Please Hold For New Technology

The basic science is in hand; costs are not

Donald Green spends his days selling the kind of high-speed communications gear that is helping pave the Information Superhighway. But at night, when he tries surfing the Internet from his home in Santa Rosa, Calif., he finds that the out-of-date phone circuits serving his town, 50 miles north of San Francisco, won't let his modem go any faster than 2,400 bits per second--too slow to make Net cruising worthwhile. So, says Green, when he hears that the I-way is just around the corner, "I kind of laugh a little. You have to recognize the limitations of capital and existing plant."

Limitations? In the New Age of telecommunications? But weren't microchips and optical fibers supposed to make all of these little problems go away?

Not quite. Sure, the science for the I-way is well in hand. In March, for example, three labs announced that they had managed to pump 1 trillion bits of data per second over a single hair-thin optical fiber. That's enough to carry 12 million simultaneous conversations--or download 100 two-hour movies in a second. For the airwaves, companies such as Analog Devices Inc. in Norwood, Mass., are developing chips for cellular-phone base stations that can understand calls in any format just by switching to the right decoder program. They call it "software radio."

But paying for this kind of technology is another matter. Time after time, communications companies that sketched out bold plans for extending advanced digital networks to consumers have retreated when it became clear that they could not do so profitably.

Case in point: Around 1990, the phone companies were going to snake an optical fiber to every home. But that would have cost well over $1,000 per home. So they decided to extend fiber only partway and use coaxial cable for the rest. By last year, that, too, seemed a bit rich, so a few phone companies began thinking of wringing higher speeds out of their existing twisted-pair copper wiring, which they would leave in place for the "last mile" to the home.

Cable operators got cold feet, too. Two or three years ago, they announced plans to install millions of digital set-top boxes. Most of those plans have been cancelled or postponed as the operators began to realize how much it would cost to buy the boxes and upgrade their networks to use them. Some operators would have to boot analog channels off the air to make room for digital, angering viewers. Now, Scientific-Atlanta Inc. has cashed in on the resistance to digital by selling more than 900,000 "advanced analog" boxes that enable such features as extra-informative program guides. They cost operators $150 to $200, vs. more than $500 for a digital box.

CHEAPWARE. Lately, cable operators have de-emphasized interactive television--a costly scheme that depends on banks of computers and disk drives to serve up movies and home shopping. Instead, they're jumping on the Internet, ordering speedy cable modems. But those will require upgrades of cable systems, too.

The technology retreat, however, may be ending. The cost of cutting-edge gear such as the optoelectronic components for fiber networks is dropping. And now that the deregulation bill has passed, companies have a far clearer picture of where they can invest in such technology and expect a profit.

For example, prices are falling for high-speed data connections to the home--higher-speed ones, anyway. Donald Green, the Santa Rosa resident with the bad phone line, is chief executive of Advanced Fibre Communications in Petaluma, Calif., which sells boxes that bring digital signals closer to rural homes. Typically, rural phone systems have long runs of wire between switches and homes. That distance wreaks havoc with high-speed communications signals. Using Advanced Fibre's device, which costs $4,000 plus $150 per line, phone companies can upgrade rural lines so customers can use high-speed modems or subscribe to ISDN (integrated services digital network), a data service that operates at 128 kbps.

Meanwhile, companies such as Aware Inc. of Bedford, Mass., are pushing technology that would let phone companies squeeze more capacity out of existing twisted-pair phone wiring. Asymmetric digital subscriber line (ADSL) sends megabits of data over ordinary twisted-pair phone wiring by exploiting frequencies as high as 1 million cycles per second. Signals on copper wires lose power rapidly at those frequencies, so ADSL requires sensitive signal-detection chips. Aware uses detection methods it developed for the U.S. military and spy agencies. To hedge its bets, it is testing versions that work over coaxial cables. Says CEO Jim Bender: "We're supplying armament to both sides."

DATA BROWNOUTS. Of course, once the high-speed capacity is in place, people are sure to use it. Even now phone companies are beginning to see that the Info Highway is going to be a lot like the concrete kind--the more you build, the more people will get on, and the more traffic congestion will result.

Already, the explosion of Internet use has pointed up the shortcomings of a phone system designed for voice calls when the online crowd plugs in. Pacific Bell, for instance, offered ISDN service with no per-minute charge on weekends, only to learn too late that Web surfers who use it to reach their Internet access providers tend to stay on for hours.

To keep Internet traffic from knocking voice callers out of the box, Pac Bell and others are having to spend hundreds of millions of dollars beefing up their switches and the trunk lines between them. The phone companies see asynchronous-transfer-mode switches as their ultimate salvation. For efficiency, ATM breaks messages into packets and stuffs data packets into the spaces between voice and video packets. But with so much invested in their conventional switches, phone companies are likely to keep voice calls on those devices for years to come.

The Net has its own technical limitations. To cope with exponential traffic growth, companies and Internet operators keep adding routers--specialized computers that read packets' addresses and send them on their way. But that means that a message or a Web page may have to hop through 20 or more routers to reach its destination. Routers can spend more time figuring where to pass the hot potato of a packet than doing the actual passing.

That's one cause of Internet "brownouts," in which response times become intolerably long. NetStar Inc. of Minneapolis is pushing a solution: an Internet redesigned for fewer, more powerful routers, so packets take fewer hops. Something has to change, says Mark E. Garver, NetStar's vice-president for sales and marketing: "The U.S. Internet is about as reliable these days as the phone system in Russia."

Those are hardly the only hurdles. There's also security and the ability of diverse networks to work together. For the phone companies, one of the biggest problems is their antiquated computers for call-handling, maintenance, billing, and the like. It costs a company such as Nynex Corp. some $200 million to $300 million a year simply to maintain mainframe software programs, which are in some cases decades old. The way it's stored now, a phone company's treasure trove of data on customers is largely inaccessible.

That problem, like the others, will eventually be solved because there's an economic incentive to solve it. But only when the pain of inaction becomes greater than the pain of action.

Before it's here, it's on the Bloomberg Terminal.