In September of 1994, Sony began selling a handheld device that was, by any honest measure, years ahead of everything around it. The Magic Link had a touchscreen, downloadable applications, email, animated emojis, and a graphical interface that let users navigate through virtual rooms by tapping familiar objects. It cost roughly eight hundred dollars, weighed just over a pound, ran on a sixteen-megahertz processor, and connected through a dial-up modem slower than nearly anything available at the time. Fewer than four thousand people bought it.
The story of how that device came to exist, and why it failed, begins in 1989 inside Apple Computer’s Advanced Technology Group in Cupertino, where a forty-two-year-old researcher named Mark Porat wrote a paper describing a machine no one had ever seen. Porat called it the Pocket Crystal. It would be a tiny computer, a phone, a very personal object that knew your name and your schedule, sent your messages, and managed your money. It would fit in your hand, and it must be beautiful.
In 1989 there was no World Wide Web, cellular phones weighed two pounds and cost three thousand dollars, and the average American had never sent an email. Porat was not guessing. A Stanford PhD who had spent a decade studying the economics of information, he saw computing as inevitably leaving the desk, entering the pocket, and merging with the telephone. He presented the paper to Apple’s CEO, John Sculley, and asked for something unusual.
Not funding, but permission to leave Apple, take some of the finest engineers the Macintosh project had produced, and build the Pocket Crystal as an independent company. Sculley agreed. In May of 1990, General Magic was born. It was staffed with the engineers who had built the Macintosh, and backed by the largest alliance of technology and telecommunications companies ever assembled for a single consumer product.
Within four years they would ship a handheld device with features that would not reach the mass market for another thirteen years. Porat had not come to technology through engineering. His 1976 Stanford dissertation introduced the concept of the information economy to American policy, arguing that more than half of the nation’s gross national product already came from the production, processing, and distribution of information rather than physical goods. He arrived at Apple in 1988 after years at the Aspen Institute and Booz Allen Hamilton advising on the breakup of the telephone monopolies.
He had spent twelve years studying the future of information without getting his hands on the hardware. Now he would. The people who would build it were already inside Apple. Andy Hertzfeld had written the core system software for the original Macintosh.
Bill Atkinson had created MacPaint and HyperCard. Susan Kare had drawn the icons that gave the Macintosh its face. Joanna Hoffman had sold the machine to a skeptical world. By the late 1980s they were restless, and when Porat circulated his paper, they recognized the same future he did.
But Apple’s revenue came from selling Macintosh computers to businesses and schools, and a pocket device that might cannibalize the desktop was a risk the board had no appetite for. Sculley let the team walk out the door, keeping a stake in the venture and contributing the talent. General Magic set up in Mountain View in an office where there were no titles, no dress code, and no performance reviews. One engineer brought his pet dwarf rabbit, which became the company’s unofficial mascot and wandered through conference rooms chewing cables while the team tried to invent the next century of computing.
The mission statement read like a love letter to a device that would not be commercially viable for nearly two decades: a dream of improving the lives of many millions of people by means of small, intimate life systems that people carry with them everywhere. Porat challenged his team to build something that, once used, a person could not live without. The talent he assembled was staggering. Tony Fadell arrived in 1992 at age twenty-three as a diagnostics engineer.
Andy Rubin joined the same year as a lead engineer. Pierre Omidyar worked in developer support. Kevin Lynch was writing software. Megan Smith was engineering systems.
Darren Adler, who had worked on the Macintosh Finder at Apple, came aboard as chief software architect. None of them were famous yet. All of them would be. The funding matched the ambition.
Porat built what he called the General Magic Alliance, a consortium including Sony, Motorola, AT&T, Matsushita, Philips, and Apple, each investing up to six million dollars for the right to build hardware that ran General Magic software. By 1992 the alliance had expanded to sixteen companies, and total capital raised would eventually reach two hundred million dollars. No startup in Silicon Valley had ever launched with this kind of corporate backing. The mission was to create two things.
First, an operating system called Magic Cap, a graphical interface designed for a handheld touchscreen where users navigated through virtual rooms, tapping on familiar objects to perform tasks no handheld device had ever attempted. Second, a programming language called Telescript, designed by chief scientist James White, which allowed small programs called agents to leave a device, travel across a network to a remote server, perform a task autonomously, and return with results. The original plan called for a nine-month development cycle, with shipment by the end of 1991. That timeline collapsed immediately, because nearly nothing the team needed existed.
There were no small touchscreens on the market, so engineers had to invent their own custom touch controller. There was no low-power processor suitable for a handheld, so Motorola designed a custom chip. There was no wireless data infrastructure worth building on. The fastest wireless modem available ran at 2400 bits per second, slower than what it takes to load a single photograph on a modern phone.
Batteries drained in hours, connections dropped, the screen was not backlit, and the software crashed constantly. Tony Fadell found himself working ninety to one hundred twenty hours a week, sleeping on bunk beds the company had installed in the office rather than going home. In August of 1993, while General Magic’s team worked under the assumption that Apple was a partner, Apple launched the Newton MessagePad, a handheld personal digital assistant with a touchscreen and handwriting recognition. The Newton was John Sculley’s project, built inside Apple using concepts that had originated in the same advanced technology group where Porat had written the Pocket Crystal paper.
The General Magic engineers felt it as a betrayal. Sculley had approved the spin-off, invested in the company, retained a board seat, and then built a rival product. The alliance partners watched Apple launch a competing device and began to wonder whether their money was protected. Trust inside the alliance cracked, and the secrecy that had felt like protection now felt like isolation.
By 1993, General Magic faced a decision that would determine whether the company survived, and they made it wrong twice. Tony Fadell proposed pivoting to business users first, executives and sales people who already understood email and calendar management and would pay for a device that put those tools in their pocket. Management said no. The vision was a consumer device for everyone, and the vision would not bend.
That same year, a young intern looked at the proprietary AT&T system the team was building on and told the engineers they were missing it, that the web was where it was at. The World Wide Web was arriving. Mosaic, the first graphical web browser, launched in 1993, and within eighteen months Netscape would trigger the largest technology boom in American history. General Magic chose instead to build on AT&T’s PersonaLink, a closed proprietary network that AT&T controlled and charged for.
In a narrow technical sense, Telescript’s autonomous agents were more sophisticated than anything the early web could do. But the open internet had something the closed network never would: everyone. When the Sony Magic Link finally shipped in September of 1994, the timing could not have been worse. The American wireless landscape was primitive.
The first digital cellular networks were designed for voice calls, not rich multimedia content. A dial-up modem running at 2400 bits per second meant sending an email with an attachment could take minutes. Fewer than ten percent of American households were online at all. The buyer received a device with a four-and-a-half-inch grayscale LCD, no backlight, four levels of gray, one megabyte of RAM, and connectivity through that modem.
The Motorola Envoy, developed under Andy Rubin and launched in early 1995, was worse. It carried a radio modem, infrared networking, a touchscreen, and wireless email at a time when most telephones could not send a text message, but it cost fifteen hundred dollars plus roughly four hundred dollars a month for wireless data service. The math was unforgiving. Two hundred million dollars in private investment, sixteen of the world’s largest technology companies, and some of the most talented engineers in Silicon Valley all produced fewer than ten thousand devices sold.
The alliance collapsed in stages. AT&T shut down PersonaLink entirely in July of 1996, taking with it the backbone of General Magic’s product vision. In September of 1996, Mark Porat resigned as CEO, leaving a company that had lost more than seventy-four million dollars. By the end of that year, eighty of its roughly one hundred employees were gone.
General Magic stock, which had doubled on the day of its February 1995 IPO, fell to one dollar by 1997. The company announced its closure on September 18, 2002. What survived was not the company but the people. The engineers who had worked in that Mountain View office with the rabbits and the bunk beds scattered into the industry and built the future they had imagined too early.
Tony Fadell joined Apple in 2001 to build the iPod, and when he assembled his team he hired engineers from General Magic. He is credited as one of the fathers of the iPhone, the device Mark Porat had described in 1989. Andy Rubin, the lead engineer on the Motorola Envoy, went on to found Android Inc. , acquired by Google in 2005, and the first Android phone shipped in 2008.
Kevin Lynch left General Magic, helped create Dreamweaver at Macromedia, became chief technology officer at Adobe, and was hired by Apple to lead software development for the Apple Watch. Pierre Omidyar founded eBay. Megan Smith served as Chief Technology Officer of the United States under President Obama. General Magic did not fail to invent the smartphone.
It invented the smartphone, and then it invented the people who would build it again. The Motorola Envoy now sits in the permanent collection of the Cooper Hewitt Smithsonian Design Museum, recognized as a working first draft of the device that would become the center of modern life.