Will Hearst and Kleiner Perkins's @Home is promising to deliver Internet broadband to a million homes by the first quarter of 1997.We've heard claims like this before. What makes this one different is that legendary networking wizard Milo Medin is leading the charge.
Milo Medin has little tolerance for low bandwidth. In computer networks, in human brains. There's not much Medin can do to grease the neurons of feeble minds that just can't keep up. But he can apply his prodigious brainpower to the creation of high-powered communication networks that blast data at speeds he calls "unreasonably fast." He's taken this on as his sacred mission.
Medin, 32, already ranks as a leading light in the development of the Internet. When he was just 23, Medin led a team in building a NASA data network that ultimately grew into the huge NASA Science Internet, linking researchers in 16 countries and six continents. Later, as the aggressive champion of the TCP/IP protocol, the key standards that set the foundation for the Internet's phenomenal growth, he beat back the status quo. These and other contributions have gained Medin enormous respect. Tony Rutkowski, executive director of the Internet Society, refers to him as a living legend.
Today, Medin is engrossed in a baldly commercial endeavour to link the Internet with cable TV. Lured away from NASA by the promise of a new creative challenge - and with an equity stake that could make him rich - Medin is bringing the Internet and the World Wide Web to the masses. As vice president for net-working at @Home, a joint venture of TCI Technology Ventures Inc., a subsidiary of cable giant Tele-Communications Inc., and the venture capital firm Kleiner Perkins Caufield & Byers, his mission is straightforward: to make the trickle of data that flows into a Net surfer's computer a torrent.
Anybody stuck with a 14.4-Kbps modem knows how soporifi-cally slow Net cruising can be, especially when there are graphics, video, or sound files to download. A 28.8-Kbps modem performs better, but not much. ISDN lines, assuming you can get them where you live, can boost access speeds to 56 Kbps. But that's peanuts compared with what coaxial cable can deliver. Coax is the black wire that snakes out of your sitting room wall and into your TV set delivering 24-hour news, music videos, and a surfeit of second-rate films you'd never want to rent. Theoretically, you could splice that cable into a specialised modem, hook it into your computer or to a set-top box on your TV, and watch Net data start flying into your machine at peak speeds of 10 Mbps - up to 700 times faster than a 14.4 modem on a phone line. That's fast enough for a head-snapping delivery of graphics, animations, sound files - even films.
Medin is trying to make it happen, sooner rather than later. The speed of cable modems, Medin believes, will so alter the online experience that online will mean something new - not television, not computing, but something else, tremendously engaging and rich. "It's like evolution is pushing online into a new space," he muses.
His boss has an equally enthusiastic, if more market-driven, view of high-bandwidth Internet access via cable. William Randolph Hearst III - grandson of the media baron, partner at Kleiner Perkins, and chief economic officer of @Home - says the speed of cable modems "will make the Web and the Internet more like an entertainment medium, so that players like CNN, Hearst, Paramount, ABC, and Time Warner can begin to program for this space in a way that is different from the traditional dial-up services" such as America Online. That sounds suspiciously like the information superhighway of the early 1990s. You know, interactive TV. Remember? It would deliver films on demand and interactive services like shopping and banking from expensive centralised video servers to set-top boxes on a couch potato's TV. Expensive is the key word. Most of the planned info highway pilot projects have been delayed or scrapped, largely because the price of the hardware required to provide films on demand is still too high.
What @Home does, in effect, is hybridize the Internet and the information superhighway. By linking up with @Home, the cable companies can offer consumers a high-speed pipe to the Internet - swift enough that Hearst thinks customers will cough up US$30 (£19) a month or more for the service. @Home would split the take with any cable company that offers its service.
Most of @Home's content will pour in over the Net through a Netscape browser customized for @Home. The look and feel of @Home's gateway is now being crafted by famed magazine designer Roger Black (with plenty of space left aside for revenue-producing advertising.)
But @Home wants to be much more than a simple Internet access pipe. @Home will also add proprietary servers called replicators to its network and scatter them around the country. Net content providers could take the material that resides on their own servers and replicate it on @Home's far-flung servers, too - after paying a fee to @Home, of course.
Because the replicant servers would be connected to @Home's high-speed backbone network, and be located relatively close to the customer's home, bottlenecks that slow down the "public" Internet could be avoided. These servers could dish out programming exclusive to @Home customers - similar to the infobahn model. @Home could also rake in a percentage of any commerce transacted through its system. Even films on demand could be added when video servers become affordable. Under @Home's model, Internet service - instead of films-on-demand - becomes the main draw to get the info highway rolling.
@Home isn't the only company making such plans, although its relationship with TCI and the cable company's base of 13 million subscribers gives it a big leg up. Time Warner Cable has launched a cable modem pilot in Elmira, New York, and other trials are underway. What will differentiate @Home? Its network, for one thing. When @Home was formed in May 1995, its founders aimed to set it apart by creating the fastest, most efficient cable-modem network possible. When they started headhunting, one name kept popping up: Milo Medin.
About a year ago, some guy from a firm called Kleiner Perkins kept calling Medin's NASA office and clogging his voicemail with messages. Medin ignored him. "What do I know?" he says. "With a name like that, I thought they were a law firm." But the caller - John Doerr, a partner at Kleiner Perkins, perhaps the most highly-regarded venture capital firm in Silicon Valley - wouldn't give up. Medin grabbed the phone one day as he hurried to leave the office. Doerr was on the other end, and his pitch was persuasive. The next Saturday, Medin found himself breakfasting on wild rice pancakes at the Good Earth restaurant in Palo Alto, California.
The two laid out their business plan. The Internet, they said, was growing rapidly in 100 per cent annual bursts - even though getting access to the Internet meant tricky installation of special software and downloading graphics, video, and audio took forever. Imagine the bonanza waiting for a company - it could both simplify access and deliver all that good content at blazing speeds. Cable modems were the ticket, they said. Although optical fibre today, at roughly 2.4 Gbps, is the speed champ of data delivery, practically nobody has fibre optic strung into the home. Cable wires, however, snake past 97 per cent of all households in the US. And 65 per cent of all homes are cable subscribers. @Home had the idea and the cash. But it needed somebody to put its network together.
Medin was less than ecstatic. The overall idea, he said, was intriguing; too bad the implementation would fail. Connecting cable operators directly to the Internet was the wrong way to go. Computer geeks might put up with Internet failures and time delays, but the cable-ready public would not, not in huge numbers, anyway. Medin told the pair that @Home should create its own national high-speed data backbone network running Internet protocols by leasing or buying sections of bandwidth from providers such as MCI, AT&T, and Sprint and attach replicant servers throughout the system. Otherwise, its system would be too slow, too unreliable, too hard to expand, and too hard to manage. And thanks for the job offer, he said, but I'm happy where I am. Doerr persisted. "I had to," he says. "In two minutes, he had changed our business model. And besides that, there are only half a dozen people in the world who make the Internet work, and Milo is one of them. We absolutely needed him."
Medin lives to network. Even his office makes that plain: there are no potted plants, no family photographs. He describes himself as "just a farm boy from Fresno, California, raised on 20 acres of grapes." His father emigrated from Serbia in 1930, when it was part of Yugoslavia, to buy a vineyard and join Serbs who had settled nearby. He returned to the home country in 1960 to bring back Medin's mother through a pre-arranged marriage. When Medin was 5, his father died suddenly, putting his mother in a precarious position. But neighbours pitched in and helped her save the farm.
Medin pitched in, too. But early on, he found computers a lot more fascinating than fruit farming. By the age of 11, he used a programmable Olivetti calculator to construct a program his teachers could use to process test scores. By secondary school, he owned an Apple II and a Novation Cat 300-bps modem; after school, he'd sit in his room overlooking the vineyard and try to connect to other intelligent life through chat groups on a computer based at California State University in Fresno. Following a buddy's tip, Medin soon figured out how to dial into a remote "terminal interface processor" and log on to a much bigger network. Looking back, it was one of those cinematic moments. Medin had found his way to the now legendary Arpanet, the first iteration of the Internet, sponsored by the Department of Defence's Advanced Research Projects Agency. But he had also found something else. Unknowingly, he made his first contact with the source of his future professional success. The site of that access controller was none other than the NASA Ames Research Centre.
In 1981, Medin took up computer science at the University of California, Berkeley. Even for a Berkeley student, Medin had an especially good vantage point to see first hand the rough stuff of the Internet in the making. He worked part time at Nasa Ames and at the Lawrence Livermore National Laboratory, writing programs to simulate the effects of nuclear bomb blasts. The job at Livermore got him onto the world's most powerful computers and the lab's advanced network connections.
The Internet in those days was a primordial soup of individual networks like Arpanet, Bitnet, HEPnet, MFEnet, and SPAN, many of them running different communication protocols. There were few local area networks at the time; there was no Gopher, no WAIS, no Archie, and nothing even remotely like the World Wide Web. Outside of a small, dedicated community of scientists, researchers, and computer science students, few other people knew or cared that the elemental Internet existed. Says Medin: "Unless you knew where stuff was in the first place and then knew exactly what commands to mutter, you couldn't find squat."
But to somebody like Medin, compelled by his own nature to organise random data into structure and meaning, the Internet, or at least what there was of it, looked like the perfect place to impose a little order. The chance came his way in 1985 when he graduated from Berkeley and got a full-time network engineering job at NASA Ames, in the heart of Silicon Valley.
Within three years, Medin was leading a team of engineers to connect the local area networks at three NASA sites - Ames, the Jet Propulsion Laboratory in Pasadena, California, and the Goddard Space Flight Centre in Greenbelt, Maryland - that stored earth science data from around the world. The job seemed straightforward enough, but it soon deteriorated into a brawl over what communications standards to use. The Protocol Wars, the participants called it. Medin backed the TCP/IP protocol suite, today's Internet standard, because it wasn't restricted to a specific computer platform and because it was easy to build new applications on top of it. Medin took to pronouncing, often and loudly, that TCP/IP was the "one true way."
It didn't help his case that some people thought him a little bit odd. He had given a pet name to his computer workstation, CINC SAC, short for Commander in Chief, Strategic Air Command. "I used to kid Milo that he thought he was Dr. Strangelove," says Bill Jones, Medin's former boss at NASA. "He liked that." Medin, who has been called "the Rush Limbaugh of networking protocol," has never shied away from making his views on American foreign policy known. For a long time, he admits, he boiled down his favoured approach to dealing with America's foes to the phrase "Nuke Ôem till they glow, then shoot Ôem in the dark." Time has softened his stance a bit, he says now.
But during the Protocol Wars, Medin came off like a zealot. He was beating up the unbelievers so mercilessly that top NASA networking chief Tony Villasenor had to be called in from Washington, DC, to resolve the conflict. "Here's this young kid who seemed to have an answer for everything," Villasenor recalls. "Initially I was put off by it. Here's this guy telling his bosses what to do. But by god, he could explain why his ways were better." That was Medin's saving grace - the soundness of his ideas. "He was simply one of the loudest and most effective voices saying TCP/IP was the only way to go," says Steve Wolff, director of networking for the US National Science Foundation at the time. "And if it wasn't for that, there wouldn't even be an Internet." Not as we know it, anyway.
John Doerr's persistence after the pancake breakfast paid off. He and Hearst continued their sell job, finally persuading Medin that @Home was the place to push the Internet to the next level. The equity stake he offered Medin was a powerful incentive, but Medin insists that wasn't the real attraction. "Most of all, I wanted to build a new network, from scratch, that will change society," he says. "And how many places can you go to do that?"
Medin set about helping Doerr and Hearst develop their plan. @Home will launch its pilot early in 1996 through a TCI cable system in Sunnyvale, California, that has access to 55,000 homes. A national rollout will follow, assuming things go well. By 1997, Hearst expects to have signed up 1 million customers, enough to achieve critical mass. "You exist as an advertising-size entity when you hit a million customers," says Hearst. "I think this business has to expend its energy while growing as fast as possible and not conserving cash."
@Home must grow fast if it's to establish itself as a leader in the market. High-speed Internet access will grow from a piddling $123 million (£76 million) today to a $4.2 billion (£2.6 billion) business by 2000, according to Forrester Research in Cambridge, Massachusetts. "Everybody wants it," says Paul Sagan, senior vice president of Time Inc. New Media.
In fact, Time Warner Cable, the second largest cable company after TCI, beat @Home to the punch with a cable modem trial of its own. In July 1995, 500 sites - schools, libraries, local government offices, and homes in Elmira - were wired into the Internet through TV cable. Customers get a mix of information services, including Time Warner's Pathfinder, one of the most popular sites on the Web; national and international news; and for an extra fee, America Online and CompuServe. In addition, Time even sent two of its Path-finder editors into town to help local people produce their own highly personalized brand of community news. Customers pay $14.95 (£9.30) a month, which includes rental of a cable modem for basic services, and an additional $9.95 (£6.22) for unlimited, high-speed access to the Net. The test is expected to last six months, and if it's successful, Time has some national plans that sound more than a little like @Home's. "We want to take it to all of our cable systems," says Sagan. "And we would make the same thing available to any other cable company."
The phone companies are getting into the act, too, providing ISDN in limited markets and stringing fibre and coaxial cable of their own. Pacific Bell, which covers California, is especially aggressive. It says the market for Internet services is the largest in the US and among the largest in the world. "It's a critical time for us," says David Dorman, president and CEO of Pac Bell. "We don't intend to shrink back into a local phone company shell."
BellSouth Corp., meanwhile, is planning a cable-modem pilot for 12,000 customers. Andrew Dietz, manager of programming and content for BellSouth Interactive Media Services, the subsidiary running the test, says BellSouth was compelled to develop online access through cable modems because it was convinced local cable operators like Scripps Howard and US West would soon do the same thing - in addition to raiding BellSouth's local phone business. At the same time, Dietz says, his company will continue to offer ISDN, though unlike Pac Bell, he's not optimistic about its effectiveness as a Convergence Wars weapon. "If you're in a market competing against a cable company," he says, "and all you've got is ISDN, you'll lose out."
Eventually, fibre-optic cable could be strung into homes. Fibre is at least several hundred times faster even than coax, and also more versatile, allowing data to travel out of the home just as fast as it comes in. The vast majority of today's coaxial cable systems can send a river of data downstream, but only a cocktail straw's worth back the other way. That's OK for pure consumers, but anyone wanting to create video-rich material to send out over the Net won't find an ideal solution in cable modems. The cable companies could upgrade their networks to accommodate robust two-way "symmetrical" services, or even string fibre of their own. But it would be enormously expensive to do so.
For the time being, though, @Home's main competition is other cable-modem service providers. @Home will attempt to distinguish itself, in part by using Medin's network architecture to manage the entire system from a national control centre. In theory, the company will be able to identify and fix potential problems in a cable modem anywhere in the system before the subscriber has to call up and complain. This saves the cable companies a lot of service hassle, and given the cable industry's lousy reputation for service, it's got a good chance of pleasing customers.
But how will customers react to a Dr. Strangelove exerting central control over their communications pipeline? Competitor Sagan at Time Inc. New Media says that the need for such services hasn't been demonstrated. Even if it were, he adds, he could buy system management on a commodity basis from Sprint. "@Home has a great pitch," he says. "They've outlined something. But we're up and running with paying customers and we have the content today."
@Home's ultimate success depends on getting more cable companies than just TCI to sign up. But will those outfits want to turn over revenue to a huge competitor like TCI? Tom Nagel, director of multimedia applications for Cox Communications Inc., the US's fourth largest cable television provider, says he likes the idea that @Home offers economies of scale, quick entrance into the business, and nationwide management. But will he bite? "I think that's an option for us," he says. "We haven't ruled it out."
Medin, of course, has no doubts at all that @Home is the right answer. He has found the "one true way" once again, although his calling sometimes seems to overwhelm him. One afternoon in mid-September, he returned to his old battleground at NASA to take care of odds and ends he'd left undone since joining @Home. It started out as a fairly straightforward trip but gradually became more like a nostalgic pilgrimage. After revisiting the Telecommunications Gateway Facility, which houses MAE-West, the interconnection point for all Internet traffic on the West Coast, Medin went into the fenced area that surrounds the sides of the building. It's a stark environment with stones and gravel covering the ground; two large, looming satellite dishes and several smaller ones; and assorted pipes and metal boxes. But Medin seemed pleased. He was almost running from one spot to another, pointing out various features. "See here. This is the emergency generator. And see those big dishes - they're pointing at communication satellites 22,000 miles up in geostationary orbit. Isn't that a wonderful mission? Space research."
And then he looked into the sky as if he could see something up there, and said, "You know, I've always loved working on big missions. They make me feel that maybe I can make a difference." n
Lucien Rhodes (rhodes@newcentury.com) has written for several US publications. Webmaster of InfoWorld Electric, he lives near San Francisco.
By Lucien Rhodes
Questions about cable modems have become so frequent lately that Craig Strachman, marketing manager at cable-modem developer Hybrid Networks Inc., will take visitors on one by one in an unusual game to demonstrate the technology.
Think of it as a talent show for modems in which the first one to sing wins.
Side by side on a long table at the back of a conference room in the company's headquarters in Cupertino, California, are two 66-Mhz 486 PCs. Both PCs are connected to the Internet, both are running the Netscape browser, and both have Hybrid's homepage (www.hybrid.com/) up. After that, the two PCs take on distinctly different characteristics.
The PC on the right is all tricked out. Two modems are connected to it. One is a garden variety 14.4-Kbps telephone modem, but the other is a Hybrid cable modem bringing in a signal over the TCI cable TV system.
In comparison, the PC on the left looks scrawny. The only thing connected to it is an ordinary 14.4-Kbps telephone modem.
The game's about to start. Strachman sits at the cable-modem PC on the right and a visitor is seated at the PC on the left. On signal, each person will begin downloading a roughly 1-Mbyte audio file from the Hybrid homepage, a selection from "People Are Strange," a song by The Doors.
"OK," Strachman says. "Now."
His PC breaks away rapidly. Already, things look bad for the other machine. Strachman's cable modem is capable of delivering data to the PC at the rate of 10 Mbps (though what gets processed by the computer can be considerably less for various technical reasons). In this particular contest, Strachman's PC is probably getting a raw data-transfer rate of 2.5 Mbps which is still 275 times faster than the raw data-transfer rate of a 14.4-Kbps modem (about 9 Kbps).
The telephone modem sitting on top of the cable modem gets into the act here as well. When a Web surfer wants to send data "upstream" back to the server by, say, clicking on a homepage hot spot, that signal uses the 14.4 modem connection. This approach is based on a widely held view that most users want a flood of graphics, text, sound, and video coming at them quickly but don't need a lot of up-stream bandwidth to carry what are mostly tiny mouse clicks. When the speed and volume of downstream and upstream transmissions vary like this, the cable modem setup is called asymmetrical. The other approach, symmetrical, faces more difficult technical challenges in overcoming radio frequency interference going upstream.
It's over. Barely 30 seconds have passed, and the contest is already finished. Strachman's PC is singing: "People are strange when you're a stranger. Faces look ugly when you're alone."
The other modem will not finish its download for another 13 minutes. Strachman estimates that a home-based ISDN line carrying data at 56 Kbps would still require almost four minutes to download the same file.
In the future, Web surfers whose cable television companies have not upgraded their cables to accommodate two-way data transmission will probably use a system very much like the one attached to Strachman's PC. But if the upgrades have been made, they'll be able to sample the faster and more streamlined asymmetrical designs now under development. @Home's experts foresee a single cable attached to the cable modem that will transport traffic at nearly 30 Mbps downstream and close to 1 Mbps upstream. The telephone modem will be eliminated.
Every company in the increasingly competitive cable modem business, including Hybrid, is working on next generation modems that offer much more aggressive performance. At the same time, Strachman says, Hybrid and other manufacturers expect prices to end up around US$300 (£188) each by 1997. Currently, cable modems are being used only at test sites.
Strachman is shutting down his machine. His visitor is still struggling to articulate some kind of reaction.
"Wow," he says after a while.
"Yeah," Strachman says, "that's what they all say."
By Steve G. Steinberg
Reading Arpa's research entrails to determine the future direction of technology.
Arpa, the Advanced Research Projects Agency of the US Department of Defence, is best known for funding the devel-opment of the Internet. But its monetary support has also brought us important technologies such as RISC microprocessors and flat-panel displays. Although intended for military use, these devices have thoroughly infiltrated our daily lives. To find out what's ahead, it makes sense to look at what Arpa is funding today.
Unfortunately, that isn't easy. Arpa's scattershot approach to funding supports lots of small and diverse projects rather than a few big ones. So, to get some sense of the hot topics and how they interrelate, Wired mapped Arpa's research using a technique known as co-word analysis. Originally developed by sociologists studying the spread of scientific ideas, the procedure exposes the forces and structures embedded in text. First, we analysed all the Arpa project summaries related to computer technology and picked out the most common technical keywords, such as network and imaging. Then we mapped the results: words that commonly occur together in project descriptions are located near one another, and the type size of a word reflects its frequency. Linked keyword pairs are connected by lines, whose thickness indicates the strength of the connection. For example, the final map shows a thick line between ATM and network because almost every project that mentions one of these words also mentions the other.
The map exposes two main clusters of research. On the left, the focus is on parallel computing. Words like compiler, lan-guage, and memory encircle parallel, reflecting the key concerns of the field. On the right is the network cluster, largely unlinked to any of the terms that surround parallel. Here, applications such as imaging and encryption, as well as technologies like ATM and mobile, radiate out.
Co-word maps are an efficient way to quickly visualise the structure of a research field. Of course, they also have their pitfalls. They don't distinguish between multiple meanings of a word, for instance, which can produce misleading results. But as a first approximation, a co-word analysis provides a useful battle chart of scientific research and a peek at the future.
Steve G. Steinberg (steve@wired.com) is a section editor at Wired US.