It's 1996, and you're bored. What do you do? If you're one of the lucky people with an AOL account, you probably do the same thing you'd do in 2009: Go online. Crank up your modem, wait 20 seconds as you log in, and there you are—"Welcome." You check your mail, then spend a few minutes chatting with your AOL buddies about which of you has the funniest screen name (you win, pimpodayear94).
Then you load up Internet Explorer, AOL's default Web browser. Now what? There's no YouTube, Digg, Huffington Post, or Gawker. There's no Google, Twitter, Facebook, or Wikipedia. A few newspapers and magazines have begun to put their articles online—you can visit the New York Times or Time—and there are a handful of new Web-only publications, including Feed, HotWired, Salon, Suck, Urban Desires, Word, and, launched in June, Slate. But these sites aren't very big, and they don't hold your interest for long. People still refer to the new medium by its full name—the World Wide Web—and although you sometimes find interesting stuff here, you're constantly struck by how little there is to do. You rarely linger on the Web; your computer takes about 30 seconds to load each page, and, hey, you're paying for the Internet by the hour. Plus, you're tying up the phone line. Ten minutes after you log in, you shut down your modem. You've got other things to do—after all, a new episode of Seinfeld is on.
I started thinking about the Web of yesteryear after I got an e-mail from an idly curious Slate colleague: What did people do online back when Slate launched, he wondered? After plunging into the Internet Archive and talking to several people who were watching the Web closely back then, I've got an answer: not very much.
We all know that the Internet has changed radically since the '90s, but there's something dizzying about going back to look at how people spent their time 13 years ago. Sifting through old Web pages today is a bit like playing video games from the 1970s; the fun is in considering how awesome people thought they were, despite all that was missing. In 1996, just 20 million American adults had access to the Internet, about as many as subscribe to satellite radio today. The dot-com boom had already begun on Wall Street—Netscape went public in 1995—but what's striking about the old Web is how unsure everyone seemed to be about what the new medium was for. Small innovations drove us wild: Look at those animated dancing cats! Hey, you can get the weather right from your computer! In an article ranking the best sites of '96, Time gushed that Amazon.com let you search for books "by author, subject or title" and "read reviews written by other Amazon readers and even write your own." Whoopee. The very fact that Time had to publish a list of top sites suggests lots of people were mystified by the Web. What was this place? What should you do here? Time recommended that in addition to buying books from Amazon, "cybernauts" should read Salon, search for recipes on Epicurious, visit the Library of Congress, and play the Kevin Bacon game.
In 1996, Americans with Internet access spent fewer than 30 minutes a month surfing the Web, according to Steve Coffey, who's now the chief research officer of the market research firm the NPD Group. (Today, we spend about 27 hours a month online, according to Nielsen.) In the mid-'90s, Coffey was working in the R&D department at NPD. He and his colleagues had long ago perfected ways to estimate audience sizes on TV and in print, and they wondered if they could port their ideas to the Web. They came up with something called PC Meter: A focus group of a few thousand people installed an application that would silently track everything they did online, and then Coffey and his colleagues would analyze the data. (Traffic ranking firms still use essentially the same methodology.) The NPD Group spun off Coffey's work into a new company called Media Metrix. In January 1996, the firm published what seems to be the first independent ranking of the top sites online.
The biggest site, by far, was AOL.com; 41 percent of people online checked it regularly. Many didn't do so on purpose: With 5 million subscribers, AOL was the world's largest ISP, and when members loaded up the Web, they went to the company's site by default. For similar reasons, AOL's search engine, WebCrawler.com, was the second most popular page. Netscape, the Web's most popular browser, and Compuserve and Prodigy, the nation's other big ISPs, also had top pages.
Yahoo, which Media Metrix ranked No. 4, just after Netscape, was one of the few sites in the Top 10 that wasn't affiliated with an ISP or a browser. Its main feature was its directory, a constantly updated listing of thousands of sites online. To produce the directory, Yahoo employees—actual human beings—reviewed new sites and cataloged them according to a strict hierarchical taxonomy. When you typed in what you were looking for—say, "new magazine," "sexy site," or "advice on taxes"—Yahoo would search its directory and return sites that it had already reviewed. This produced pretty good results—when you searched for "White House Web site," you could be sure you'd get to the right page because someone had actually looked up the official site. Obviously, though, such a model was unable to keep pace with the growth of the Web. In retrospect, it's telling that anyone in 1996 thought this was a sustainable way to catalog the Web. (In 2003, after acquiring the search companies Inktomi and Overture, Yahoo launched its own machine-produced search engine; now, the human-edited Yahoo Directory isn't even listed on the site's front page.)
Some of Yahoo's 1996-era front pages have been saved in the Internet Archive. What's interesting about them is what they lack. First, no e-mail: The first webmail site, Hotmail, launched in July of 1996. There was no instant-messaging software; the first big IM client, ICQ, hit the Web early in 1997. The MP3 file format was invented in the early 1990s, but very few people traded music in 1996—the files were too big to cram down modems, and Winamp, the first popular MP3 player app, was published in 1997. All these innovations hit the Web suddenly, defying prediction, and each completely altered how we'd spend our time online.
Still, some mid-'90s trends do prefigure our current Web obsessions. In Media Metrix's first listing, Geocities.com, a site that let you build your own home page, was the 16th most visited site. Over the next year, it grew significantly, Coffey says, eventually breaking the top 10. "And, of course, that was a precursor to blockbusters like MySpace and Facebook—it was the first we saw of user-generated content, which drives the Web today," Coffey says.
There's a similar trend in blogging. The term wasn't coined until sometime in 1999, but several seminal blogs were already online by 1996, says Scott Rosenberg, one of the co-founders of Salon and the author of Say Everything: How Blogging Began, What It's Becoming, and Why It Matters, which will be published in July. Rosenberg points out that Tim Berners-Lee, the computer scientist credited with inventing the Web, and Marc Andreessen, the coder who founded Netscape, had both set up frequently updated, reverse-chronological Web pages by the mid-1990s. In 1994, a Swarthmore College student named Justin Hall began links.net, one of the very first personal Web sites. * Eventually, he decided that he should be writing much more frequently—what we'd call blogging today. "I think I'm gonna have a little somethin' new at the top of www.links.net every day," he wrote in his first daily post, dated Jan. 10, 1996. Hall's site—unlike so much else that was on the Web back then—lives on today.
If the Web was so completely different just a decade ago, what will become of it in the next decade? When we look back, will we laugh at how taken we were with YouTube—ooh, you can watch everyone's home movies!—and puzzle over how Google missed the rise of the Web-searching technology that suddenly sprang up to vanquish it? Maybe. On the other hand, some parts of the Web have become so deeply ingrained in the culture that it's hard to imagine any force killing them outright. In 2020, we'll get the Internet over electronic ink scrolls powered by algae or something—but we'll probably still be spending a lot of time reading Wikipedia.
Correction, Feb. 27, 2009: This piece originally implied that links.net began in 1996. It began in 1994, while daily updates began in 1996. (Return to the corrected sentence.)