How the Internet Happened

by Brian McCullough

  AOL, CompuServe and Prodigy were all what the industry liked to call “walled gardens.” They were online services that provided their users with proprietary tools and packaged content developed by the services themselves or their media partners. Little of what the online services did (with a few exceptions, such as email) interacted with the larger Internet, and none of the services were based on Internet standards. In a very real sense, online services like AOL didn’t actually want users wandering outside of their networks and their control of the content. They much preferred if users stayed to play in the garden. The rise of the World Wide Web changed all this radically.

  AOL always had a schizophrenic relationship with the Internet. The web provided a new, wilder alternative online environment, and in some ways this was in tension with AOL’s carefully cultivated online “community.” After all, would AOL prefer you researched cars on a Car and Driver “channel” on AOL proper, or by going on the web and visiting Car and Driver’s website? In interviews from the time, Case repeatedly floated the notion that the web was complicated and “niche” while AOL was targeting a mainstream audience by providing simplicity: “One disk to install. One price to pay. One customer service number to call. Building web sites and hoping people will find them is a significant leap of faith.”26

  “Their attitude toward the Web is a little grouchy,” an industry researcher said of AOL at the time. “They have a hard time getting past their own resentment that this disorganized cousin is taking over in the public’s mind. But it is a bias against inescapable realities.”27

  At the same time, however, the web also presented AOL with a rare opportunity. AOL’s millions of users were still paying by the hour to dial in, and if AOL simply turned on access to the wider web, those same users would still be paying for the privilege of going through AOL’s pipes. As AOL executive Ted Leonsis put it, AOL could become “the Carnival Cruise Lines” of the Internet, the trusted guide to places unknown.28 Prodigy was actually the first online service to allow its users to browse the web, in December 1994, but AOL soon followed suit.29 AOL then rushed headlong into a $160 million Internet-based spending spree in order to keep abreast of the changing landscape.30 A perfect example was BookLink and its Internet browser, which AOL snatched from Microsoft’s clutches in November of 1994. AOL bought companies like Advanced Network & Services Inc. to build out its dial-up network (and thereby burnish its credentials as an ISP), and it bought a website called the Global Network Navigator, an early version of a search engine/Internet directory.31 There were even very serious discussions about AOL doing some sort of investment in the young Net­scape.

  AOL’s pivot to position itself as America’s most popular on-ramp to the Internet quickly paid dividends. The number of subscribers grew to 6 million. Almost overnight, one out of every three people surfing on the Internet in the United States did so via AOL’s dial-up lines.32 This growth showed up on the bottom line. AOL recorded revenues of $1 billion for the first time in 1996, tripling what the business had brought in only a year before. AOL’s stock had risen thirtyfold since its IPO; its market cap reached $5 billion.33 While it still insisted on paying lip service to its own walled garden of content, AOL had wisely ridden the web’s growth like a bucking bronco.

  But the bronco was not always easy to ride.

  Starting at 4 A.M. on August 7, 1996, AOL’s services went down for nineteen hours.34 The outage made front-page news around the country and made AOL the butt of jokes on late-night talk shows. For AOL, it was a major public relations black eye, but at the same time, a validation of how important the service had become in a few short years. This wasn’t just an early adopter’s playground anymore; AOL was how Americans were increasingly living their online lives every day. Imagine the chaos that would occur today if there were no email, no web, no anything online for nineteen straight hours. The Internet itself hadn’t crashed, but America’s ability to access it had. Suddenly, that was a big deal. The service outage came on the same day that NASA announced the discovery of indications of water on Mars, but AOL was the lead story on CNN.

  Worse was to come. While AOL was now the country’s largest Internet service provider, it was still competing in a crowded field. In addition to Prodigy, CompuServe and MSN, there were thousands of independent mom-and-pop ISPs spread around the country. An independent ISP didn’t have the packaged content and proprietary chat rooms that AOL had. The indies gave users one thing: the Internet. You dialed in and you were on the web, quick and dirty. Increasingly, that seemed to be all people really wanted. To stand out from their online service brethren, ISPs competed on price. A low monthly fee of $19.95 got you unlimited hours of usage. This put quite a bit of pressure on AOL, which still depended on hourly rates and overages for the bulk of its revenue. Why was the Internet worth $2.95 an hour on AOL when you could browse unlimitedly elsewhere for a flat fee? The pressure from cheap competition threatened AOL’s meteoric growth. In a quarterly report at the end of 1996, AOL announced signing up 2.1 million subscribers, but at the same time losing 1.3 million subscribers who fled the service for other ISPs.35 AOL was still the market leader in terms of sheer numbers, but this competition and customer churn started to worry Wall Street.

  The hourly fee structure was unsustainable. MSN announced in October of 1996 that it would provide unlimited access to its service for $19.95 a month, copying the business models of the independents. AOL had no choice but to follow suit. Starting with the December 1996 billing period, AOL announced that it would switch all of its users over to unlimited usage plans for the price of $19.95 a month. There were concerns internally about whether or not this move would kill AOL’s hourly golden goose. “I had data and I had projections on how much money we would lose,” Jan Brandt says. “We had so many people that were paying us 50, 60, $70 [a month].”36 Flat-rate pricing would bring that to an end. More seriously, there were concerns about the network’s ability to handle the increased usage that would inevitably occur. After all, members who had previously tried to limit their time to a few hours here and there could now, if they wanted, leave their America Online connections going 24/7. AOL testing suggested that actual usage would only increase 50% or so in an unlimited paradigm, and theoretically, the network could handle that.37 But those assumptions were only taking into account existing users. Wasn’t the point of flat-rate pricing to stop the churn, win back old customers, and maybe entice new ones? Steve Case told a Wired reporter that he thought the company had the infrastructure in place to handle “runaway growth.”38 He could not have been more wrong.

  The very first day that user accounts were switched to “unlimited” pricing, member sessions leapt from 1.6 million hours to 2.5 million hours.39 The numbers would only go up from there as, over the course of the month, more member plans were switched over. In addition, December was, of course, the height of the holiday season, and plenty of new computers were unwrapped as gifts that month. Now, with the promise of unlimited usage, all those bundled AOL trial discs were suddenly a lot more enticing. AOL signed up a record half-million members that December alone.40 AOL’s daily usage numbers were now up to 4.5 million hours each day.

  There were too many people trying to log in all at once. The service couldn’t handle it. Across the country, instead of the familiar guttural noises of the modem connecting, users began to hear only busy signals. Frustrated members would try over and over again to connect, hoping to get lucky. If users did get online, they tended to stay on as long as possible because there was no telling when they’d have the chance again. Once more, there was nationwide consumer outrage. The jokes began to circulate again about “America OnHold.” CompuServe launched an advertising campaign to attempt to take advantage of its rival’s misfortune, using the phone number 1-800-NOT-BUSY.

  “We didn’t really have an internalized grip on how important we were to people’s daily lives,” Jan Brandt says of the crisis. “What we didn’t calibrate was the ferociousness of the response. It was crazy
and it was really enlightening. It was like, ‘Oh my God! People love us! They really love us!’ Or, at that point, they love-hated us.”41

  In the end, AOL would spend hundreds of millions of dollars in a crash program that attempted to increase network capacity and bandwidth. Millions more were set aside to refund users and head off lawsuits and government scrutiny. Television ads were suspended so as not to encourage too many new sign-ups until the problems were fixed. Over the first few months of 1997, the busy signals slowly went away and service went back to normal. And the especially positive news was that the users stayed loyal. Even in the face of this well-publicized fiasco, user churn subsided. And, once they could actually use the service again, members did so in ever-increasing numbers.

  AOL survived on the strength of its branding as America’s online gateway. A lot of Americans didn’t want any other way to get online; many didn’t even know there was any other way. “Long lines are endemic at Disney World,” a new AOL executive named Bob Pitt­man said. “Folks hate them. But offer Six Flags as an alternative and they look at you like you are crazy. They don’t think anything is a substitute for Disney.” AOL survived and continued to thrive for one reason, according to Pittman: “It’s the brand, stupid.”42

  4

  BIG MEDIA’S BIG WEB ADVENTURE

  Pathfinder, HotWired and Ads

  But what exactly was the big draw of the web? Why were people clamoring for AOL to add web access? Why had Net­scape gotten a billion-dollar valuation and Microsoft revamped its entire corporate strategy? What exactly were people doing on the early web? Well, it was hard to say at the time, and maybe even harder to say twenty-five years later. The early web was sort of everything and nothing at the same time.

  The entrepreneur and venture capitalist Chris Dixon has remarked that “the next big thing always starts out dismissed as a ‘toy.’ ”1 This is very often true with Internet technologies; a new site or a new tool can, on first encounter, seem gimmicky. Why would I ever want to use/do that? is many a first user’s impression of the new. The web and the Internet itself engendered this reaction among many during its early days. At the time, the most enthusiastic net cheerleaders were touting it as a revolutionary medium that would completely change our lives. But there were still others who looked at the net and saw, yes, a toy. And we have to admit that these skeptics had a valid point of view, even with the benefit of hindsight. Because so much of the early web was decidedly amateur.

  For example, one of the notorious early websites was the Net­scape Fishcam, which was maintained by Lou Montulli, one of the original Mosaic six that Marc Andreessen and Jim Clark had recruited from the University of Illinois. This was, simply, a live webcam of a fish tank. Nothing more. It still functions to this day at Fishcam.com. It had a spiritual twin in the world-famous Trojan Room coffee cam, which showed a real-time image of a coffee pot in the Computer Laboratory of the University of Cambridge, England. It was a live video feed of a coffee pot. That’s it. But for people on the early web, the fact that, at any time of day or night, you could see if a coffee pot, halfway around the world, needed to be refilled, that was just—kind of cool.

  The list of early web ephemera could go on and on. There was a site that translated your name into Hawaiian; Cows Caught in the Web featured bovine trivia for no particular reason; Interactive Frog Dissection allowed you to dissect a virtual frog; Doctor Fun pioneered web comics; the Ultimate Band List (formerly the Web Wide World of Music)2 listed information on bands and concerts and indie music in general;3 the Frank Lloyd Wright Source Page tried to catalog pictures and analyses of every work the great architect ever created; Hiram’s Inner Chamber provided info on Freemasonry; if you were a fan of the nineties cartoon Animaniacs, the Animaniacs Page! was obsessively complete; the Bonsai Home Page was all things tiny Japanese trees. The nature of the web made publishing so simple, anyone could publish a website about anything, and lots of people did.

  But there were early websites with serious utility as well. The first commercial web publication was called Global Network Navigator, or GNN. It was launched all the way back in May of 1993, under the umbrella of the technology publishing company O’Reilly & Associates. O’Reilly published computer books and manuals, and in 1992 published The Whole Internet User’s Guide and Catalog, one of the first books about the Internet targeted to mainstream users. One O’Reilly “associate,” Dale Dougherty, was tasked with creating a rudimentary website to put the online-catalog portion of the book actually, you know, online. Dougherty continued to add layers of content until it functioned as a sort of online magazine, as well as a directory, listing cool websites in one of the first attempts to bring search and discovery functions to the web.4 GNN would eventually be swooped up by AOL during its 1994–95 buying spree of early web properties. The Bureau of Labor Statistics maintained a website very early on to provide up-to-the-minute data on labor market trends. FedEx allowed customers to track the status of package shipments before most people even knew the web existed. Alamo Rent A Car was the first to allow users to book a car from its website. BankNet in Britain was the first bank to allow online account creation (if not actual online banking). Nature’s Rose Floral Services allowed you to order flowers from the web. Classifieds began migrating to the web almost from the beginning, because for years there had been digital antecedents on the message boards and newsgroups of the early Internet and online services. The company that would eventually become Monster.com began life in 1994 as a site called The Monster Board.

  Newspapers, for all their later reputation as being roadkill in the Internet Era, were actually prominent Internet pioneers as well. But then, publishers had more protodigital experience than almost anyone. For years, newspapers dreamed of electronic delivery of their product; digital would mean the elimination, or at least mitigation, of their greatest cost centers: paper, printing and physical delivery. Like the cable and telecom companies, they had sunk millions of dollars into digital experiments going back to the late 1970s. The dream of digital took its biggest step with Knight Ridder. The San Jose Mercury News was a Knight-Ridder publication, and it just so happened to be the hometown newspaper of Silicon Valley. Perhaps it was that proximity to the swelling tech revolution that led the Mercury News to launch Mercury Center in 1992. Mercury Center would offer the Mercury News’s regular content, but with more in-depth offerings online. It was sort of how shows or publications will now often say, “If you’d like to see the full interview, go online.” Mercury Center’s wares were designed to be complements to what the paper was already doing—content extensions. It carried press conference transcripts, wire stories that didn’t make the printed edition, and legal documents and notices. Codes were printed at the bottom of stories so that readers could call or log in for the additional content. This cost $9.95 a month, and users without a computer could pay $2.95 a month for phone and fax service. In other words, you could have headlines read to you or faxed to your home or office. All of this was made possible via a partnership with America Online, which handled the monthly fees.

  The Mercury Center was a small but genuine success. Newspapers around the country came to the Mercury News to see how the experiment was working out. In early 1994, the New York Times ran a profile on the Mercury Center noting that there had been 5,100 sign-ups, which represented a little under 20% of America Online’s 30,000 subscribers in the Bay Area, albeit, less than 2% of the Mercury News’s 282,000 subscribers. The New York Times article noted that one key innovation was that reporters were urged to interact with readers about their stories. Bob Ingle, who led the Mercury Center initiative, told the Times reporter, “Our communication historically had been: ‘We print it. You read it.’ This changes everything.”5 It was a lesson that all media entrants to the Internet era would have to learn, or not learn, at their peril.

  In the winter of 1994 when the Netscape Navigator browser came out, the Mercury Center quickly embraced the web. In January 1995, the Mercury News launched a website
, with access originally $4.95 a month, though the paywall was later dropped in an effort to land more advertisers. The Mercury Center again found small but genuine success on the web, with thousands of new subscribers and $120,000 a month in revenue by its first year. By 1997, the website could claim 1.2 million monthly visitors. Under the Mercury Center’s auspices, the Mercury News would continue to break ground, becoming the first daily to put the entire content of a given issue online while also being the first to use the site to break news, instead of waiting for the next day’s edition. In April 1995 when the Oklahoma City bombing occurred, a photograph flashed across the wires that would become iconic, that picture you might remember of a firefighter holding a child in his arms. The Mercury Center immediately posted it to the website, over the objections of the photo editor, who wanted to save it for the next day’s front page.6

  Similar experiments were taking place in the magazine industry. In 1995, the journalist and commentator Michael Kinsley launched a web-only publication for Microsoft (this was in the midst of Bill Gates’s “hard-core” web obsession). Formerly an editor at Harper’s Magazine and the New Republic, Kinsley very much intended the new publication, called Slate, to be a “magazine,” complete with issues and publishing dates. In an early memo to staff, Kinsley wrote: “There should be a notional moment each week when we ‘go to press’ and ‘hit the stands’ (one and the same in this medium). I would say Friday midnight. This will allow us to summarize the week, and allow people to read us ‘fresh’ over the weekend.”7 Readers would be encouraged to print up articles and read them at their leisure. Within the editorial brain trust, there were actual debates over whether any normal person could be asked to read any piece over 700 words on a cathode ray screen without eye strain or boredom. At one point, Kinsley argued that each new piece or article would replace an older piece and the old piece would disappear forever. Slate launched with page numbers and a traditional table of contents, even though, obviously, numbering pages on the web was pointless. There were even debates about whether or not to allow hyperlinks in the articles, for fear of sending people away to other sites. Most of these callbacks to print media would be abandoned shortly after Slate launched.