The death of telephony

Proclaiming “multimedia convergence,” “interactivity,” “intelligent networks,” “electronic yellow pages,” “caller ID,” “500 channels other than pay-per-view” and invoking vistas of “high-definition television”, seers in telephone and television companies give stirring speeches about the future. Contemplating their revenues of tens of billions of dollars (see chart), their laboratories full of new technology, their millions of mostly satisfied shareholders and customers, their multiplying masses of trade publications and cover-stories in national magazines, telephone and television executives all too often seem unaware that their basic technologies are dead.

Such concepts as “high definition”, “interactivity”, “intelligence networks” or “electronic yellow pages” are merely cosmetics for the corpse. In order to see the future of telecommunications, it is necessary to lug the dead body out of the way first.

Life and death in technology is a matter not of revenues but of prospects. Vacuum tubes reached a pinnacle of sales in the 1970s. Revenues from telephones and televisions are currently at an all-time peak. But the industries organized around these two machines will not survive the century. Still, telecoms executives want to believe that televisions and telephones can evolve bit by bit into the new digital world, with the same companies supplying essentially the same products and offering essentially the same jobs.

This dream cannot come true. Even to talk of “telephones” and “televisions” reflects a lexicographic lag that prevents many business leaders from detecting the onset of rigor mortis in their still-profitable products. In coming years, the very words will ring as quaintly as “horseless carriage,” “icebox” or “picture radio,” today.

Dying industries exude an increasingly fetid air. It is not just that some 100,000 acts of television violence are sen by the average American child before the age of 13, or the obtuse denial that such a diet could affect behaviour. Or the perverse drive by television magnates to ensure that the new American HDTV standards preserve the interlaced TV screen (alternately filling every other line) that is wholly unsuited to computer text or multimedia. It is not just the increasing amounts of broadcast advertising needed to support diminishing amounts of substantive programming. Nor yet the continued lobbying by American telephone companies against any encroachment by cable television in their domains, accompanied with constant trysts and cohabitations with the same cable films on foreign shores, or even in distant states.

Closer to the heart of the matter is the constant celebration of fibre optics by telephone executives in public forums, accompanies by the continued installation of millions of miles of the old twisted-pair copper wire every year. The very same trumpets that blared ineffectually for ten years for ISDN (integrated services digital networks) now toot for an amazing new redeemer of twisted copper called asymmetrical digital subscriber loop (ADSL). As developed by a tiny California company called Amati and adopted by Northern Telecom, ADSL sends a stunning 6m bits per second of full-motion video down a conventional telephone line.

As a technological breakthrough, ADSL is stupendous. But it resembles the varistor, a brilliantly crafted device excogitated in the late 1960s by the vacuum0tube people at RCA int he face of the new threat of silicon transistors. Even in strategic terms, it is a distraction. ADSL provides a new weapon for a lowing fight with cable and satellite first sending hundreds of channels of pay-per-view films and games to the household. A dying telephone technology clutches at a shiny new broadcast-industry death rattle.

As the voices die away

Telecommunications is in a transition that is too fundamental for any such quick fixes. It is moving towards an era in which person-to-person communications will give way to links between computers. And although imaginative writers persist in using anthromorphic language to describe computers, in fact they have virtually nothing in common with people. People have associative memories that are millions of times larger than computer “memories”; two eyes can do more image processing than all the supercomputers in the world put tougher. But in terms of communications bandwidth, people lag hopelessly behind computers, with nearly a billion times less communication power than the best machines.

Almost every feature of the telephone network, from its 4kHz wires to its circuit-switched lines, is designed for the sluggish human voice (which communicates at about 55 bits per second). Telephones give us dysphasics what we need: a very small bandwidth connection for a relatively long time. Computers, by contrast, need huge bandwidth for microsecond bursts. Their networks transmit digital data at a minimum rate of some 10m bits per second, going up soon to 155m bits a second. On such a digital flood, 64 kilobits per second of voice can ride as an imperceptible trickle.

Data already comprise half of the bits in a telephone network and account for 20% of the profits. Data income is growing six times as fast as voice income. As the telephone network become a computer network, it will have to change root and branch. All the assumptions of telephony will have to give way. thus telephony will die.

Television faces a similar problem. It is a broadcast analogue system which assumes that all people are essentially alike and at any one time can be satisfied with a set of 40-50 channels. In Europe and Asia, 40 or 50 channels may seem wretched excess. But compare this array to some 14,000 magazines and a yearly output of some 55,000 trade books published in America alone. Television defies the most obvious fact about its customers: their prodigal and efflorescent diversity. It ignores the fact that people are not inherently couch potatoes; given a chance, they talk back and interact. People have little in common with one another except their prurient interests and morbid fears and anxieties. Aiming its fare at this lowest-common-denominator target, television gets worse and worse year after year.

Computer networks that respond to all the human characteristics that television networks defy. Computer nets permit peer-to-peer interactivity rather than top-down broadcasts. Rather than a few “channels,” computer networks offer as many potential connections as there are machines linked to the web. Rather than a system in which a few “stations” spray images at millions of dumb terminals in real time, computer networks put the customer in control. Television will die because it affronts human nature: the drive to self-improvement and autonomy that lifted the race from the muck and offers the only promise for triumph in our current adversities.

Confidence in the new paradigm, however, does not spring only from the desire for a better culture, and it cannot be stemmed by some new global plague of passivity and tube addiction. Propelling the new order is the most powerful juggernaut in the history of technology: an impending millionfold rise in the cost-effectiveness of computers and their networks.

Early in the next decade, the central processing units of 16 Cray YMP supercomputers, now costing collectively some $320m, will be manufacturable for under $100 on a single microchip. Such a silicon sliver will contain approximately one billion transistors, compared with some 20m in currently leading-edge devices. Meanwhile, the 4kHz telephone lines to America’s homes and offices will explode into some 25 trillion possible hertz of fibre optics.

At the same time, the supposedly scarce realms of the radio-frequency spectrum will open up to a series of innovations that make communications power (bandwidth) as cheap and abundant in the air as it is in wire today. Microcells using a protocol called Code Division Multiplication Access can use the entire radio-frequency spectrum every few miles or even hundreds of yards. Billions of hertz of little-used spectrum are available in the microwave domain and can be used for television broadcasts or computer networks.

All these developments converge in one key fact of life, and death, for telecommunications in the 1990s. Television and telephone systems–designed for a world in which spectrum or bandwidth was scarce–are utterly unsuited for a world in which bandwidth is abundant. The key strategy of both systems has been to centralise intelligence in local central offices, cellular base stations, cable-television nodes, and broadcast centers, and give the user a stripped-down commodity terminal, whether a telephone or television set. For telephony, this meant making up for scarce bandwidth with powerful centralised switches. For cellular phones, it meant making up for scarce spectrum with high-powered narrow-band radio phones and smart cell sites. For television, it meant making up for scarce bandwidth with powerful centralised transmitters.

In all these cases, intelligence at the centre made up for a lack of bandwidth and computer power on the fringes of the network. But with new bandwidth galore in fibre and air, and video supercomputers on the way for under $1,000, all these structures are obsolete. Over the next decade, engineers will use bandwidth and computer power on the edges of networks as a substitute for switching and intelligence at the centre.

The computer paradigm will prevail. Just as the 1980s brought the collapse of the centralised scheme of a few thousand mainframes and millions of dumb terminals, the 1990s will see the collapse of similar structures in television and telephony. First to fall will be the broadcast system of a few thousand stations and a few networks serving millions of idiot boxes. Next to fall will be the telephone scheme of a few thousand local central offices serving millions of dumb telephones.

Telecosmonauts and their laws

Governing this transition will be the two key laws of modern digital electronics. One, the law of the microcosm, holds that the more transistors are linked together on single chips the more coolly, swiftly, and cheaply they function. Measured in power=delay product (roughly, transistor speeds times their heat dissipation, the efficiency of transistors has risen exponentially for two decades as their size dropped and chip densities rose. Judging by recent new from the laboratories–describing such exotic stuff as single-electron electronics and electron-spin transistors and such practical advances as three-dimensional and holographic memory devices–the trend is likely to continue through the next two decades. This will ensure a continued onrush of low-cost computer intelligence on the edges of all networks.

Now the law of the microcosm merges with the law of the telecoms. Just as the law of the microcosm essentially showed that linking any number n of transistors on a single chip leads to n^2 gains in computer efficiency, the law of the telecoms finds the same kind of exponential gains in linking computers: connect any number n of computers and their total value rises in proportion to n^2.

The result is twofold: the cost-effectiveness of individual computers measured in MIPS (millions of instructions per second) per dollar approximately doubles every 18 months and the value of computers in networks rises as the square of the rise in the number of networked machines. In a top-down network, such as a conventional telephone or cable system, attaching a new device may burden the central switch or head-end; in broadcasting overt the air, each additional receiver has no effect on the technical power of the system. But in a peer-to-peer computer arrangement, Each new device isa resource for the system, expanding its capabilities and potential bandwidth. The larger the network grows, the more efficient and powerful are all its parts.

Gathering irresistible momentum over the next decade, these forces will blow away at the old analogue establishments of television and telephony. Both can survive only to the extent that they transform themselves into digital computer networks.

There will be little time to spare. Judging from the sales of microprocessor central processing units so far this year, some 50m computers will be sold worldwide in 1993 at an average price close to $1,500. That price will continue to fall, sustaining a continued pellmell pace of sales far exceeding the sales of analogue televisions or telephones. Meanwhile, American experience offers a portent of the future of networks. Between 1989 and 1993m, the proportion of computers in America connected in networks rose from under 10% to over 60%.

In sum, both television and telephony suffer from information hierarchies that are totally unsuited to the coming era of billion-transistor chips and terahertz nets. These digital computer networks will function both over wires and in the air. Indeed, the most common personal computer of the next decade will be based on the digital cellular telephone. Called personal digital assistants, among many other coinages, they will be as portable as a watch and as personal as a wallet. They will recognize speech and navigate, open the door and start the car, collect the mail and the news and the pay-cheque, connecting to thousands of databases of all kinds.

As for the descendant of television, the dominant traffic force the future will be store-and-forward transmission of digital data among millions of telecomputers. These machines will be capable of summoning or sending films or files, new stories and clips, courses and catalogues anywhere in the world. Whether offering 500 channels or thousands, television will be irrelevant in a world without channels, where you can always order exactly what you want when you want it, and where every terminal commands the communications power of a broadcast station today.

The same couch potatoes who, in the absence of a better choice, now settle for a Donahue show on lesbian nuns or for a sanguinary stew of cops and prostitutes, will turn to favorite films or local sports or career education. Many of the same people who now sink into a passive stupor before the tube will find themselves using telemeters to travel around the world, taking courses, conducting transactions, and shaping their own programs and software.

The new computer networks will have virtually nothing in common with existing hybrid networks that combine analogue and digital functions. Telephone companies currently use their digital switches to relieve the pressures on their current copper wires. most television companies plan to use digital electronics to increase the choices of couch potatoes, allowing them to choose a pay-per-view film or play a video game without going out of the house. But end-to-end digital systems will bring something quite different, and a true paradigm shift.

As Richard Solomon of MIT’s Media Lab pointed out four years ago, the new broadband networks will reverse the current relationship between computers speeds and network speeds. At present, telephone networks work far more slowly than the internal communications “buses” of individual computers (the electronic pathways linking processors to memory and screen). Transferring data far faster internally than externally, computers are decoupled from the telephone network; they can legitimately be seen as outside the system, interconnecting to it only through complex modems. The new broadband networks, however, will dissolve this boundary between the computer and the network. As Professor Solomon puts it, “With end to end digitisation, the public switched network will be transformed into one large processor.” In it, computers may be able to tap remote databases more readily than they can reach their own hard disks or CD-ROM drives.

This vision of the network as one colossal processor goes well beyond mere metaphor. With the new generation of microprocessors, such as the DEC Alpha and the Silicon Graphics Cray-on-a-chip, computers are shifting form processing data in 32-bit clumps (“words”) to 64-bit words. This change means a 4-billion-fold rise in the ability of computers to address memory directly. Since few computers command storage above the current 4-gigabyte limit, this new increment of capacity may be irrelevant locally for some time. But it will let telemeters enter and interact with huge databases of digital video–tens of thousands of films, art exhibits, courseware, three-dimensional experiences and other possibilities bounded only by the reach of the mind and the span of the global ganglion of computers and cables, the new worldwide web of glass and light.

Within these miraculous mansions of imagination, however, one thing is clear. There will be no room for televisions or telephones, or for the companies that make them.