“This book is precious. It makes a real contribution to our struggle for liberty and democracy in Vietnam.”
"I was deeply impressed by this book."
—John le Carré
"A disturbing and totally compelling account."
"In Reinventing the Future we get close to the minds of eleven restless individuals who do not dot the 'i's in sentences written by others."
—James D. Watson
After seven years of blue-sky exploration, Interval Research Corporation - the Palo Alto, California, think tank financed by Microsoft cofounder Paul Allen - is coming in for a landing. Open-ended research in information technology is the only life it has ever known, but now the lab is leaving behind the thin air of advanced ideas to work on creating marketable products. As Interval scientist Bill Lynch earnestly put it recently, "We're moving from the 'Let a thousand flowers bloom' stage to being more tightly focused on commercializing our technology."
Interval's earthward trajectory actually began some time ago, when the information age took a terrestrial dive of its own. In the early '90s, everyone thought the wonders of digital broadband entertainment would beam down to us from the heavens. This is how Starwave, Paul Allen's online news and sports producer now owned by Infoseek, got its name. While satellite was the belle of the ball, cable TV was the ugly duckling of telecommunications, reserved for cultural bottom-feeders and telephone linemen. But in 1998, Allen, for one, changed his mind. In a few short months, he spent more than $20 billion buying his future flagship, Charter Communications, and other cable companies.
AT&T and Time Warner quickly followed his lead, fueling a multibillion-dollar rush to deliver broadband via cable. In October, Allen's investment company, Vulcan Ventures, made another bet on cable, taking a $1.7 billion stake in RCN, a telecom newcomer with its telephone, Internet, and TV fiber-optic cables planted firmly in the ground. Now the fourth-largest cable operator in the United States, Allen boasts more than 6 million subscribers, and the party is still open to anyone with a spare $3.5 billion - which is what he expects to raise from a public offering of Charter stock this year.
Freshly endowed with sister companies and a fee-paying audience, Interval is being summoned, after an adventuresome adolescence, to pull its weight in the family business. Its new mandate is surprising to some and inevitable to others: to develop products for Allen's cable empire. Either way, the priorities have suddenly become clear. "Once you have a fat digital pipe into the home," explains Interval researcher Pierre St-Hilaire, "you have to create value for the people who buy your pipe." This shift in direction rocked the lab all the way to the top in mid-September, when David Liddle, Interval's founding director and CEO, stepped down, leaving the top slot empty and making room for two new copresidents, Arati Prabhakar and Doug Solomon. In a three-sentence statement his press office released, Paul Allen acknowledged Interval's "shift in focus" from pure research to product development. He vowed to continue his "investment" in the lab and announced that Liddle would continue serving as chair and adviser to the company.
The Interval team members held their breath through project reviews in October and are waiting to see what the "new" Interval will look like when next year's budget is finalized in December. Change is in the air, and not everyone's embracing it. Said one researcher, "If our project gets starved, we'll bail."
For four days, the secret lab opened its doors to me. "All these years I haven't been able to speak to anybody," said the project leader of "Fantasma."
The question is, Where to? The tale of Interval's past and its impending future goes to the heart of the debate over the value of pure - or at least purer than commercial - research. "When Interval began, we just did cool things," says Michael Naimark, a video artist who's worked there since 1991. "It was 100 percent research, 0 percent development." Interval came to be revered as perhaps the sole surviving link to the great industrial research facilities of yore - the labs at IBM, AT&T, and Xerox PARC, which themselves have become increasingly commercial and product-driven. Liddle, a veteran of PARC, strove to take the hard lessons of that famous institution - which managed to foster brilliant ideas but not profit from them - and reinvent the form. An unusual hybrid between an industrial-research lab and a venture capital fund, Interval was conceived to live off the proceeds of its ideas. According to the original vision, this perpetual-motion machine would be primed to sustain itself by 2002. It would seek out commercial applications without sacrificing creative leaps. Liddle has described this hybrid model as "a PARC without a Xerox."
If you do a back-of-the-envelope calculation of what it costs to keep a vested scientist employed in Silicon Valley - about $200,000 a year - it would appear that Allen's original $100 million commitment to Interval has doubled by now (not including the money he spent last year to buy the Page Mill Road office complex that houses the lab). So far, neither he nor the public has much to show for it: some art installations and videos, a touring tent full of computer games, a musical "stick" played by Laurie Anderson, and five spinoff Valley startups. Some of this stuff looks loopy, and none of it - so far at least - has made much money. It didn't help matters when, a couple of weeks after Allen's announcement, the game company ePlanet became the third of the spinoffs to go belly-up.
Liddle emphatically supports the new direction, at least in public, and we'll probably never know whether his original idea would have succeeded. Refusing to speculate, he says he prefers to keep his mind focused on the future. "We've done enough with games," he says. "We've done enough with music. Now we're trying broadband and cable - where suddenly we have a chance to get in."
With 116 scientists and 54 staff members, Interval is not the largest research laboratory in Silicon Valley, where both Xerox and IBM have a major presence. But Interval has always had a special buzz and a collection of talent that, even in the annals of technological genius, stands out.
There's Lee Felsenstein, moderator of the Homebrew Computer Club, and Rob Shaw, cocreator of chaos theory, and Max Mathews, the first person to make a computer play a tune. (Stanley Kubrick paid tribute to this musical milestone in 2001, when the dying HAL breaks into song.) There's Joy Mountford, who designed high-speed tactical-fighter displays for the F-16 before heading up Apple's GUI group. And there's Jim Boyden, the distinguished-looking man who directs the company's audio research. Liddle refers to Boyden as one of Interval's Jedi - senior research scientists who mentor junior colleagues. There are great inventors from PARC, Atari, and Apple, and there are a host of younger researchers plucked from Stanford and the MIT Media Lab. These people are famous - at least in the Valley - for inventing or developing key aspects of the PC revolution. In Boyden's case, it was the inkjet printer.
From its founding, Interval claimed the leadership role in designing next-generation technology. Paul Allen had struck it rich with Microsoft; Interval would get him up and riding the next big wave.
The lab has also been resolutely private. On the day it opened its doors, it closed them, wrapping itself in a cloud of secrecy. Even outside scientists who collaborate with Interval's researchers typically remain in the dark. "I've been visiting Interval since it opened," says Jim Crutchfield, a physicist at the Santa Fe Institute who has worked with Rob Shaw, "and I still have no idea what it does."
In the midst of adjusting its altitude and direction, Interval opened its doors to me for four days. I entered the sanctum to discover old-style Interval projects cohabiting with a series of new schemes responding to the needs of Allen's other holdings, in particular Charter Communications. Instead of games, musical instruments, and location-based entertainment - arcade games and theme parks - these projects aim to exploit the new generation of digital television, interactive video, and high-speed online connections that cable will bring to our houses. If America's corporate giants are right, this is the next "wow" technology headed for explosive growth. With companies like Charter pumping data into our homes, the crude TV of today is slated to morph into a technology capable of delivering three-dimensional and high-definition images. Some of the ongoing projects, such as basic research into the human auditory system, fall under what copresident Arati Prabhakar calls "unbounded research," as opposed to the "fairly standard corporate R&D - usually D instead of R - embedded in a cash-generating business. But we're moving," she says, "a couple of steps toward the latter end of the spectrum."
Located in a complex of low-slung buildings with red tile roofs on Page Mill Road, the lab sits at the edge of the Stanford University campus. A vine-covered trellis arches over the front door, and the tastefully decorated offices look out over a garden landscaped with fountains and waterfalls. The only visible wrinkles in this corporate paradise are big retrofit beams that reinforce hallways buckled in the 1989 earthquake, and a shortage of space. The latter problem should be resolved in the next few months, when Interval moves to a bigger building in its office campus. (Earthquakes, for now, are still beyond Interval's control.)
The tale of Interval's past and its impending future goes right to the heart of the debate over the value of pure research.
A group of researchers invites me to join a meeting. They are tackling two related projects, Holodrome and Sashimi. All of Interval's research projects have similarly whimsical, opaque names. Holodrome and Sashimi are both concerned with building cameras to help Interval fulfill its quest to reinvent TV.
The Holodrome camera looks outward, capturing a 360-degree panorama at such remarkably high resolution that each part of the image is itself nearly the equivalent of an HDTV image. If television this rich could be broadcast, instead of using your zapper to change channels, you could use it to zoom in on the action or fly around a scene from any angle.
Paul Allen has suggested using this technology to broadcast theater-in-the-round. Another application would be to build it into one of Allen's sports stadiums (he owns the Portland Trail Blazers and the Seattle Seahawks). This would allow a viewer to focus on the action anywhere in the stadium, thereby creating an infinite number of crane shots and close-ups. No longer will people be watching the ABC version of Monday Night Football; they will be making their own.
"It's the old dream of telepresence or virtual reality; you get transported into a scene, which you want to be as realistic as possible," says Pierre St-Hilaire, the French Canadian who leads the project. St-Hilaire, who came to Interval from the Media Lab, is a young man of infectious enthusiasm. He built his camera from scratch; many of the finer pieces were lovingly assembled on a workbench next to his desk. "If I had known how big a job it was going to be, I probably never would have started," he says with a Gallic shrug.
The Sashimi camera looks inward. When it's completed, it will actually be more than a hundred cameras, which capture an array of images. This could lead to holographic television, where objects appear on the screen in three dimensions. As you walk "around" them - again, maneuvering via a new kind of zapper - the sparkling diamond rings on the Home Shopping Network will beckon irresistibly. This camera is being built by Gavin Miller, an English scientist who did graphics algorithms and animation at Apple for eight years. Miller talks about some of the new methods he has found for "putting a robust vision system behind every camera." Then, as the meeting winds down, he cracks, "When our projects are finished, do you know what Pierre and I are going to do with our two cameras? Point them at each other and take each other's picture."
Elsewhere, an Interval scientist named Malcolm Slaney pads around his office barefoot, working on other ways to enhance TV. Collaborating with fellow researcher Michele Covell, who's also Slaney's wife, and Christoph Bregler, now a Stanford professor, who arrived at Interval four years ago as a summer intern, Slaney has worked on audio morphing (seamlessly splicing one sound into another), on the cocktail-party effect (tuning in to a single voice in a crowded room), and, most recently, on video rewrite, which animates human faces in videos and movies by driving them with speech. With a lot of number crunching, video rewrite can rearrange all the pixels or dots in a moving image according to the words a character is speaking: Roberto Benigni's lips, for example, could be in synch when Life Is Beautiful is dubbed from Italian to English. Or CBS could broadcast a synthetic Dan Rather, a simulacrum who reads the news 24 hours a day - even while the real one is home in bed.
Then there's Tom Ngo, who is working on a project called Centaur. A crisp, methodical man, Ngo is tackling the great beast of broadband: video-on-demand. Another Centaurian, Bill Lynch, has also worked on a new kind of video compression. Lynch's technology has one killer application: It's symmetrical. As MPEG and other forms of video compression work today, decompression is done cheaply, on computers or set-top boxes, but compression on a commercial scale happens in big, centralized facilities filled with expensive machines. What if you could do both sides of the equation - the compression and decompression - at home on inexpensive equipment? Lynch says that, using just a small part of one chip, he has cracked the problem.
One of the more impressive endeavors preoccupying the Interval brainiacs is not, at the moment, on Page Mill Road. Down in Palo Alto's commercial district - near Fry's Electronics, the local outlet for everything from motherboards to Furbys - the company converted an old house into an Interval skunk works. The house has been gutted to make a big room full of network analyzers, spectrum analyzers, digitizer scopes, and signal generators, each one costing $30,000 to $50,000.
Here, a half-dozen people are working on a project called Fantasma, Italian for "ghost." Project leader Roberto Aiello, a particle physicist from Trieste, explains that he and his team are building a wireless home network for connecting all our soon-to-be-installed digital appliances, including audio, video, TV, computer, and telephone.
Aiello begins talking, then interrupts himself: "All these years I haven't been able to speak to anybody. How do I explain?"
Endowed with sister companies and a fee-paying audience, Interval is being summoned to pull its weight in the family business.
He tries again. Fantasma, Aiello says, is meant to take all the datastreams coming into our homes through broadband cable and distribute them. Aiello's wireless network is based on transmitting data as ultra-wideband radio signals (UWB), a kind of digital broadcasting that uses pulses instead of conventional radio's modulated waves. The rapid-fire pulses, which mimic the firing of 1s and 0s in a microprocessor, can pack tremendous amounts of information into their signals. Aiello knows all about these pulses because they resemble what physicists look for in atom-smashing accelerators. On one side of the room, a system demo runs computers that open files in computers along the opposite wall and pipe music from a CD player to remote speakers. "It's some kind of Italian rock music," says Aiello. "I don't know; it's my wife's."
Central to building a wireless home network are the antennas that transmit and receive its signals. In this case, Aiello's antennas are hidden under paper cups. I am not at liberty to reveal the size of the paper cups. "You might call them 'a commercially interesting size,'" one researcher suggests. Aiello kindly lifts one of these cups to allow me a quick peek. Underneath, I find an inspired example of the black art of antenna design.
The only problem with Aiello's home network is that it's illegal to operate, at least outside a research laboratory. (The US military has been using similar technology for 30 years in radar and secure forms of communication.) Aiello is broadcasting way down in the microwatt range, which is lower than the unintentional radiation your TV set gives off. But UWB transmits in a part of the spectrum controlled by the FCC, so Aiello finds himself in a catch-22.
"It is OK to broadcast here by accident, but it's not OK if it's intentional," he says, throwing up his hands in dismay. Interval has submitted testimony attempting to show there's no danger of serious interference, and the FCC is now holding hearings on whether it should allow civilian use of UWB signals. Company copresident Prabhakar intends to push Fantasma hard. "I'm quite bullish," she says, "that the regulatory struggle will be overcome."
Eventually, if all goes according to plan and the regulators give a green light, Fantasma will move into "advanced development," joining another five projects in this stage back on Page Mill Road. Here, researchers work on burnishing their fastballs before heading out to search for venture capital or launch themselves as startups. A dozen people - market researchers and business staff - coach the scientists, helping get the best pitch schematized into a business model.
Out the other end of this process come Interval startups like Avio Digital. Unlike Interval's other ventures, most of which focused on games, Avio is directed dead center at home broadband. Given the lab's new mission, this should be the harbinger of Interval businesses to come.
The idea behind Avio is simple. A fabless chip company - one that designs microprocessors produced in someone else's billion-dollar factory - Avio has created a chipset for a home network called MediaWire. This efficient multimedia system can distribute all the video, audio, computer, telephone, and other digital feeds the "local" cable company will soon be pumping into our electronic cottages. This media network for the home is synchronous - governed by one clock - making it free of jitter and other phase problems that affect packetized data. The network is capacious and cheap - and, best of all, it runs off telephone wire, either the stuff already installed in the walls or a slightly better grade of wire, which optimizes performance but is still dirt cheap.
To show this broadband system in action, Avio has turned its San Carlos headquarters into a mock digital home. A television plays video clips in the "living room." John Lee Hooker, at higher-than-CD quality, is thumping in the "kitchen," while out in the "family room" the kids are listening to Larry Mullen and Adam Clayton's funky version of the theme from Mission: Impossible. An intranet built around Avio chips and some telephone wire pumps these audio and video streams from one room to another, while simultaneously answering the telephone, downloading email, and scanning the front steps to see who's ringing the video-equipped doorbell.
Avio's first-generation system handles 100 megabits per second, 10 times more than a standard corporate Ethernet. It can connect a hundred devices over 2½ miles of wire. It has enough capacity to distribute 8 video streams, 32 higher-than-CD-quality audio streams, 16 telephone or ISDN lines, and 12 Mbytes of computer data. In case you don't rush out and buy all the new digital goodies as soon as they hit the market, Avio's system is engineered with enough adaptability to keep your old analog components working, too.
While not yet ready to reveal their names, some big players in the computer industry and in consumer electronics have signaled their willingness to embrace Avio's technology. If it prevails as the standard - and this is a big if - a standard MediaWire connector, like a telephone jack, will go into all the new TVs and stereos we buy, and a single wire will carry their signals around our homes. Avio, which hired its new president from Philips, is trying to convince a critical mass of people in the Netherlands, Japan, and Korea that Avio's one-plug, one-wire solution is the way to go.
Four of Avio's founders - Bob Hoover, Tim Ryan, Roger Meike, and Keith Crosley - gather in a conference room to explain how the company got started. "We're technogeeks," says Meike, who helped design the International Space Station's networking system. Tim Ryan spent a summer building the control system in Bill Gates' house, and Bob Hoover is a synth geek who has built electronic music and video systems. The fifth founder, Glenn Edens, who invented the laptop computer (he patented the clamshell design), is now vice president of broadband technology at AT&T. "I like to innovate and move on to something new," says Edens.
Avio began as an Interval project to imagine what the digital home of the future would look like. "We built a little cartoon house out of foam core and moved into it," says Hoover. "It was George Jetson's house of the future."
The team drew pictures of all the interesting things they wanted in their house, then built them out of foam core and eventually created the devices themselves. "We were surrounded by all this neat stuff when we realized something was missing," says Meike. "It was the enabling technology that would pull it all together and allow the pieces to talk to each other.
"In order for the TV to shut up when the phone is ringing and let you know who is calling, the TV needs to know about phone calls. How do you get this information into a TV set? Well, you connect it to the phone cable. It slowly dawned on us that the ability to get information around to all of our devices was the underlying problem."
Inside the skunk works is a high-power microwatt home network - one illegal outside a research lab.
"We wanted a network so simple and fundamental that it's like the power grid," says Hoover. "Wherever there's an outlet in the wall, you just assume that what comes out of it is power you can use. In the future, wherever there's a MediaWire outlet in the wall, you will just assume that what comes out of it is data."
Already, there are competing technologies for wiring home networks. Systems produced by Tut and Epigram, for example, are being promoted by the members of the Home Phoneline Network Alliance. This packet-switching approach to moving data around the house on telephone wires is good for sharing files and modems, but it runs at just 1 Mbps. The next generation of HomePNA, recently released, still chuffs along at 10 Mbps - far short of what's required to pipe media around a house.
FireWire - or IEEE 1394, its official name - is another kind of home network created by Apple and embraced by Sony, which has its own brand name for 1394 connections, i.Link. This computer bus zips along at 400 Mbps. But IEEE 1394's fiber-optic cable is expensive and fragile across distances.
With outfits like Sony and IBM (the latter is researching similar technology in a project called Pervasive Computing) engaged in the battle for the home, and with a slew of companies pushing wireless solutions, Avio is hardly a shoo-in. But MediaWire is doing its best to stand clear of the fray by positioning itself as the backbone of home media networks, with enough flexibility to run all the analog, HomePNA, IEEE 1394, and wireless devices you might want to attach to this backbone. The first commercial MediaWire products will appear next year, but it will take another year or two before consumer electronics and computers are happily married and living together in everything-digital households.
"This is my second Ethernet," says Liddle, referring to his glory days 25 years ago at Xerox PARC, where he helped develop the key technologies that run today's computer networks. "Local area networks were a new idea back then. Home media networks are a new idea now. Without any prior research, we've made an out-of-the-blue invention that solves the problem from scratch."
Despite Liddle's optimism, Interval must concede that Avio is one of only two Interval startups still standing. The other, Zowie Intertainment, is fully rooted in Interval's past. Zowie, which has just released its first products, is a toy company with interesting technology but little relevance for broadband. In Redbeard's Pirate Quest, a model pirate ship loaded with electromagnetic sensors takes the place of the computer keyboard. Behind it, on the screen, appear pirate-infested waters, looming rocks, and peaceful lagoons through which Redbeard and his cronies navigate. The view changes as players move action figures, which are loaded with sensors, around the deck, or when they turn a spyglass atop the crow's nest. The company also just released Ellie's Enchanted Garden, an electronic play space for girls.
Zowie, it must be said, is faring better than Interval's other startups. In two earlier attempts, both launched in November 1996, Interval spun off Carnelian, which produced Web publishing technologies, and Purple Moon, which made Rockett's New School and other CD-ROM games for girls. Although Carnelian disappeared quietly (Interval reabsorbed the technology), Purple Moon's demise in February 1999 hit the lab like a seismic blast. (The spinoff was later bought out of bankruptcy by Mattel, which plans to ship a new Rockett product by December.)
Run by researcher and software designer Brenda Laurel, Purple Moon aimed to create computer games that appealed to girls. However, the company foundered on the reality of what it takes to sell games to kids in today's market. Says Interval market researcher Bonnie Johnson, "If we had researched business models as thoroughly as we researched girls, I don't think we would have created a CD-ROM business."
"We were really discouraged by the closing of Purple Moon," says Bill Verplank, who researches human-computer interaction at Interval. "It was a critical point in the organization's life."
Liddle is forthright about what Allen and Interval have learned from their failures: "No more music, no more games." As for the shift to broadband, "We just blocked off what have turned out to be some dead ends."
Interval's history can be divided into three periods: In the early days, everything in Paul Allen's wired world was open to exploration. Then in the middle years, the lab went its own way and they saw less of Allen in Silicon Valley, except when he came down to play his guitar at the summer picnic. Now there's this new phase, coinciding with Allen's push into the cable industry. "For the first time," Bonnie Johnson says, "he has articulated broadly what he'd like to hear from us."
Johnson admits that some of Interval's researchers are troubled by this new phase in Allen's life. But she thinks these qualms will pass when people understand that Allen is less interested in cable as it now exists than in its expanded, broadband form, which includes pumping "art and fun things over the wires." In May, she herself told me, "I don't like television, and I don't subscribe to cable." Since then, she has decked out her home with cable connections throughout.
"There was never a falling-out," says Liddle friend Bob Metcalfe. "Development just wasn't what interested David."
One of the worriers is Michael Naimark, who got his start in the 1970s as a filmmaker working with Nicholas Negroponte on a filmed map financed by Darpa. It charted the city of Aspen, Colorado, with such fidelity that a prospective spy, who could steer through the map at will, would end up knowing the city inside and out without ever having been there. Filming in stereo and creating immersive environments of breathtaking verisimilitude, Naimark has been making video maps ever since. Now he wonders whether Interval still welcomes his art. In the meantime he has chosen to switch from stereoscopic movie cameras to webcams, a medium better suited to the lab's new direction.
Another researcher expresses similar worries anonymously. "There is a social contract here," he says. "We don't make as much money as we could working for a startup company. Nor do we get public recognition because our research is kept secret. In exchange, we get to work on stuff really at the edge. That's why we're here. This is a watershed moment for a lot of us. We wonder whether the new management understands this social contract."
David Liddle gets impatient with such remarks. "It's revisionist history," he snorts. "The truth is that we've always been an archly commercial place." Coming from the opposite direction, copresident Prabhakar avers that Interval won't be entirely cable-centered. "We'll intentionally continue to support some deep, long technology research that won't be realized until several years out."
Researcher Lloyd Watts, for example, continues to study the neurobiology of human hearing, learning how the ears and brain process sound and then implementing these computations in hardware. In an office humming with computers, he gives a demo of how a little camera outfitted with "ears" will turn toward you if you clap your hands. Next he must get it tracking not just loudness but the frequency of a particular voice or sound. "This is the basis for a powerful multisensoring system, a computer that can see me and hear me," he says.
How can the changes at Interval best be measured? Perhaps by watching Liddle himself, who might be finding the place less welcoming than before. "David has wanted to move on for some time, ever since it was clear that Interval was going to do less research and more development," says Bob Metcalfe. A close friend of Liddle's, Metcalfe worked for him at PARC, where he invented Ethernet before leaving to make his fortune at 3Com. "Development," Metcalfe says, "just wasn't what interested David."
Whatever Liddle decides to do, he will remain one of Silicon Valley's senior statesmen. A big man at 6' 4", with a bald head and broad forehead, silver-framed aviator glasses, blue wide-beam eyes, and a white, close-trimmed beard, Liddle is a forceful writer and charismatic leader who sits on numerous boards and sometimes teaches at Stanford. Speaking with his hands, which are long-fingered and quick, he cuts the air with decisive sweeps or frames it like a basketball player ready for a layup. In fact, Liddle briefly played basketball at the University of Michigan, and if he had been a couple of inches taller, he says, he would have stuck with sports instead of going on to graduate school in computer science at the University of Toledo. "Of course, by now I'd be selling insurance somewhere," he jokes.
Liddle joined PARC in 1972. Soon he was running the Systems Development Division, formed to sell the Star, Xerox's first commercial workstation. Xerox introduced the first GUI with icons, the desktop metaphor, dialog boxes, object-oriented programming, the laser printer, and the Ethernet LAN - all the great inventions that made Apple, Microsoft, Adobe, 3Com, and other competitors fabulously wealthy. Why Xerox let its technology walk out the door is a long story. Suffice it to say that, for Liddle, the experience left a lasting impression; at Interval, things would be done differently. In 1982, Liddle joined the exodus from PARC and, with Don Massaro, his former boss, founded Metaphor Computer. They made a workstation like the Star but faster, and sold a lot of machines and software until IBM bought them out in 1991 for $122 million. Liddle, with some money in his pocket, was wondering what to do next when he got a breakfast invitation from Paul Allen.
"Allen and Liddle are Viking buddies," says Interval scientist Bill Verplank, who also worked on the Star. The observation echoes a theme. Allen calls the company that directs his investments Vulcan Ventures, after the Roman god of fire. An early suggestion for the research center's name was Valhalla, but someone pointed out that this was where slain Norse heroes went to fight their battles in perpetuity. They then settled on the name Interval, referring to the space between, or the interregnum, dividing the old order from the new world yet to be born - a process Allen and Liddle originally thought would take a decade. To bridge this interval, Allen planned to fund the lab for a decade. The years 1992 to 2002 were even printed on company name-tags. Today, the end date has been removed.
Joining forces in '91, Allen and Liddle began staffing Interval with an all-star team of Silicon Valley players. It was Liddle's inspiration to organize Interval's brainpower around projects, with everyone expected to work on two or three of them at once. Today, researchers are still scattered throughout the building randomly, and everyone is encouraged to work together. This is accomplished by dividing time into points and giving everyone 20 to spend. A project will be budgeted for so many points, and a project leader will recruit fellow researchers by signing up teammates for primary (14 points), secondary (6 points), or lesser (3 points) commitments. "If we took our signal-computation people and put them all together, they'd start making 10 percent improvements in things," says Liddle. "But if we match them one at a time with a mechanical engineer and a videographer, they'll do great new things."
Borrowing his organizational model from God, Liddle started Interval with seven projects. When it grew bigger, he grouped these projects into seven "areas." Bonnie Johnson, his third hire, describes the founding: "We spent most of our days in meetings thinking about what we would do, and then one day David said, 'There shall be seven.'"
"When Interval grew to over a hundred researchers, David took the seven gray-haired staffers and said, 'Ye shall be area chairs,'" adds Johnson, who is one of these chairs (although she is blond). Interval's seven fields of research, currently scattered throughout areas codenamed Alpha, Bravo, Charlie, and so on, included computer graphics and image processing, new computer design, signal processing, audio research and wearable computing, human-computer interaction, market research, and electronic assembly in Interval's shops.
That was then. And now?
Allen planned to fund the lab for a decade: "1992 to 2002" was printed on company name-tags. Today, the end date has been removed.
It did not go unnoticed that in Paul Allen's terse announcement of the shift in direction, there were no words of praise or support for his old friend. No thanks for years of loyal service. "It's a golden-parachute statement, a classic phaseout," says a Silicon Valley press agent with long experience in reading corporate tea leaves.
Liddle insists that giving up daily operation of the lab signals nothing about his longtime friendship with Allen. "Every relationship has its ups and downs," he says. "We're not alienated."
For his part, Allen chose not to comment on the state of the friendship (or on the future of Interval), but perhaps Bob Metcalfe describes it best: "There was never a falling-out between Paul and David - although, on second thought, maybe a falling-out is exactly what it is, in the literal sense of falling out of step. Paul wants to go one way, David another."
"This may be the end of the golden age at Interval," says a friend of Liddle's. "But David is an incredibly loyal and persistent guy who will stick with them through the transition."
Whatever happens to their relationship, Allen's commitment to Interval appears, if anything, to be stronger than before. "It's an increasingly important part of our portfolio," says Bill Savoy, president of Allen's Vulcan Ventures. "Now that we have to find the breakthrough technologies that will make cable valuable to consumers, a lot of what Interval is doing could be a huge windfall for us."
Metcalfe, for one, agrees: "As Paul's empire grows, the leverage of the Interval brain trust only increases. This is an extremely valuable asset David has given his friend."
To complement its researchers, the lab is beefing up its business team. Prabhakar's copresident, Doug Solomon, joined Interval in October 1998 as VP of development. In a 15-year career at Apple, Solomon had moved from managing educational products - such as Logo, an early programming language for children - to directing strategic planning. His assignment at Interval is to assemble a team of MBAs and bundle the lab's ideas into startup companies and shippable products.
Working directly with Allen's companies on joint projects, or giving his cable subscribers first dibs on Interval technology, the lab is supposed to feed its ideas, inventions, and breakthroughs into our homes across the Paul Allen universe. Does this make Charter Communications the Xerox that Liddle promised would never interfere with his PARC? Is Interval slipping into the kind of corporate research driven by a parent business and its quarterly reports?
Perhaps not. Perhaps Interval, steering a trickier course, will pull off a cultural coup: Amid the Valley's knee-deep capital and high-speed innovation, it will manage to fly high with its feet on the ground.