,'" Warnock answered, "which has 'slots' for several 'cards.' The first card we (Adobe) built was a graphics card." He talked about other network services, a file server card. The printing work, Densmore came away understanding, was just the first component; the real goal was a programmable network. The man had built the industry standard for printing and regarded it as a peripheral.
An Artist's Soul
Marva Warnock needed better tools.
She was a working graphic designer in the 1980s, which meant her craft was still mostly physical — rulers and French curves and technical pens and the endless manual labor of drawing and redrawing precise shapes. Her husband had, in his company's font-development software, an internal tool for drawing exact curves on a screen. In late 1986 he set out to turn it into a product that would automate the tedious manual parts of her work. The result, Adobe Illustrator, used lines and Bézier curves to render graphics that could be scaled infinitely without degradation. It shipped in early 1987 for the Macintosh. It was Adobe's first application — the first time the company had sold anything directly to the people who made images rather than to the companies that made printers — and it altered the company's course. It also, not incidentally, sold a great many PostScript laser printers, because Illustrator's output was PostScript and only a PostScript printer could do it justice.
The origin is so domestic it borders on the sentimental: a husband building a drawing program for his wife. But the domestic detail is the business insight. Warnock built for an expert user he could watch every day, whose frustrations he did not have to infer from a focus group. Decades later, Shantanu Narayen, Adobe's CEO, would write in his memorial email that "John and his wife, Marva, who is a graphics artist, used our products constantly and set the standard for customer empathy."
"I draw, I paint, I studied geometry, I'm a mathematician," Warnock once said. The Computer History Museum put the line atop its memorial, under the heading "A Mathematician with an Artist's Eye."
— John WarnockI draw, I paint, I studied geometry, I'm a mathematician. I've never separated my visual world and the world of mathematics.
Jonathan Seybold, the publishing analyst whose seminars were the trade's annual gathering, put it more compactly: "He's trained as a technologist, but he has an artist's soul." The soul was evident to anyone who stepped into the Warnock house in Los Altos, where an elegant stone carving of the Latin alphabet stood in the entryway and a page from an illuminated manuscript hung in a corner of the living room. It was evident in the kitchen, where, Forbes reported in 2000, he made "a mean potato-and-butternut-squash gnocchi with fresh sage and white truffle butter." It was evident in a back porch with beehives kept by his son Jeff, from which came, Warnock reported with paternal precision, "incredible honey and beeswax candles."
And it was evident in his contempt for things he found ugly — a contempt that could be useful, and could be ruinous.
Adobe's first products after PostScript were digital fonts, released in a proprietary format called Type 1, worked on by Bill Paxton. Adobe kept key elements of the Type 1 format secret. It was, Warnock argued years later, the right call technically, and for a while commercially — a way of protecting the crown jewels while publishing everything else. But the secrecy invited exactly the response it was meant to prevent. Apple developed a competing format, TrueType, and licensed it to Microsoft. The industry split. Michael Miller, then editor of PC Magazine, remembered watching Warnock "get teary and later combative in fighting for PostScript, although in the end he lost." The formats were eventually reconciled in OpenType, but Warnock never conceded the aesthetic point. Type 1, he insisted to Knowledge at Wharton in 2010, had "always been … a better solution than TrueType."
The same taste that made PostScript beautiful made the World Wide Web look, to Warnock, like an affront. "The early versions of HTML — from a design point of view — were awful," he said. "There was nothing beautiful about it." Adobe, by its own later admission, sat on the sidelines of the web longer than most companies. Warnock avoided Microsoft software because the company "has never had good taste." Its products weren't "cool." He thought Tim Berners-Lee had made "a fundamental error" in the design of the web, and when the subject turned to fonts online, he could sound genuinely bereaved. "It's really sad," he said in 2007, of the bleak typographic landscape of the web. "It's not that [kids on MySpace] wouldn't like to use different typefaces. It's that you're prohibited from doing anything."
In February 1996, the cost of all this came due. Adobe announced an $11.8 million loss for the quarter ended December 1, the hangover from an acquisition binge — chiefly its 1994 purchase of Aldus, which had made Adobe a company with $525 million in annual revenue and drawn the Federal Trade Commission's attention, forcing the divestiture of Aldus's FreeHand illustration program. The stock fell $13.50 in a single day, 23 percent, to $45. The Los Angeles Times found Warnock, "normally reserved," animated on the subject of the internet, "his pale blue eyes flickering with intensity."
"There's no chance we won't dominate the Internet," he said. "Adobe's success is that we have not ignored 500 years of graphic-arts history. That's a mistake being made by the computer industry."
It is a sentence that is simultaneously right and wrong, and rarely has a single quote contained a whole career so neatly. Adobe would, in fact, come to dominate the tools used to make the visual internet. It would not do so on the terms Warnock imagined — through PageMill, a now-forgotten web-page builder, or through the elegant, fully specified documents he wanted the web to become. The web would stay ugly, more or less, and Adobe would sell people the tools to decorate it.
Camelot
The memo was dated spring 1991 — April 1, according to one archive, though nobody seems to have treated it as a joke. It ran six pages. Warnock called the project Camelot, which tells you something about how he regarded it.
"This project's goal is to solve a fundamental problem," it began. "There is no universal way to communicate and view ... printed information electronically." The fax machine existed. Email existed, in a rudimentary way, for the technically inclined. But a document created in one program on one computer could not be reliably opened, viewed, or printed on another. Send someone a word processor file and they needed the same program, often the same version; even then the fonts might change, the text reflow, the pages break in different places. Every machine was its own island, and documents crossed between them badly or not at all.
— John Warnock, "The Camelot Project," 1991What industries badly need is a universal way to communicate documents across a wide variety of machine configurations, operating systems and communication networks. These documents should be viewable on any display and should be printable on any modern printers. If this problem can be solved, then the fundamental way people work will change.
The solution came straight out of the tax form. PostScript, the language, was powerful because it was a full programming language — loops, procedures, subroutines — and that same power made it slow and unpredictable to display. But Graph Bind had shown, back in 1984, that you could run a PostScript program once, catch every mark it made, and write those marks out as a simple, flat description. Strip out the programming; keep the page. That flattened page, extended with fonts embedded inside it so the receiving machine didn't need to have them, became what Warnock first called Interchange PostScript and the world came to know as the Portable Document Format.
The memo also contained, almost in passing, a prediction that reads today like a description of the next thirty years of the internet. "Imagine if the IPS viewer is also equipped with text searching capabilities," Warnock wrote. "In this case the user could find all documents that contain a certain word or phrase, and then view that word or phrase in context within the document. Entire libraries could be archived in electronic form..."
In February 1992 he demonstrated the technology at the third TED conference, opening the same file on a Macintosh and on a Windows machine. On June 15, 1993, Adobe officially introduced the Acrobat product suite and the PDF. And almost nobody cared.
"Nobody got it," Warnock told Knowledge at Wharton at the product's twentieth anniversary. The software lost money for years. The Reader, the program needed merely to view a PDF, was originally priced at about $50 a seat — a toll on the very universality the format was meant to provide. Adobe's own board of directors, according to multiple accounts, repeatedly sought to kill the product. Warnock persisted. He had already articulated, to the New York Times in 1991, the argument that would eventually win: "I think the crossover point is if I can go to General Motors and say, 'I can deliver your information more quickly and more cheaply than you can on paper.' You're talking about savings of tens of millions of dollars."
What turned it was a pricing decision that reversed the whole economics of the product: Adobe made the Reader free. Anyone could view a PDF; only the people creating them had to pay. The format became a de facto standard as the web spread, precisely because a PDF was the one thing on the web that looked the same everywhere — the thing HTML, by design, refused to be. Adobe kept the format proprietary until 2008, when it became the ISO standard 32000-1, but it had published the specification all along, trusting that it could profit from the tools to make and manipulate the files rather than from the files themselves. Acrobat went from the board's candidate for execution to one of the company's most successful products.
Note the shape of the argument. Warnock disliked standards bodies — "horrible, horrible, horrible things," he called them, that "design by committee" and produce "a hodgepodge of stuff." But he loved standards. "The only way to make standards," he said, "is to get them out and just compete." PostScript had become the standard by winning; PDF did the same, slowly, against the indifference of the market and the skepticism of his own directors. The lesson he had half-learned with Type 1 — that secrecy invites rivals — he applied to PDF in full: publish the format, give away the reader, sell the pen.
"It had been a holy grail in computer science," he said in a 2019 interview with Oxford University, "to figure out how to communicate documents."
A holy grail. Camelot. The Arthurian vocabulary is not accidental, and it is not modest. The man who regarded HTML as aesthetically awful had given the world, in the end, a format for mortgage disclosures. He seemed entirely satisfied with that. Thirty years after Acrobat's launch, a few months before his death, he attended the anniversary celebration and spent the time afterward talking with Narayen about what Adobe's new generative image tool, Firefly, might do next.
A Gentlemanly Discussion
On the morning of May 26, 1992, Chuck Geschke pulled into the Adobe parking lot in Mountain View and was taken at gunpoint by two men — Mouhannad Albukhari, twenty-six, of San Jose, and Jack Sayeh, twenty-five, of Campbell. They drove him to a bungalow in Hollister and chained him. They called his wife, Nan, and demanded a ransom. The FBI listened to the calls.
Geschke's daughter left $650,000 at a drop-off point. When Albukhari picked it up, agents arrested him, and then, in the deadpan phrasing of the FBI agent who described it afterward, "[a]fter a gentlemanly discussion," Albukhari led them to Hollister. Geschke was released after four days, unhurt apart from what the chains had done. Both men were eventually sentenced to life in state prison.
It is the most violent episode in Adobe's history, and Warnock almost never spoke about it publicly — which is consistent with everything else known about the partnership. The two men had, by 1992, been working together for fourteen years, had founded a company together, divided its titles in a single conversation and never revisited the division. The Los Angeles Times, profiling them in 1996, described them as "an anomaly" in an industry "where few partnerships have withstood the test of time": two bearded computer scientists who had "set ego aside." Warnock, in a 1994 Wired interview, was described as "modest in the extreme, instantly sharing each ounce of praise, crediting his team and longtime partner Chuck Geschke at every possible opportunity" — and also, in the next paragraph, as "the ferocious competitor, the 20/20 visionary, the strategist with boardroom battle skills and the demeanor of Lao Tzu."
The company they built reflected a decision they had made before they hired anyone. "When we began, John and I wanted to build a company that we would like to work in," Geschke said in a 2008 interview that Warnock later quoted in his memorial tribute to his partner. "Why would you build a business you didn't want to work in?" Geschke spelled out the logic in a talk at Santa Clara University. A software company's "only significant assets are people and the ideas they discover and invent. Every night, those assets go home." Unless the company excited and rewarded them, "they may not return the next day." So they hired the best people they could find, "male or female, independent of their lifestyle choices, their ethnic background, or their race." They gave early employees a terminal, a modem, and a phone line so they could telecommute when family required it — in the early 1980s, when telecommuting barely had a name. Employees should "work smart not long."
Warnock was blunter about the alternative. "Unfortunately, I think people too often see the business ethic out there and say, 'This is the way you have to run a business to be successful. You have to be cutthroat and go for the kill.' I have never believed that." Good products did well. "The money takes care of itself."
He also kept himself close to the work. In 1993, Wired reported, Warnock "abandoned his glitzy CEO showplace in a new Mountain View, California, headquarters building" for "a no-frills desk in a corner of the engineering building that allowed him to 'escape from the accountants and attorneys.'" He was, by then, running one of the largest software companies in the world. He was still writing PostScript routines.
His case for the technologist-as-CEO was not sentimental. "I believe I have a fundamental understanding of what computers can do and what they can't do," he told Forbes, "where they can be used and where they shouldn't be used." Most technologies, he thought, were "really solutions looking for a problem," and it was "much more interesting to take technology and apply it to real economic problems that are causing people heartburn." But he distrusted the academic temperament too, his own former tribe. Researchers turned executives, he said, "may put too much faith in the rational process. I don't know that a lot of academics are looking to help people rather than solve the problem for the problem's sake."
By March 2000, Forbes called him "an endangered species in Silicon Valley's turnstile culture." Adobe had 1999 sales of $1.02 billion and a profit of $238 million; it was the world's fourth-largest desktop software maker. "The average tenure of high-tech CEOs is 2.4 years," Warnock observed. "I figure I've lived nine lives." The secret, he said, was "understanding your shortcomings and adapting. Many [young high-tech CEOs] figure they've learned the truth early in their careers, and that is their undoing."
Geschke retired as president in April 2000. In December, Warnock handed the chief executive's job to Bruce Chizen. The following spring, at sixty, he stepped down as chief technology officer too, staying on as chairman — a position he had held since April 1989 and shared with Geschke from September 1997. "I've had a wonderful career at Adobe," his statement read, "and I am proud of the products and technology the people of Adobe have created."
The partnership outlasted the job. They received the National Medal of Technology and Innovation together — the 2008 medal, presented by President Obama at the White House in 2009 — and the Marconi Prize together in 2010. They co-chaired the board until January 2017. When Geschke died of cancer on April 16, 2021, at eighty-one, Warnock's statement was eight words longer than it needed to be and not one word too long: "I could never have imagined having a better, more likable, or more capable business partner. Not having Chuck in our lives will leave a huge hole and those who knew him will all agree."
There is a coda to the partnership that belongs to Geschke's father, the photoengraver. In Adobe's early days, Geschke senior checked the company's color separation work with his engraver's loupe — the same instrument his own father had used to judge halftone plates in the letterpress era. He approved of the halftone patterns. Chuck Geschke called it "a wonderful moment." Three generations of men whose life's work was making images reproduce faithfully on paper, and the oldest of them, peering through glass at a computer's output, found it good.
The Rare Book Room
The collecting began in a London bookstore, with a 1570 edition of Euclid's Elements.
Of course it did. The foundational text of geometry, the book from which every subsequent mathematics descends, the book that George Barton's method — solve every problem, then move to the next — most resembles in its relentless sequence of proofs. Warnock bought it in the 1980s, and it was the first of many. By 2000 Forbes reported that he owned first editions of Darwin, Copernicus, Machiavelli, and Galileo, and one of the original copies of the Bill of Rights. In 1995 he founded a company, Octavo, to scan rare books at high resolution and sell the images on CD-ROM — Milton, Shakespeare, Galileo — drawn from his own shelves and from the greatest libraries in the world. In 2003 Oxford's Bodleian Library awarded him its Bodley Medal.
Octavo failed. "The company was not successful for a number of reasons," he said flatly in a 2013 interview, "but I had all the book files." So he put them on a website, rarebookroom.org, free to anyone with a browser — more than 400 volumes. The Camelot memo's offhand prophecy, "Entire libraries could be archived in electronic form," turned out to describe his retirement.
He and Marva collected Native American art too, beginning casually on family trips to the Four Corners with their three children — baskets, then pots, then moccasins, shirts, and beadwork from more than two dozen tribes, hundreds of objects that toured the country and were catalogued in Splendid Heritage: Perspectives on American Indian Arts, published by the University of Utah Press in 2009. "We think the early history of the United States is interesting," he said, "and the art produced by Native Americans is incredible." John and Marva personally assisted with cataract surgeries led by the ophthalmologist Geoffrey Tabin on missions to restore sight in the world's poorest countries, and endowed a chair at Utah's Moran Eye Center to fight preventable blindness.
He was an investor in odd and brave things. He backed Salon, the early online magazine, and when it turned its reporting on Kenneth Starr during the Clinton impeachment saga, the Wall Street Journal editorialized against the publication and named Warnock, its principal investor, in the condemnation. There was a bomb threat; there were advertising boycotts. "Throughout all of these dark times," wrote David Talbot, Salon's founder, "John Warnock never once pressured me to go easier." Talbot remembered, above all, the smile — the one that seemed to say: "you'll get this Internet thing right, you'll figure it out, you'll get to breakeven."
He kept going back to Utah. In 2003 he and Marva donated 200,000 shares of Adobe stock, worth more than $5.7 million, as the cornerstone gift of a $13 million campaign for a new engineering building at the University of Utah. It opened in 2007 as the John E. and Marva M. Warnock Engineering Building. They endowed chairs in computer science, mathematics, and fine arts. "I remember John telling me," the engineering dean, Richard Brown, said, "that, when he was a young man, he really needed the College of Engineering, so when he learned the College of Engineering needed him, he couldn't say no." In December 2012 Adobe opened a 280,000-square-foot campus in Lehi, Utah, the anchor of what came to be called Silicon Slopes — the company named for a California creek, back in the state of the boy who wasn't supposed to have the aptitude. At a 2019 tech summit in Salt Lake City, asked for advice, he did not talk about sales or marketing. "Always hire people that are smarter than you are," he said. "You have to let people create. You have to invent all the time."
The company grew past him, as companies do. By 2025 Adobe reported $23.77 billion in revenue and 31,360 employees. It moved its software to subscriptions, a shift that enraged many of the designers who had once regarded the company as one of their own; in 2021 a customer posted a $291.45 cancellation fee and set off a small revolt; in 2024 the Federal Trade Commission and the Department of Justice sued over hidden termination fees. Its $20 billion attempt to buy the design-software upstart Figma collapsed under regulatory pressure in December 2023, costing a $1 billion termination fee. Beneath a 2014 blog post celebrating the history of Illustrator, a commenter summarized the arc with a line borrowed from the opera Nixon in China: "Founders come first, then profiteers." It is unfair, as epigrams usually are, and also not entirely wrong, as they also usually are. A hastily assembled large-print tribute, The Influence of Dr. John Edward Warnock, a Well-Known American Computer Scientist, appeared within three weeks of his death, its subtitle running on for most of a line — a fate that would perhaps have amused a man whose doctoral thesis ran thirty-two pages.
He had pancreatic cancer for nearly two years. He kept appearing. In January 2023 he and Marva cut the ribbon on the Adobe Experience Museum, built with the Computer History Museum on the ground floor of the new Adobe Founders Tower in downtown San Jose. In March 2023 he was back on the Utah campus, receiving an IEEE Milestone award as one of the pioneers of computer graphics, for the algorithm that divides a picture until every piece is simple. Marva stood beside him in the photographs. He told the audience about an idea that had started at the U and traveled to Evans & Sutherland and PARC and Adobe: that printed images should be the product of a computer program.
He was working with the Computer History Museum on one last project: the transcript of a two-part oral history he had recorded in April 2018, nearly three hours on the first tape alone, with the museum's curator David Brock. They were finalizing it when he died at home in Los Altos on August 19, 2023. He was eighty-two.
Three days later the museum posted the transcripts on its website, two links at the top of Brock's memorial, labeled Download transcript here. They are PDFs. Version 1.7, according to the header in the file's first line. Anyone can open them, on any machine, in any country, and see the same pages: the margins where the margins were set, the line breaks where the line breaks fell, the words of a man from Holladay, Utah, laid out exactly as intended — no waiting.
Part IIThe Playbook
Warnock's career offers a rare thing in technology history: a founder who remained recognizably himself across four decades, from a 32-page thesis to a $1 billion company to a free library of rare books, and whose methods can be traced from the domain where he learned them to the domains where he applied them. The principles below are drawn from the decisions described in Part I. Some of them made Adobe. One or two nearly unmade it. Several cut against prevailing startup orthodoxy, then and now.
Twelve Principles
- 1.Subdivide until the problem is trivial.
- 2.Never bind the description to the machine.
- 3.Leave the sandbox when the sandbox won't ship.
- 4.Sell the stake, not the company.
- 5.Build for the expert at your kitchen table.
- 6.Name the category when the product stalls.
- 7.Give away the reader; sell the pen.
- 8.Compete standards into existence — but don't hoard the keys.
- 9.Fund conviction past the point of consensus.
- 10.Taste is a moat and a blindfold.
- 11.Settle the title question once, then never again.
- 12.Remember that your assets go home every night.
W
The Long Arc
Each major invention reused an idea from the one before it.
1964
Solves the Jacobson radical for row-finite matrices, posed in 1956.
1969
Thirty-two-page thesis introduces recursive subdivision for hidden surfaces.
1976
At Evans & Sutherland, builds New York Harbor as a text database, seeding PostScript.
1978
Hired by Chuck Geschke at Xerox PARC; JaM follows in 1979, then Interpress.
1982
Leaves PARC, co-founds Adobe in December; refuses Jobs's $5M offer, sells him 19%.
1985
PostScript ships in the Apple LaserWriter; desktop publishing is named.
1987
Illustrator ships, built to automate Marva Warnock's design work.
1991
Writes the six-page Camelot memo describing the PDF.
1993
Acrobat and PDF launch June 15; lose money for years.
2000
Hands CEO role to Bruce Chizen after sixteen years.
2008
PDF becomes ISO 32000-1.
2023
Dies August 19; oral history released as PDF three days later.
Principle 1
Subdivide until the problem is trivial
The Warnock algorithm asks a region of the screen one question — is this simple enough to draw? — and, if not, splits the region into four and asks again. The method "solves the problem of rendering a complicated image by avoiding the problem." That phrase is the most useful thing in this profile for an operator, because it reframes difficulty. Hard problems are rarely uniformly hard. Most of any picture is empty tabletop; the difficulty concentrates where the pencils cross.
Warnock applied the same move outside of graphics. Faced with a tax form that took two and a half minutes to print, he did not try to make all of PostScript faster; he isolated the one thing the demo needed — a fast, flat list of marks — and wrote Graph Bind to produce it. Faced with the impossible-sounding goal of universal electronic documents, he decomposed it into capture, transmission, viewing, and printing, and noticed that PostScript already solved most of the pieces.
The discipline here is resisting the urge to attack the whole. Most strategy failures are attempts to solve the full scene at once.
Tactic: When a problem resists you, divide it into four and identify which quadrant is actually hard — then spend all your effort there and solve the other three by rote.
Principle 2
Never bind the description to the machine
On the New York Harbor simulator, the team "decided the most stupid thing we could do was to design this database in a form that would be used directly by the simulator." They described the harbor in text and wrote a compiler to translate it later. That decision is the intellectual root of PostScript, which describes a page abstractly and lets each printer render it at its best resolution, and of PDF, which describes a document so that any machine can display it.
The principle generalizes far beyond file formats. Any time a company encodes its core asset — customer data, a product specification, a process — in the idiosyncratic shape of today's tooling, it creates a migration cost it will pay later, with interest. Warnock's career was built on the economic value of the layer that sits between the description and the device. Whoever owns that layer captures the value of every new device that arrives.
Tactic: Store your most important asset in a form that outlives the systems currently consuming it, and treat translation to any specific system as a replaceable compiler step.
Principle 3
Leave the sandbox when the sandbox won't ship
Warnock and Geschke spent "two years of our lives going to meetings" trying to get Xerox to license Interpress. The phrase Warnock chose for PARC — "the world's best sandbox" — is precise. A sandbox is a place where things are built but not released. The research was real; the path to the world did not exist.
The non-obvious part is what they did next: they did not take Interpress. They built a new language from scratch, unencumbered, improving on the "somewhat compromised" design that committee-style corporate process had produced. Leaving was also an opportunity to fix what the institution had forced them to compromise.
The tell, for anyone inside a large organization, is not rejection — rejection is normal — but the structure of the rejection. Xerox adopted Interpress internally while refusing to sell it. That combination, internal validation plus external indifference, is a near-perfect signal that the opportunity belongs to someone else.
Tactic: If your organization uses your idea internally but refuses to take it to market, set a deadline for one more attempt, then leave and rebuild the idea without the compromises you were forced to accept.
Principle 4
Sell the stake, not the company
In 1982 Steve Jobs offered $5 million for all of Adobe. The founders refused; their investors pushed for a deal; they sold Apple 19 percent at five times the company's valuation, plus a five-year PostScript license paid in advance. Adobe became profitable in its first year. The structure solved three problems at once: it capitalized the company without venture dilution, it locked in the most important customer of the decade, and it preserved independence — which mattered enormously when, decades later, Apple and Adobe found themselves on opposite sides of the Flash fight.
Strategic minority investments are often treated as a lesser outcome than acquisition. Warnock treated a prospective acquirer's eagerness as a pricing signal and converted it into prepaid revenue.
Tactic: When a would-be acquirer shows serious interest early, counter with a minority stake at a premium plus a prepaid commercial agreement, so their enthusiasm funds your independence.
Principle 5
Build for the expert at your kitchen table
Illustrator began as an attempt to automate the manual tasks Marva Warnock performed as a professional graphic designer. Narayen later credited the couple, who "used our products constantly," with setting "the standard for customer empathy." The domestic origin is not a charming footnote; it is a product method. Warnock could observe a demanding expert user daily, see exactly where her tools failed her, and judge whether his software actually helped.
This inverts the common advice to build for a broad market from day one. Adobe's first application was built for one person, a working professional, and succeeded because professionals like her existed in the tens of thousands. Breadth followed from depth.
Tactic: Identify a single demanding expert whose work you can watch directly, build a tool that removes their most tedious manual task, and refuse to generalize until that one person would pay for it.
Principle 6
Name the category when the product stalls
The LaserWriter launched to excitement and then, as Geschke admitted, sales "began to flatten out and really slowed down." The technology was not the problem. The market did not know what the product was for. The fix was not engineering but language: Adobe, Apple's John Sculley, and Aldus's Paul Brainerd jointly coined desktop publishing, and Apple repositioned the Macintosh as a printing press for graphic arts. A $10,000 platform suddenly replaced a $150,000 typesetter.
What makes this instructive is that it was done as a coalition. No single company owned the category; three companies with complementary products named it together and each profited.
Tactic: When a strong product stalls, convene the makers of complementary products and jointly name the job the bundle does — then reposition each product as part of that category.
Principle 7
Give away the reader; sell the pen
Acrobat floundered while the Reader cost about $50 a seat. It took off once the Reader was free and only the tools for creating PDFs carried a price. The pricing change reversed the network dynamics: every free reader increased the value of every paid writer. Adobe extended the logic by publishing the PDF specification openly while selling proprietary software to make and modify the files — what Knowledge at Wharton called the "delicate balance between publicly documenting its intellectual property and gaining a competitive advantage through proprietary software."
The principle is now familiar, but Adobe executed it before the term "freemium" existed, and it is worth noting what Warnock gave away: not a crippled version of the product, but the entire consumption side of the network.
Tactic: Identify which side of your product creates network value by its mere presence, make that side free and complete, and charge only the side whose output becomes more valuable as the free side grows.
Principle 8
Compete standards into existence — but don't hoard the keys
Warnock's view of standards was paradoxical and coherent: "Standards bodies are horrible, horrible, horrible things," but "the only way to make standards is to get them out and just compete." PostScript and PDF both became standards by winning in the market long before any committee ratified anything; PDF reached ISO status in 2008, fifteen years after launch.
The counterexample is Type 1 fonts. Adobe kept key elements secret, and the secrecy provoked Apple to build TrueType and license it to Microsoft, splitting the market. Warnock "lost," as one witness put it, despite believing to the end that Type 1 was technically superior. The lesson is that a market-made standard survives only if competitors find it cheaper to adopt than to replace.
§
Four Format Decisions
How openness determined the outcome of Adobe's standards bets.
| Decision | Openness | Outcome |
|---|---|---|
| PostScript licensed to printer makers | Published, widely licensed | Industry standard |
| Type 1 fonts with secret elements | Partially closed | Provoked TrueType |
| PDF with $50 Reader | Spec published, viewer priced | Stalled for years |
| PDF with free Reader | Spec published, viewer free | De facto, then ISO standard |
Tactic: Before keeping any part of a would-be standard proprietary, estimate what it would cost your largest potential partner to build a rival — and if that cost is within their budget, open the part up.
Principle 9
Fund conviction past the point of consensus
Acrobat lost money for years. Adobe's board, by several accounts, repeatedly tried to kill it. "Nobody got it," Warnock said. He persisted because he had articulated, in concrete economic terms, the condition under which the product would win — the moment he could tell General Motors he could deliver its information "more quickly and more cheaply than you can on paper," with savings in the "tens of millions of dollars."
That specificity is what separates conviction from stubbornness. Warnock's faith in the PDF was not faith in its elegance (though he believed in that too); it was faith in a crossover point he could name. The same founder who stayed wrong too long about HTML was right about PDF, and the difference is that one belief was aesthetic and the other was economic.
Tactic: When you defend an unprofitable product against your board, state the specific customer, the specific saving, and the specific condition that will mark the crossover — and kill the product yourself if that condition stops being plausible.
Principle 10
Taste is a moat and a blindfold
Warnock's aesthetic standards built PostScript, Type 1, Illustrator, and the PDF. "Adobe's success is that we have not ignored 500 years of graphic-arts history," he said in 1996. The same standards kept Adobe on the sidelines of the early web, because "early versions of HTML — from a design point of view — were awful." He disdained Microsoft for having "never had good taste." That year Adobe posted an $11.8 million quarterly loss and its stock fell 23 percent in a day.
The useful distinction is between taste as a product standard and taste as a market filter. Applied to your own products, high standards create durable advantage. Applied to the market — refusing to engage with a platform because it is ugly — they create blind spots, because ugly platforms that spread fast are often where the next customers are. Adobe eventually prospered on the web by selling tools to make it less ugly, not by waiting for it to become beautiful.
Tactic: Hold your own products to your highest aesthetic standard, but when a new platform offends your taste and is growing anyway, build tools for it immediately rather than waiting for it to improve.
Principle 11
Settle the title question once, then never again
"I said, 'This might not work long term, but since I've done it before, I should probably be chief executive,' and we've never talked about it again." The Warnock–Geschke partnership lasted from 1978 until Geschke's death in 2021, through an IPO, an acquisition binge, a kidnapping, and the internet. The Los Angeles Times called it "an anomaly." The decision about roles took one sentence and rested on a plain, unflattering criterion — prior experience, not ability or ego.
The rhyme with the algorithm is close: settle the trivial region and move on. Cofounder conflict is often not about the titles themselves but about the decision remaining open. Warnock closed it with a stated rationale and an acknowledgment that it might need revisiting, which is precisely why it never did.
Tactic: Decide cofounder roles early, on a single explicit and impersonal criterion, state aloud that it can be revisited if the criterion stops applying, and then refuse to relitigate it informally.
Principle 12
Remember that your assets go home every night
Geschke's formulation was that a software company's "only significant assets are people and the ideas they discover and invent. Every night, those assets go home." Adobe's founders responded with policies that were unusual for the early 1980s: telecommuting equipment for early employees, hiring irrespective of "lifestyle choices," "work smart not long." Warnock rejected the "cutthroat" ethic outright and moved his own desk into the engineering building to "escape from the accountants and attorneys." His standing advice: "Always hire people that are smarter than you are … You have to let people create."
A
Conventional Wisdom vs. the Adobe Founders
How Warnock and Geschke departed from the startup norms of their era and ours.
| Conventional wisdom | Warnock and Geschke |
|---|---|
| Startups require 80-hour weeks. | Work smart, not long; the money takes care of itself. |
| The CEO's office belongs with finance and legal. | Put the CEO's desk in the engineering building. |
| Learn to code as early as possible. | Get a liberal education first, then the computer. |
| Accept the acquisition offer. | Sell a minority stake at a premium and keep control. |
| Charge for every user. | Give away the reader; charge for the writer. |
| Standards come from committees. | Standards come from shipping and competing. |
Tactic: Audit each policy that governs your engineers' time by asking whether it makes your best person more or less likely to come back tomorrow, and remove any policy whose only defense is that other companies have it.
Free playbook
Get The Business Model Playbook
58 business models, one visual page each: how the money flows, the metrics that matter, and who runs it. Free when you join the Faster Than Normal email.
Free. No spam. Unsubscribe anytime.
Part IIIQuotes and Maxims
In their words
— John Warnock, to Steve Jobs's demand for a faster demo, as recounted to Knowledge at WhartonSteve loved the tax form and wanted to use it on stage, but he said, "John, I can talk — but I can't talk for two minutes and 30 seconds waiting for something to print."
— John Warnock, Los Angeles Times, 1996We spent two years of our lives going to meetings. Chuck and I were totally frustrated. One day, I went to Chuck's office and said, 'We can't continue to live in the world's best sandbox.'
— John Warnock, Programmers at Work, 1986My saving grace in life is that I was not introduced to computers at an early age.
— John Warnock, Computer History Museum Fellow citationIn building a high-tech business, it has always been my basic instinct to out-invent and out-engineer the competition. After all, customer satisfaction ultimately depends on the quality of the products.
— John Warnock, University of Utah commencement address, April 30, 2020The rest of your life is not a spectator sport. Your job in life is to be an active player, to make the world a better place.
Maxims
- Avoid the problem by dividing it. Split any hard problem until each piece is simple enough to solve by rote, then look at what remains.
- Describe the thing, not the machine. Encode your core asset in a form that survives every device that will ever read it.
- Internal adoption plus external refusal means leave. When your institution uses your idea but won't sell it, the opportunity belongs to you.
- Turn suitors into financiers. A premium minority stake with prepaid revenue beats a quick sale.
- Build for one expert you can watch. The most demanding user within reach is better market research than any survey.
- Name the category, not just the product. Stalled technology often needs vocabulary more than engineering.
- Free the readers; charge the writers. Give away whatever makes the network larger and sell whatever the network makes more valuable.
- Secrecy invites rivals. Any part of a standard you hoard is an invitation for a large partner to build a replacement.
- Tie conviction to a crossover. Defend an unprofitable bet only with a named customer, a named saving, and a named condition for success.
- Your taste is for your products, not your markets. Refusing to serve an ugly platform is how elegant companies get left behind.
