At 16,Raymond Lau wrote StuffIt, which soon became the prevailing software for compressing files on Macintosh computers so they take up less space. But Lau really heard his calling when he realized “the mathematical models for data compression are pretty similar to those for language processing.” He joined MIT’s Spoken Language Systems Group in 1994 and was central to its Galaxy project, producing software to recognize speech and interpret language, then deliver database information. He followed with Galaxy II—software that lets U.S. marines access information hands-free. Lau then used Galaxy II as the backbone of the MIT lab’s most ambitious project: Mercury. The system allows anyone to call the lab, speak to a computer and book flights on 23 airlines, as if talking to a travel agent. In 1999 Lau became chief technology officer of startup iPhrase Technologies in Cambridge, MA, to apply his expertise to written words. IPhrase programs have advanced search capabilities for Web sites such as Yahoo! Finance and Schwab.
Internet and Web
In 1994, Richard Barton devised a plan to revolutionize the travel industry. He convinced Bill Gates, his boss, that online travel planning had a future and in 1996 launched Expedia. In 1999,Barton spun the company off from Microsoft and has since grown it into a thriving dot com. Today the site receives queries from 15 million people a month. In February, USA Networks—which owns Ticketmaster and Citysearch—acquired majority interest in Expedia for more than $1.3 billion. Barton, who helped develop the MS- DOS 5.0 and Windows 95 operating systems, says, “Our competitive differential is all about technology. ”Under his leadership as chief executive officer, Expedia developed an award-winning algorithm that compares prices on billions of flight combinations and allows customers to find and buy the lowest fares. Barton now wants to make the customers’ transactions even easier and more secure while customizing services to each person’s buying habits. “Helping people take a trip is fundamental to our long-term dream, ”he says.
Internet and Web
Today’s World Wide Web is a jungle. How to speedily and smartly sort through it? More than 150 million times a day, users turn to Google, the four-year-old search engine developed by a pair of Stanford University graduate students. Sergey Brin and Larry Page (p.84), PhD candidates in computer science, often found themselves stymied when hunting for data. “Innovation in search had halted,” recalls the Russian-born Brin, who had been researching data mining. Brin and Page dropped their doctoral work and came up with PageRank. The software measures the importance of a given Web page by how many other pages link to it— and by how important those linked pages are. As soon as Mountain View, CA-based Google went live in 1998,it attracted Web surfers who wanted rational search results. Today nobody lists as many Web pages (over two billion) or sorts them as fast (a typical search takes under a second).Now that Google is a success, Brin, once known as a jokester, says he has turned serious. “Jokes are no longer allowed—that’s what our PR people tell me,” the copresident says.
For a decade, John Carmack, cofounder of id Software, has revolutionized the computer game industry with immersive first-person shoot-’em-up games where players maneuver through 3-D worlds as characters on the screen. Since 1992 the self-taught programmer has attracted a devoted following of millions and has broken sales records with Wolfenstein 3-D, Doom I and II and the three- part Quake series. His work raised the standard from simple games to complex, role-playing scenarios, which are so compelling that the U.S. Marines have used the games to train fighters. To make possible more realistic environments, Carmack has used leading-edge graphics hardware to create game engines, and he freely allows developers to improve them. “A great many people in the industry got their start modifying our games,” he says proudly. Carmack—who also builds small rocket-powered vehicles—and his crew in Mesquite, TX, are working on a new engine that improves the depth and texture of 3-D environments. His ultimate goal: achieve a level of rendering equal to those “of film and television graphics.”
Ignore the bare feet. Josh Coates may look like an exuberant techie grad student, but he is a serious business player who has convinced investors to pony up $55 million for his 1999 San Francisco startup, Scale Eight. The chief technology officer has a paradigm-shattering idea that says the right software deployed over the Internet or local networks will let large corporations dramatically cut their data storage bills. Right now, data storage involves expensive, proprietary hard drives that are usually deployed at a few central sites; it’s a $20 billion market set to grow inexorably as more computers produce ever more information. But Scale Eight challenges that inevitability. Coates says his software will let customers use networks to route data to scores of cheap, off-the-shelf hard drives, where they can be stored inexpensively and securely. “I’m trying to sweep hardware out of the way and thereby commoditize storage, really lowering the costs, ”says Coates, who counts Microsoft among his two-dozen customers. “Software has no bounds, ”he adds. “If you can think of it, you can do it in software.”
Paul Debevec’s rise to computer graphics stardom sounds like a fairy tale. In 1996 Debevec presented a paper on Façade, a system he developed as a student that digitally generates 3-D scenes from 2-D photographs. Soon after, he was flown to Hollywood to present his technology to John Dykstra, the visual-effects supervisor on Batman and Robin. Effects companies have since used Debevec’s techniques in several films, including The Matrix. Debevec now directs the graphics laboratory at the University of Southern California, where he is perfecting the Light Stage. Inside this three-meter-wide spherical structure, actors and objects are illuminated by 156 light-emitting diodes that duplicate light from any environment. For example, an actress can be illuminated with light recorded inside the Sistine Chapel, and her image can be simultaneously superimposed on a scene set there. The technique yields far more realistic results in less time than the standard method of adjusting concocted lighting frame by frame. “The idea is to use the light from the actual scene, rather than manually try to approximate it,” says Debevec, who admits to being under Hollywood’s spell.
Internet and Web
Not many undergrads write world-changing code, appear on the covers of Time, Fortune and BusinessWeek, or testify in front of U.S.Senate committees before they can legally buy beer. Shawn Fanning has done all three since founding the cultural juggernaut Napster in a Northeastern University dorm in 1999. Fanning transformed a software script he wrote to help a roommate retrieve digital music files from the Internet into a full-featured online swap service millions of users strong. The free program enabled users to post MP3 digital music files they had on their computers to an online index supported by Napster, and to access files from any other person’s computer. That way, users could swap music files directly. The application became so popular that the Recording Industry Association of America effectively shut it down in 2001 through lawsuits alleging copyright infringement. Fanning is busy relaunching his company as a paid subscription service. His concept, though, continues to challenge the status quo. Music industry giants are scrambling to mimic Napster’s success on a pay-per-use basis— but to no avail, as free copycat sites constantly spring up.
Internet and Web
Justin Frankel has brought music to desktops in numerous ways. After dropping out of college in 1997 and returning home to Arizona, he wrote Winamp, a program that let people play downloaded MP3 music files on their PCs. It was much easier to use than existing MP3 players. He and partner Tom Pepper also devised Shoutcast, which enables computers to broadcast like radio stations over the Internet. To vastly expand music’s availability online, Frankel then created Gnutella, a system that lets Internet users swap MP3s and other files. Unlike Napster, Gnutella does not pass files through a central distribution point—and recording companies can’t track them. By the time Frankel released Gnutella in 2000, he had sold Nullsoft, the company under which he developed Winamp and Gnutella, to America Online.AOL paid $400 million for Winamp and online-radio pioneer Spinner Networks and merged them under AOL Music in San Francisco. But AOL became wary of Gnutella because it let people acquire music they hadn’t paid for and pulled the program. What’s next from Frankel? “Just stuff that hopefully will make a difference,” the rebel says. That’s a tune he’s played before.
Vinay Gidwaney wrote the software that his Calgary, Alberta, high school used to teach his classmates word processing. Resellers expressed interest, and Gidwaney, only 16, started a small company to supply it. But he found himself spending a lot of time meeting customer requests. Youthfully impatient and eager to reserve his time for writing code, he created software tools to automatically handle certain customer support tasks .Gidwaney soon realized he could develop versions of the software tools to sell to other companies, to enable them to provide live support to their customers over the Internet. So the Canadian started Control-F1 in Calgary. Gidwaney, chief technology officer of the 40-person company, calls his tools “better than being there.” That’s because a remote customer can continue to work on her computer while Control-F1 software is solving her support problem in the background: no need for her to step away from the computer for a human technician. Several organizations now use Control-F1 to provide customer support, including Novell, Unisys and IBM.
Choosing a car color is hard enough. Imagine trying to make strategic purchasing decisions for a huge corporation. Robert Guttman’s knack for softwareagents—autonomous, personalized programs that facilitate better-informed decisions—has made such buying chores easier. With an artificial-intelligence degree and four years at Motorola, Guttman arrived at the MIT Media Lab in 1996 to plan the world’s first agent-mediated marketplace experiment. His idea was to create software agents that could find certain goods for their masters at preferred prices, then negotiate and close sales on the buyer’s behalf. The successful experiment left Guttman wondering whether similar agents could function in real-world marketplaces. In June 1998,along with two MIT colleagues, Guttman founded Frictionless Commerce in Cambridge, MA, to commercialize his technology. The software is now used by operations like the U.S. Army for large purchase orders of laptops, truck brakes, even lumber. With Frictionless’s success secure, Guttman has left his post as chief technology officer— though he remains a board member— and is shopping his talents around.
In school, Ramesh Hariharan found biology boring. But once he became a computer science professor at the Indian Institute of Science, he got excited about the race to map the human genome. So he cofounded Strand Genomics in Bangalore, where he designs software tools to efficiently analyze the ever increasing volume of data about the makeup of genes. One U.S. customer is applying Hariharan’s data-crunching innovations to proteomics—the analysis of protein structures to aid in the discovery of new drugs. Strand Genomics expects to grow from 35 to 100 employees this year. Wearing another hat, Hariharan also works to bridge the digital divide. With colleagues from the university and from a local software firm, he started the nonprofit Simputer Trust to develop a simple, cheap (under $200), portable, battery-operated computer to bring the Internet to the developing world. The trust’s first targets are rural Indian village schools, hospitals or community centers that have phone lines. Villagers get smart cards that give them access to a shared Simputer, while touch-screen icons and the Dhvani text-to-speech system Hariharan developed empower illiterate users.
Mar Hershenson came to Silicon Valley from Barcelona, Spain, for a summer job, met her future husband and stayed, bringing a bit of her native city to California in the form of Barcelona Design, which she cofounded in 1999.The Sunnyvale, CA, company produces software and intellectual property, developed by Hershenson, for quickly optimizing the design of analog circuits for cell phones, TVs and DVD players. Previously, engineers could spend a year designing a single analog chip. With Barcelona’s solution, custom analog circuits can be finished in hours. Hershenson’s breakthrough was to represent circuits with equations that can be solved mathematically. She learned the technique in a course taught by Stephen Boyd, the Stanford University professor with whom she launched the company. The 45- employee firm has raised $44 million and lined up several large clients, including chip-making giant STMicroelectronics. Bursting with ideas, Hershenson plans to apply Barcelona’s technology to a wider range of circuitry.
Internet and Web
Travis Kalanick is good at escaping sticky situations. In 1998,he launched Scour.com with six buddies at the University of California, Los Angeles. What began as a Web search engine morphed into a popular peer- to-peer file exchange system with 250,000 simultaneous users trading movies and music. Everything was looking up until more than 30 media companies sued Scour for $250 billion for copyright infringement. Scour settled and eventually sold its assets. Then, in 2001, Kalanick founded Los Angeles-based Red Swoosh with Scour’s engineering team. They’ve developed software that streamlines the way content—documents, music, videos—is moved around on the Internet. Typically, when you request a file from a Web site, it is delivered from a centralized server. Red Swoosh’s software continually updates a directory that lists which files are on which servers and end-user desktops and transfers the file to you from the closest source, speeding delivery. The scheme also saves big bucks in server infrastructure for the company that posted the file. Several media moguls with busy Web sites are now testing his software.
Lydia Kavraki made her first move between worlds when she left Greece to do a PhD in computer science at Stanford University. Drawn to the human potential of robotics, Kavraki studied how robots—from assembly line “arms” to autonomous machines—assess the obstacle-laden world and move around in it. She then created an algorithm that rapidly generates a path for a robot to follow through a given environment, using descriptions of how the robot moves, the space it’s in, obstacles it must navigate and its beginning and end points. Today, most papers on robot-path planning cite her algorithm, and engineers in the automotive industry are using variations of it to build better robotic assembly lines. Kavraki, meanwhile, has moved to a new research world, applying the rules of her algorithm to predict how two molecules will move through space and interact with each other—crucial to designing drugs. Intense and determined, Kavraki finds the two problems closely linked: “There is a potential here for solving problems that could affect our lives, whether it’s a robot that helps disabled people get out of bed or a tool that helps find a compound to treat disease.”
Internet and Web
The Internet is a great set of parts. Reiner Kraft wants to make them a more valuable whole. One way is to exploit the many computers linked to the Internet to solve massive computing tasks that no single computer could handle well. Kraft coinvented a program that parcels out such tasks over the Internet to thousands of PCs; each solves its morsel, and the program integrates the solutions. In 1998,IBM applied for a patent based on Kraft’s distributed- computing scheme, which is used increasingly to crack computing problems. Kraft then created jCentral and xCentral—custom search engines for IBM’s programmers. They search and return Java and XML programs, respectively—the software behind many Web applications— and nothing else. This allows the company’s programmers (and non-IBMers, too) to much more efficiently build libraries of code for creating advanced applications that leverage the Internet’s capabilities. Kraft has cranked out numerous other programs that integrate Internet functions and has filed 80 patent applications. Despite his prolific youth, however, Kraft frets about what’s been left undone. “I worry,” he says, “that I am missing some good opportunities.”
Innovator of the Year
After emigrating from Ukraine to Chicago as a teenager, Max Levchin enrolled as a computer science student at the University of Illinois so he could create and break codes. He moved to Silicon Valley after graduation to start a company based on his cryptography passion. In 1999, he cofounded PayPal in Palo Alto, CA, which quickly became the Internets leading person-to-person payments processor. One in four transactions on eBay is settled using PayPals system for debiting and crediting checking accounts and charge cards. In February, the company went public, raising $70 million.As chief technology officer, Levchin not only manages servers that store encrypted data about the companys 15 million members but has led the development of an antifraud program called Igor, named after a Russian fraudster it helped apprehend in 2000. Igor monitors PayPals transactions for unusual behavior, alerting personnel to freeze suspicious accounts or head off cash en route to dubious destinations. The FBI has also enlisted Igor to combat wire fraud. Citibank and Bank One, and even eBay itself,have launched rival online payment services, but none has matched PayPals market share.
In 1997, after finishing her PhD and starting up Imagen in Cambridge, MA, Pamela Lipson would get phone calls from her mentor Alex d’Arbeloff, chairman of the MIT Corporation. “Focus,” he’d always tell her. Lipson had devised algorithms that could rapidly identify and classify digital images. Venture capitalists wanted them, but for far-flung applications: to improve Web searches for images, or for face recognition, video-database indexing or pharmaceutical R&D. But it was not clear any of these emerging markets would embrace Lipson’s technology. In a quest for real customers, Lipson bet on inspection of printed circuit boards. She adapted Imagen’s software so it could identify production errors from a digital snapshot without misidentifying normal variations in parts. She designed a straightforward interface so users could easily modify the software. Inspections using Imagen software enhanced productivity without introducing lag. “What used to take five minutes now takes 20 seconds,” says Paul Keating at Teradyne, which has rights to use Lipson’s technology. Now that’s focus.
Internet and Web
While Rob—don’t call him Robert—Malda may fit the irreverent hacker stereotype, his finest hack does not. Malda is founder of Holland, MI-based Slashdot, a Web site cum online community cum Internet Zeitgeistmeter visited by more than 250,000 surfers daily. What started in 1997 as an online hang-out for Malda’s cronies to trade banter on geek subjects is now “the number one site for tech news and geek ranting, ”according to the Washington Post.Contributors recommend news items to Slashdot, where Malda and his small staff create links to the stories and write introductory paragraphs. Readers post comments, which are then graded by other readers. Many times, Web sites whose addresses are cited experience the “Slashdot effect”—an increase in traffic so sharp that their operations sometimes halt. The open-source program that runs Slashdot, which Malda created and regularly works on, is intuitive enough to have attracted 500,000 registered users. Countless others have downloaded it to run their own online discussion groups. As Malda continues to refine the Slashdot experience, he will refine the way the world experiences the Internet.
Internet and Web
Steve McCanne’s career as a rock star fizzled in high school. But noodling on a synthesizer did spark his interest in digital signal processing, which blossomed into graduate work at Berkeey National Laboratory. There he helped his mentor, Van Jacobson, invent the “Internet multicast backbone” (Mbone), which led to Internet standards for streaming media and enables people at scattered locations to collaborate using video, audio and a whiteboard. Among Mbone’s first users: NASA engineers. In1998,McCanne cofounded FastForward Networks and pioneered the first scalable techniques for live Internet broadcasting. In 2000 Internet giant Inktomi bought FastForward for $1.3 billion to get its multimedia tools—and McCanne, now chief technology officer. Inktomi, in Foster City, CA, performs cataloguing and searching for huge portals like America Online and MSN. McCanne is now devising systems to let big businesses, including Ford Motor, McDonald’s and Merrill Lynch use video webcasting throughout their own networks. Someday he’d like to write a book about “how the Internet really works.”
Internet and Web
It’s one thing to devise a key innovation for the Internet. Lou Montulli has designed half a dozen. While a computer science major at the University of Kansas in 1991,he wrote Lynx, a program that enabled a computer user to automatically link text documents. It became one of the earliest and most popular World Wide Web browsers. At the same time, Montulli was a leading figure in the grass-roots effort to improve several fundamental computer languages and protocols, including the hypertext transfer protocol—the addressing scheme that links Web pages—and HTML, the language for creating text and images on Web pages. In 1994 he moved to California to work as a founding engineer at what became Netscape, developing the first commercial Web application. Not all of his innovations have been universally embraced: he is responsible for cookies—data files that enable Web sites to recognize returning users—as well as blink tags—those endlessly flashing words on Web pages. Shrugging off the burden of being named People magazine’s sexiest Internet mogul of 1999,the freelancing Montulli continues to experiment with new ways to exploit the Internet.
Internet and Web
Commuters in Japan are staring into their hands—and Kazuho Oku is to blame. Oku used personal digital assistants in high school but only experienced the Internet when he enrolled at the University of Tokyo as a geology major. He was struck by how much more useful the Internet could be—especially to idle commuters on subways and trains—if it were easily accessible over handheld devices. He was soon spending his time in the university’s computer department, devising a way to compress Web pages and developing software to convert them into a format for handhelds. The result was Palmscape, one of the world’s first Web browsers for handhelds. Oku distributed Palmscape—intended for the Palm Pilot’s Palm operating system—free over the Internet. Before finishing his studies, Oku was lured to Ilinx, a software company in Tokyo, where he developed his successor product, Xiino. It comes installed in a wide range of handhelds and is a leading browser for Palm products in North America, Europe and Japan. Oku is now adding capabilities that allow corporate clients and individuals to write their own custom applications.
Internet and Web
Google combs more Web pages, faster, than any other search engine. But perhaps just as impressive is that despite the dot-com meltdown, the company has never veered from its mission. Competitors have tried to reposition themselves as all-purpose “portals,” only to slip from the radar screen. That thrills Google cofounder Larry Page, a feisty roller hockey player. Google has continued to expand the kinds of data it searches, recently adding Usenet news groups and retail catalogues. And it keeps expanding its tool set: for instance, Google now serves its results to cell phones. Copresidents Page and Sergey Brin (p.69) have worked as equal teammates since they first devised their unique search software and went live in 1998.They are both competitive but know how productive they are working together. Their responsibilities often overlap, and they still share an office. For them, Google is about solving intellectual problems. Indeed, they recently recruited Novell and Sun veteran Eric Schmidt as CEO tomanage their 300-plus employees, so they can continue to focus on technology. “Our goal is to keep innovating,” Page says.
Internet and Web
Joseph Reagle bikes rather than use polluting transportation. He eats vegetarian so animals are not killed, and brings a quiet but strong sense of social conscience to bear on issues like trust, privacy and intellectual-property rights on the World Wide Web. After earning his graduate degree from MIT’s Technology and Policy Program in 1996,Reagle established himself as a creative thinker at the World Wide Web Consortium, based at MIT. He has driven several initiatives that will dramatically affect online interactions. He led the group that developed a standard way for Web sites to disclose their privacy policies, telling people what might be done with personal information. He coordinated input from far-flung institutions to create rules recognizable by all Web browsers for signing online documents, so people can leave unique stamps verifying that documents have been made or approved by them. Sound like fun? It was for Reagle, who says innovation is not just for technology but for culture. Web inventor Tim Berners-Lee, who oversees the consortium, says Reagle is “continually” looking to better the relationship of the Web tosociety.
Internet and Web
When Tim Tuttle was 30,he quit his job at Lucent Technologies’ Bell Labs, moved into a dilapidated apartment in Cambridge, MA, and began to reinvent the Web. It had a freshness problem. If the information on a Web page changes while you’re reading it, you don’t know—unless you hit the “refresh” button. Tuttle saw a different possibility: a virtual network, overlaying the Web, that lets sites send you live updates on information that changes, the moment it changes. And he was sure he could get it to function on ordinary Web browsers using ordinary Internet protocols—no extra software needed. Working with no funding or source of income, he built the first node of such a network: the prototype of the “Bang object router.” After securing $10 million in July 2000,Tuttle,an active ultimate Frisbee player, moved to San Francisco to become an Internet entrepreneur. Six months later, his network was up and running, used chiefly by financial-services companies that need continually updated information from dozens of sources. Tuttle’s company, Bang Networks, has thrived through the dot-com collapse: it raised almost half of its $32 million in November 2001.
Internet and Web
Handheld wireless devices are great for voice and simple data but are frustratingly limited when it comes to handling video—mostly because today’s networks were designed for wired computers with robust processors and full-sized screens. Susie Wee, R&D manager for Hewlett-Packard Laboratories’ streaming media systems group—and an avid hockey player—is skating around those constraints. Her first move was to devise algorithms that adapt data-heavy video streams to the capabilities of different online computers. The result:a handheld device can receive video at a lower resolution than a workstation, allowing it to display the video much faster. Wee is now developing protocols for moving streamed content away from central Internet servers to cache servers geographically closer to end users. Doing so would reduce network congestion and interruptions, making video and audio flow more easily to wireless devices. Wee’s goal is to turn your cell phone into a full-blown multimedia player—a goal she is speeding toward.
Internet and Web
When Ethan Zuckerman went to Ghana in 1993 as a Fulbright scholar in percussion, he immediately tried to get online; he was a Usenet junkie and eager to e-mail his girlfriend (now his wife). But in bustling Accra, he found only one temperamental Net connection. Zuckerman later became vice president of R&D at Web-hosting company Tripod, which made him a dot-com millionaire, but he never forgot Ghanas inadequate communications. In July 1999 he left Tripod and in February 2000 cofounded Geekcorps in North Adams, MA. Geekcorps sends volunteers with information technology expertise to underdeveloped countries for four-month stints, where they help businesses-from furniture factories to radio stations-get online, expand sales and thus create jobs.One volunteer even helped launch the Ghanaian parliaments Web sites. Funded by foundations, aid agencies and private donors, Geekcorps has sent 35 tutors to Ghana and several other countries.A recent merger with the International Executive Service Corps gives Zuckerman the support to expand much further. Theres no shortage of volunteers; more than 1,100 people are on Geekcorpss waiting list.