Is the Web "contracting"?

The numbers show a bigger slice for the top sites -- but most of the pie remains in the hands of the little guys.

By Scott Rosenberg
Published August 26, 1999 8:00PM (UTC)
main article image

Any time the words "study" and "Internet" appear in the same sentence, beware -- there may well be some nonsense afoot.

Readers with long memories (by Internet standards) will remember the infamous Marty Rimm "study" of pornography on the Net, which made the cover of Time magazine, despite vast problems with its methodology and conclusions that online critics painstakingly detailed. More recently there was the Carnegie Mellon "study" purporting to link Internet use with depression -- another investigation in which the results were far less definitive than the press coverage warranted. Just this week we've been barraged with coverage of the latest in a series of "studies" of "Internet addiction," which tallies up a whopping 11.5 million Net addicts out there -- by adopting a comically loose definition of just what constitutes a Net junkie.


Everyone is starved for research about the Internet, of course -- that's what keeps dozens of consulting firms in business, and allows them to charge thousands of dollars for often-dubious reports cobbled together out of surveys and guesswork. But the pressure to put numbers on Net behavior continues to lead to questionable conclusions.

The latest and subtlest example arrived in Monday's Los Angeles Times, in
a study that claimed that the World Wide Web is "contracting." Well, OK, Charles Piller's article doesn't call itself a "study"; rather, it's based on "research conducted for the Times by Web tracking companies."

The conclusion, trumpeted in the headline "Web Travelers Follow Beaten Paths to Similar Sites," is that the Web universe grows narrower by the day: More and more of people's time on the Web, it seems, is spent at a small handful of leading sites.


It sounds like a dire situation for those of us who have cherished the Web's openness -- or, in the article's words, its reputation as "the ultimate exponent of diversity and choice." And, in fact, unlike so many previous "studies," the L.A. Times piece doesn't suffer from any blatant methodological flaws or obvious factual gaffes.

But it's still off base. Here's why.

The L.A. Times article reports that, as of June, users spent 35 percent of their time on the Web on the "50 most popular sites" -- "up sharply" from about 27 percent a year earlier. Now, this information is supplied by Media Metrix, a company whose Nielsen-like methods of surveying Web use remain highly skewed in a variety of ways.


But let's put aside lingering questions about Media Metrix's approach, which tends to downplay Internet use at work. Assume, for the sake of argument, that these numbers are good. What do they tell us?

The Times focused on that jump from 27 to 35 percent -- a significant number, to be sure. The statistic probably reflects a couple of trends. First, there's the continued consolidation among the big Web companies: When Yahoo acquires Geocities and, or America Online buys Netscape, their slices of the "top sites' traffic" pie inevitably expand. Second, there's the strong incentive commercial Web sites have to maximize time spent on their pages and clicks per visit: These companies are devoting all of their considerable ingenuity and resources to push this number higher.


To me, though, the message these numbers shout is that despite everything -- despite the vast sums being spent to advertise Internet companies offline ($1 billion projected for this year's fourth quarter, according to the Wall Street Journal), despite the enormous stock market attention focused on the big portals, despite the vast energy being devoted to making commercial sites "sticky" for users -- two-thirds of Web use is still eluding the "50 most popular sites"!

My headline, in other words, wouldn't be "Web Travelers Follow Beaten Paths"; it would be "Web Travelers Continue to Go Their Own Way."

The L.A. Times is not alone in its vision of a "contracting" Web. Piller's article echoes a June story in the New York Times, based on a study from the Xerox Palo Alto Research Center: That report found that "the most popular Web sites command by far the biggest share of Internet traffic," with the "top 119 sites" receiving 32.36 percent of all visits. It painted the Web as a medium in which popular sites can parlay their existing traffic into more traffic -- or, as the Times' John Markoff put it, a place where "the rich are getting richer."


Yet none of this should be shocking news: These are examples of "network effects." Once you have a lot of traffic, it's much easier to funnel that traffic to new offerings and services. In other words, if you're Yahoo, and you've got millions of people coming to your existing sites already, you can start up a new auction site and rest confident that you can point tons of people its way.

Still, the site that got Yahoo, Amazon and every other big Web company to start thinking about auctions was a start-up company called eBay that essentially, in true Web fashion, came out of nowhere. Genuinely new ideas, services and business models still tend to bubble up from below on the Net. Before they become standard features on every portal, these innovations help launch major new sites and companies and change the makeup of that charmed inner circle of "top 50 sites."

Though I may see a half-full glass where they highlight the emptiness, I wouldn't dismiss the kind of research reflected in the L.A. Times and New York Times stories -- there's value in tracking these usage trends. I just wouldn't rush to the conclusion that the Web's fundamental nature is changing, that "diversity and choice" are on the wane, simply because the big sites are increasing their chunk of the overall traffic.


Let's assume the worst: Say that the commercial Web does indeed "contract" as a smaller and smaller group of portals and mega-sites continue to grow their share of total Web usage. Say that share even tops 50 percent. Even under this scenario, we're still left with a medium in which a huge percentage of what's being expressed on the Web has eluded the control of big companies and remains in the hands of individuals and smaller outfits. That remains a historically unprecedented level of "diversity and choice" compared to every preceding mass medium.

As long as putting up a basic Web site remains simple and cheap -- and it seems to be getting simpler and cheaper by the day -- the Web's ability to serve as a petri dish for new ideas and a vast commons for human interaction will remain unimpaired. What the "contracting Web" alarmists seem to be forgetting is that not everyone who puts up a Web site is hoping to strike it rich, build a massive corporation or aggregate billions of "eyeballs."

E-commerce is fine and dandy, but there are millions of people who simply want to use the Web as a medium for interpersonal communications. They don't expect their family photo page or fan site or special-interest bulletin board to make a mint. They don't care that their Media Metrix "reach" rating is infinitesimal, as long as the 10 or 100 or 1,000 people they'd like to reach have found their URL.

The good news is that, though individually such small site traffic doesn't amount to much, collectively it still accounts for a huge slice of the Web-use pie. The TV analogy, imperfect but telling, would be a universe in which two-thirds of viewership went to public-access programming. Far from a depressing indication of a Web sliding fast toward mass-market conventionality, these numbers define a Web that remains historically unique in its openness to the offbeat start-up, the eccentric enthusiast and the passionate individualist. As long as that individualist doesn't develop delusions of Disney-scale empire that will inevitably disappoint, the little guys and the big sites should continue to coexist just fine.

Scott Rosenberg

Salon co-founder Scott Rosenberg is director of He is the author of "Say Everything" and Dreaming in Code and blogs at

MORE FROM Scott Rosenberg

Related Topics ------------------------------------------