America Online's crash may portend constant crises unless the Internet is revamped

It was the day America Online Inc. became America Offline. For 19 hours on Aug. 7, the world's biggest online company and its 6 million customers were blacked out while technicians tracked down what they described as two crippling problems: a faulty roadmap of Internet addresses and a bug in the software of a powerful switching computer called a router. In an apology to subscribers, Chairman Stephen M. Case wrote: ``I would like to be able to tell you that this sort of thing will never happen again, but frankly, I can't make that commitment.''

That's for sure. Preventing blackouts may actually get tougher as online services such as AOL--and the global network of networks known as the Internet--run ever closer to the limits of their capacity. Outages at overburdened Internet service providers have already become dismayingly common. Says Allan H. Weis, CEO of Advanced Network & Services, which sold its Internet service business to AOL last year: ``Maybe for the first time in the history of the Internet, the demand is exceeding the supply that technology can deliver.''

The Net is hardly going to pieces. But the growing pains are so severe that stories about online screwups could easily carry this headline: ``Internet Becomes Victim of Own Success.'' And service failures are just one problem. Devoted Internauts can reel off a slew of other annoyances: Their favorite Web sites are always busy. Their brand-new modem delivers data like a Volkswagen Beetle on a steep hill. Every newsgroup they visit has been ``spammed'' with ads for get-rich-quick schemes. They're stymied by fruitless searches for information. They waste hours idly clicking from the Poisons Information Database to the Exploding Head page to Aaarrgh!

Yes, the Internet. You can't live with it, you can't live without it. On one hand, the Net aids scientific collaboration, builds virtual communities, and streamlines commerce--most of the time, without incident. In fact, a BUSINESS WEEK/Harris Poll of 1,186 people taken June 21-30 finds that 79% of those who use the Internet rate the experience excellent or good (page 66). But it's a long way from a mass medium. Today's users are early adopters who tend to be tolerant of glitches. The poll found that while 87% of the U.S. public has heard of the Internet, only half the overall group use a computer at home or work, and only 19% say they have used the Internet or the Web in the past year.

MORE TOLLBOOTHS. So the Internet still has some growing up to do. Security and reliability must be improved. Technical standards must be established faster, even if less democratically. Lines of control must be clarified. The Internet community must develop ways to settle debates over sensitive issues such as contract law, privacy, and pornography--or risk having governments step into the breach. Intellectual property must be protected--forget that dippy slogan bandied about the Net that ``information wants to be free.''

Above all, financial incentives for investment must come into line. Right now, customers who pay a mere $20 a month can blast the Net with untold megabytes of data, voice, and video. Without usage-based charges, service providers are called on to upgrade their infrastructure with no clear promise of a return on their investment. For that reason, today's Internet is destined to collapse under its own weight, predicts Edmond J. Thomas, executive vice-president of Nynex Science & Technology Inc., in Price Waterhouse's book, EMC Technology Forecast: 1997. It will reemerge, says Thomas, ``but there will be a lot more tollbooths on that highway than there are now.''

The Internet's chaotic present stems from its anarchic past. It was never designed to be an all-purpose, for-profit communications network. Its predecessor, Arpanet, was started in 1969 by the Defense Dept. to connect incompatible computers at universities doing military research. The Internet emerged in the 1970s to connect Arpanet with other networks. There was no need or way to bill for usage, since the government paid for pretty much everything. Besides, the Internet was designed strictly for data, which it breaks into packets that travel disjointedly, to be reassembled at their destination. That's a liability now that the Internet is handling live traffic such as phone calls and videoconferences, where unpredictable delivery times are intolerable.

That's no rap against the pioneers who designed the Internet more than 20 years ago. Phones and electricity fail sometimes, too. The Net has held up remarkably well considering its rapid growth: The number of computers connected to it worldwide is rising 72% a year, according to a recent survey by Network Wizards in Menlo Park, Calif. Operators have to switch out equipment on a monthly or even weekly basis. Such changes can have poorly understood ripple effects, says Columbia University computer science professor Yechiam Yemini. ``You don't change one of the engines under the wings of a jet with one 10 times faster and expect the plane to fly. It will crash,'' says Yemini.

While there aren't widespread blackouts, there are frequent brownouts. The Web is already derided as ``slow television,'' or the ``World Wide Wait.'' Imagine the traffic jams when couch potatoes can click through the Web with remote controls--and thousands decide to check out the Friends site during a commercial break from the TV show.

Computer experts have worried about Internet overload since its earliest days. But the concerns were easy to dismiss as long as the state of the art was well ahead of the demand. It isn't anymore. Advanced Network & Services' Weis points out that the speed of the Internet backbone rose 700-fold between 1987 and 1992. Another 700-fold increase in the next five years is hard to imagine. ``The big problems of the Internet haven't hit yet,'' says Weis. ``We don't know what they are.''

Locating what you want on the Internet is already a challenge, and threatens to get harder. The Internet is like a library with no filing system, one that's constantly receiving more books in random piles. Search engines like Yahoo! and Lycos create indexes of the Net's contents by sending software ``robots'' on search missions--rummaging through the growing piles again and again. That works O.K. on small databases, but 'bots can be mighty annoying on big ones. When they hit Los Alamos National Laboratory, some poorly programmed robots get lost inside a vast database of physics research, tying up the lab's computers by issuing up to 10 information requests a second.

DIRTY SECRET. The Internet is beginning to suffer from a lack of manners as well. One dirty little secret is that most phone calls and videoconferences ram their way past data transmissions by using a bully of a communications method called User Datagram Protocol, or UDP. Unlike the more polite Transmission Control Protocol, TCP, which drops back when it detects congestion, UDP continues at full speed, elbowing ahead of TCP traffic. Yet UDP customers aren't paying anything extra for their fast lane. What's to stop it from being abused? ``It's really basically altruism and peer pressure and people knowing each other,'' says Jeffrey K. MacKie-Mason, an associate professor of economics, information, and public policy at the University of Michigan.

Altruism is hardly a workable ethic for an Internet that has become huge, impersonal, and profit-minded. The Net's deficiencies are reflected in the lives of people like Rick Cunnington, a mechanical engineer from Chandler, Ariz. After discovering the Internet, he says, ``I jumped into it with both feet,'' using it to plan a vacation, exchange E-mail, and weigh investments. But when he researched the purchase of a water softener, all he got was product puffery. And he spent five hours over three days searching for historical stock data. Finally, he got off the Information Superhighway, into his car, and onto Interstate 10, down to the Phoenix public library. There, he spent less than an hour getting the information from microfilm. Cunnington's advice about cruising the Net: ``Take your sleeping bag.''

It's a bad sign when the likes of Cunnington get disillusioned. A survey by Mercer Management Consulting Inc. identified only two groups making up 17% of the population--the ``Wired Elite'' and ``Upgrade Families''--that account for two-thirds of the market for electronic commerce and network-based services. Says Mercer Vice-President Michael E. Smith: ``You're probably talking about a cap of 25% to 30% of the households that are really willing to step up to the plate to put up with the difficulties.'' Forrester Research Inc. in Cambridge, Mass., predicts that the average Web publisher will lose money until the year 2000. ``Content providers who joined the Web gold rush find themselves tumbling down a long, dark mine shaft,'' writes Senior Analyst Bill Bass.

CLEARER PATH. Internet skeptics--or perhaps realists--are redirecting their money away from the Net and toward private ``intranets,'' where they see a clearer path to profitability. These internal networks work like the Internet and can communicate with it but are built for a company's own employees, big customers, and suppliers. Intranets aid collaboration by pulling together data on incompatible computers. Zona Research Inc. in Redwood City, Calif., predicts that in 1999, spending on host computers and software for intranets will exceed Internet spending by 6 to 1 (chart).

The Internet, in contrast, suffers from topsy-turvy financial incentives because of its legacy as a government-subsidized enterprise. Big Internet service providers such as MCI Communications Corp. and Sprint Corp. have long accepted traffic from many of the smaller ones without demanding payment. But that practice has become onerous. So they're raising standards for admission to the club of ``peers'' that exchange traffic for free. All others must pay.

That change will spill over to the way carriers charge their customers. If one carrier is charged by others for sending them lots of, say, video signals, it will turn around and raise the fees for that kind of traffic. Such usage-based payment will discourage capacity-hogging traffic that doesn't make economic sense. And it will raise money for investment to support other services that do make sense--phone calling, perhaps.

The next step is to get the phone companies and cable-TV operators to invest more. Today, says Roger S. Siboni, national managing partner of KPMG Peat Marwick, the carriers that provide Internet infrastructure ``are fearful that they'll be segmented out of the value proposition--they'll be reduced to a commodity while others will add value.''

If phone companies don't hold back the Internet, standards wars might. Net standards have long been set by a collegial group of volunteer experts known as the Internet Engineering Task Force. But more companies with a stake in the Net's development are sending delegates to IETF committee meetings. That bogs things down because anyone who attends has a vote, no questions asked.

Meanwhile, standard-setting for the World Wide Web is beset by fighting between Microsoft, Sun Microsystems, and Netscape Communications. Tim Berners-Lee, the research scientist who is credited with inventing the Web and now directs the World Wide Web Consortium, says he worries that in their efforts to one-up each other with unique features, these rival companies could spoil things for their customers.

The battling creates headaches for people who need to place bets on which technologies will take hold. Trouble is, no one really runs the Internet. ``The framework that people cooperated in before is collapsing, and a new framework has to emerge,'' says Robert G. Moskowitz, a technical support specialist for Chrysler Corp., a big Internet user.

So the Internet is at a turning point. It has certainly impinged upon the public's consciousness--AOL's Aug. 7 crash made front-page news across the country. Moskowitz recalls that in the early 1950s, the American public turned against the space program after a Redstone rocket blew up on the launchpad on live television. ``There is a concern in the Internet technical community that we can lose the consumer. This is completely recognized.'' His prediction: ``We'll get things straightened out.'' The community has a powerful incentive to do just that. After all, worrisome as they may be, the Internet's problems stem not from decline, but growth.

By Peter Coy in New York, with Robert D. Hof in San Francisco, Paul C. Judge in Boston, and bureau reports


Updated June 14, 1997 by bwwebmaster
Copyright 1996, Bloomberg L.P.
Terms of Use