Viral Loop

by ADAM L PENENBERG

  Software lagged a decade behind hardware, with application fiefdoms holding sway over a small, captive computer population. Gopher helped users locate documents, FTP let them download and transmit them, and Telnet hooked them up to other computers. Those who had taken the plunge online connected to one another through communities like the Well in San Francisco, signed up for newsgroups, and frequented chat rooms, many devoted to sex and fetishes. To navigate the Internet required a working knowledge of the Unix operating system and a communications protocol called TCP/IP. The Web wasn’t searchable; you had to know the addresses of the few sites out there you could visit. Then you had to type arcane commands and addresses like “$ find / -type l -print’ perl -nle ‘-e|| print’” (a command to find links that point to nothing). The schism between the pros and the rest was further exacerbated by the prevailing computer systems of the time. Geeks spoke Unix, a staple of college networks, while the public tooled around on Macs and Windows machines. Despite its grandiose name and intentions, the World Wide Web wasn’t very web-like.

  That’s precisely what Marc Andreessen, a twenty-one-year-old computer science student at the University of Illinois in Urbana-Champaign, aimed to change. Raised in rural Wisconsin—his father was a seed salesman and his mother was a shipping clerk at Lands End, the mailbox-stuffing catalog retailer—Andreessen figured he would spend his life with computers even before he had actually touched one. When he was ten, he checked out a computer book from the local library and taught himself Basic (a programming language). At eleven he coded his first application: a virtual calculator to assist him with his math homework, until the janitor shut off power for the night, wiping out his handiwork. Soon after, his parents plunked down $595 for a Commodore 64, which he plugged into his television, storing files on a cassette tape recorder. He quickly learned that hardware had constraints but with software he was only limited by his imagination. Anything was possible, yet nothing was doable if the program didn’t function.

  Andreessen enrolled in college to study electrical engineering, thinking it would pay better, but ended up in computer science because it required less work. Working part-time for $6.85 an hour creating 3-D graphics programs at the National Center for Supercomputing Applications (NCSA), Andreessen burned through hours in chat rooms and roaming around the Web, frustrated with how hard it was to navigate. Berners-Lee’s browser, introduced in February 1991, ran solely on NeXT machines, which made up a fraction of the world’s computers, and displayed text one line at a time like a teletype machine. Other browsers followed, all with shortcomings. Erwise, released by students at the Helsinki University of Technology in April 1992, was the first to work on non-NeXT machines, but development stalled when the students graduated and moved on to other projects. Worse, it was coded in Finnish. Viola originated at Berkeley and incorporated style sheets and tables, plus a bookmark function so users could track where they had been and which sites were their favorites, but it was notoriously hard to install. MidasWWW, developed at the Stanford Linear Accelerator Center, only worked on Unix and Vax computers while MacWWW was the first designed expressly for Macintosh.

  He decided to fashion his own browser, one that would meld Telnet, Gopher, and FTP into one program and secret it behind an attractive, easy-to-master graphic interface. This would build on Berners-Lee’s work, which had created a standard for the Web, key to the Internet’s evolution. Europe of the eighteenth century didn’t move from separate agrarian and cottage industry-based economies to massive industrialization until countries agreed on standardized weights and measures. In pre-Revolutionary America, each colony minted coins and issued its own paper money, which slowed the march to commerce. Eli Whitney reaped a fortune by inventing interchangeable parts for muskets in the 1790s, making machinery reparable and more efficient. Berners-Lee’s standards had the potential for a similar impact on digital communication, except on their own they weren’t far-reaching enough. As with cross-border trade in seventeenth-century Europe, surfing the Web was inefficient, time-consuming, and largely restricted to those fluent in the medium of exchange (in this case, arcane computer systems).

  That’s where Andreessen came in. He wanted to democratize the Internet so that anyone with a computer and a modem could sally forth online. Then users need not be fluent in Unix, Linux, Perl, or DOS. The Web would be open to all, and, if it followed the dictates of history, would grow far beyond the cloistered academic community. It would truly become a World Wide Web.

  Until 1866, when the transatlantic cable was laid, the quickest way to send news internationally was to stash carrier pigeons on ocean-skimming ships and release them near shore. In the early twentieth century, stock exchange runners called “pad shovers” rushed from one broker’s office to the next, shouting the latest prices, which were transmitted by telegraph and published in the next day’s newspapers while pneumatic tube networks fueled by compressed air conveyed small but urgent packages like mail, stock purchases, or money over short distances. Then came the telephone, radio, TV. Yet it wasn’t until the Internet that instantaneous visual and audio communication without regard to geography was possible. Anyone and everyone would be interconnected, and this would have a profound impact on…everything. The idea boggled Andreessen’s mind. But he needed help. At first, Eric Bina, a full-time programmer at NCSA, told him he was too busy. Andreessen persisted, sketching out an almost mystical vision for the World Wide Web, a place where people could access information in a completely different way. He egged his colleague on, realizing the best way to convince Bina to do anything was to tell him how impossible it would be. They were ideal foils. While Andreessen had a voracious appetite and a cosmic view of the world, Bina was short, shy, and wiry, the kind of guy who peered down at a desk and saw molecules. Andreessen was a natural risk taker, “evangelical” compared to the “apprehensive” Bina. George Gilder described them thus: Andreessen was “expansive” and “prodigal with bandwidth” Bina was “bitwise” and “focused.”

  [ MOSAIC ]

  After receiving approval for the project, the two started coding in late 1992. Andreessen wanted it to process images, but Bina, like Berners-Lee, was concerned that people would abuse it, and since graphics files were much larger than text, he feared they could bog down entire networks. Andreessen, the antithesis of the pie-in-the-sky academic, argued it was about time someone brought some color to the Web. He believed that most users, the ones with whom he communicated in chat rooms and forums, would love having pictures, and this would attract others to the Web. Bina gave in. Sketching out their browser’s characteristics, they agreed it had to be reliable, easy to use (reminiscent of the pull-down tabs in Apple’s operating system), and extremely forgiving. Unlike other browsers, if there were errors in a page’s HTML code, theirs would still access the content. It would also enable users to create personal Web pages, which would encourage people to create online identities.

  Andreessen’s timing was born of luck, being at the right place at the right time. NCSA’s supercomputing program originated in the mid-1980s because the government believed it was crucial to provide scientists with a powerful computing infrastructure. Around 1990, however, the nation’s supercomputing centers began to shut down their mammoth Crays. The centers simply couldn’t afford to keep them running, especially with the advent of cost-efficient microprocessors. Because it cost $500,000 to remove a single supercomputer from the premises and scientists preferred crunching data at their desks on a DEC Alpha workstation or Silicon Graphics machine, they just sat there, dead. At this pace of innovation vast computing power would be available to almost anyone in just a few years, and that would profoundly alter the technological landscape. Computers would go mass-market and users would need a simple, intuitive tool to navigate the Internet.

  For the next six weeks, Andreessen and Bina ensconced themselves in the basement of the school’s Oil Chemistry Building, pounding out code from dawn to dusk to dawn again, once for four days straight, Andreessen subsis
ting on Pepperidge Farm cookies and milk and Bina on Mountain Dew and Skittles. Each worked on the parts that most interested him, so much so it didn’t feel like work. Working in the C programming language, Bina compiled the actual rendering engine that displayed the content, and Andreessen designed the interface—the look and feel of the browser—as well as the network backend. They christened it Mosaic after Andreessen laid out the graphics as a mosaic of icons for users to point to and click on. Their Unix version was a lithe nine thousand lines of code, a fraction of the size of Microsoft’s Windows 3.1, the prevalent operating system of the time, which weighed in at 3 million lines. After a dozen friends tested it for bugs, Andreessen unveiled it to the world.

  On the morning of January 23, 1993, he posted a message to several online bulletin boards, which read, in part:

  By the power vested in me by nobody in particular, alpha/beta version of 0.5 of NCSA’s Motif-based information systems and World Wide Web browser, X Mosaic, is hereby released.

  Cheers,

  Marc

  X Mosaic (they later dropped the X) was the Internet’s first smash hit, with enthusiastic early adopters conjuring a powerful word-of-mouth marketing campaign. Even the Web king himself, Tim Berners-Lee, found Mosaic “brilliant,” “exciting,” and “full of features,” promising to advertise it inside the community. Andreessen was overwhelmed by the kudos and queries pouring into his email box. It was a chore just to keep up with the correspondence—people with technical questions, suggestions for more advanced features, bug test results. Companies inquired about licensing the browser or hoped to commercialize it. Just when he thought things were calming down, a new wave would roll in. Within six months of its release, the Unix version of Mosaic, by Andreessen’s count, had one hundred thousand users.

  Even before finishing Mosaic, Andreessen and Bina had rounded up five NCSA colleagues to code versions for Windows, Mac, and software to run a Web server. Crack programmers all, a friendly competition developed. Not only hell-bent on speed, they added additional features like a cursor that changed shape over hot links to make them easier to see. During Thanksgiving week 1994, on the first day Mosaic was available for PCs, the NCSA server overloaded and crashed. By the time Andreessen graduated in December, more than a million people were surfing the Web with Mosaic in Unix, Windows, and Mac flavors.

  [ NETWORK EFFECTS ]

  Andreessen had uncorked a “network effect,” a term first used to describe the spread of the telephone in the early years of the twentieth century. Simply put, the more people who own a telephone, the more valuable an added line is to each person already on the network. The potential number of connections grows exponentially in relation to the number of people on the network.

  With two people there are two communications paths:

  With three, there are six potential paths:

  A network with five members yields twenty possible connections, while twenty results in 380 potential connections. Thousands of users result in hundreds of millions of connections. Millions of users…well, you get the idea.

  Mosaic forged a positive-feedback loop: the more people who discovered Andreessen’s browser, the more who spread it, and this unleashed an ancillary virality stream expressed in the growing online population and the number of websites. The more who downloaded Mosaic, the more websites they created. This attracted additional users, who built more webpages. The greater the online population, the greater the utility of being there—both for those already there and for those entering for the first time. The release of separate Macintosh and Windows versions spiked the virality, as the general public began to pierce this once impenetrable academic sphere.

  In 1992, 4.5 million people were plugged into the Internet and there were perhaps fifty websites. By the end of 1993, 1 million people had downloaded Mosaic, there were 6 million total users, and 623 websites. Within a year the online population jumped to 13 million with 10,000 websites. Web traffic shot up more than 300 percent, with users creating their own home pages, uploading photos, and setting up chat rooms. Today’s Web, with more than 1.5 billion users worldwide and almost 200 million websites, owes its existence to three men: Paul Baran of the Rand Corporation, who conceived the Internet; Tim Berners-Lee, creator of the World Wide Web; and Marc Andreessen, who figured out how to navigate it. In essence, Baran provided the land, Berners-Lee built the roads, and Andreessen manufactured the vehicles (see Figs. 1 and 2).

  FIG. 1. Number of Internet users increased exponentially between 1992 and 2007.

  FIG. 2. Growth of websites increased exponentially in tandem with increase in growth of Internet users.

  In a few years Andreessen would know precisely what to do. In 1993, however, there were no clear paths for capitalizing on a free Web browser, and start-up companies didn’t exactly sprout up in Illinois cornfields. What’s more, since Mosaic was coded under the aegis of NCSA, it didn’t belong to him. The supercomputing center was not shy about trumpeting its ownership. Just as Andreessen was set to graduate in December 1993, a New York Times article on the front page of the business section called Mosaic “a map to the buried treasures of the Information Age” that was being hailed as the “first ‘killer app’ of network computing,” and listed companies, including Digital Equipment Corporation, Xerox, and Novell, that were exploring ways to exploit it for online commerce. NCSA director Larry Smarr, quoted as saying that Mosaic was “the first window into cyberspace,” received the lion’s share of credit, complete with a photograph, while its creators weren’t even mentioned. To add insult to injury, NCSA, which inked a million-dollar licensing deal for Andreessen’s handiwork, asked him to stay on full-time as a manager—but not on Mosaic.

  Andreessen didn’t bother to pick up his diploma before skipping town. Later he would claim he wished he had never studied computer science and instead had taken philosophy and history courses. Trying to forget all about Mosaic, Andreessen relocated to Silicon Valley, where he took a job at a small company specializing in artificial intelligence and defense consulting.

  He wouldn’t last three months.

  [ MOSAIC KILLER ]

  Across the Valley another technology visionary was feeling used and abused. James H. Clark, a year shy of fifty, was packing up his office, about to depart Silicon Graphics, the company he had cofounded a dozen years earlier. Clark had divined a method to coax supercomputer performance out of networked desktops and in the process transformed three-dimensional computer graphics, reinventing the way everything from cars to jets, buildings to suspension bridges and Hollywood films were made, producing the special effects for movies like Steven Spielberg’s Jurassic Park.

  When he first set out on his own, however, Clark was a neophyte in the ways of business. To get SGI off the ground he sold a 40 percent stake to Glenn Mueller of the Mayfield Fund for $800,000, giving his company a laughably low valuation of $2 million. Less than a year later Clark went back for more, trading away another slice of SGI for $17 million. A third round required him and his engineers to dilute their equity even further. At the same time the company sustained a growth rate of more than 40 percent a year, with revenue jumping from millions to billions. When it went public in 1986, its stock leapfrogged from $3 to $30 a share.

  Although Clark was chairman, SGI’s chief executive officer, Ed McCracken, who was brought in by the company’s venture capital backers, ran operations. By 1994, cut off from influence, Clark walked away with a severance of $15 million, a fraction of what Ed McCracken and other executives were worth. Red with rancor, Clark plotted his nemeses’ comeuppance, wanting more than anything to start a new venture that would make SGI look like a mere garage hobby. He couldn’t do it alone, though. On his last day, he asked Bill Foss, a young SGI colleague who often functioned as Clark’s jack-of-all-digital-trades, if he could recommend any brilliant software engineers. Foss tossed out the first name that came to mind.

  Clark had never heard of Marc Andreessen. Preoccupied with his dealings with McCracken and
SGI’s board, he hadn’t been paying attention to the browser craze sweeping the tech world. Foss downloaded Mosaic on Clark’s computer and told him, “You’ll figure it out.” Walking out of the room, he watched Clark squint at the screen, tentatively clicking around. A few minutes later Clark came upon Andreessen’s home page, where the former University of Illinois programmer had posted his re-sumé. Clark marveled at the browser’s simplicity, and it crossed his mind that the first time he was using Mosaic was to contact its creator.

  He typed a now famous email:

  Marc,

  You may not know me, but I’m the founder and former chairman of Silicon Graphics. As you may have read in the press lately, I’m leaving SGI. I plan to form a new company. I would like to discuss the possibility of you joining me.

  Jim Clark

  Then he went back to packing. On the other side of the ether, it took Andreessen all of ten minutes to reply. He knew who Clark was. In college he had whiled away many happy hours on SGI machines. The next morning they met at seven thirty for breakfast in Palo Alto to plot their future—except they had no idea what they wanted to do. All Clark knew was whatever it was, he wanted to do it with Andreessen. Beneath the façade of the gloriously grimy, bleary-eyed twenty-two-year-old was vast experience in programming on a bevy of machines and a vision that jibed with Clark’s. Both searched for ways to simplify technology to make it attuned to the needs of a mass market. Both believed others had cheated them by wresting away control of their creations, and in the process skimming vast sums of wealth off their sweat and inspiration. Both were capable of seeing the big picture and burned with a desire for sweet revenge. Clark wondered about building on the browser, but Andreessen was firm: “I’m finished with all that Mosaic shit,” he said.