|N E G R O P O N T E||Debunking Bandwidth: From Shop Talk to Small Talk|
When I was an Assistant Professor of Computer Graphics at MIT in the late '60s, my career had little meaning at a dinner party. Computers were totally outside everyday life. I recall one Boston Brahmin who thought that a joy stick was a sex object.
Today, I hear 60-year-old tycoons boasting about how many bytes of memory they have in their Wizards, and the capacity of their hard disks. Others talk half-knowingly about the speed of their processors (thanks to "Intel Inside") and affectionately (or not) about the flavor of their operating systems. I recently met one socialite who provides consulting services; her business card reads "I do Windows."
Bandwidth is different; it remains a mystery to most. This is true because we often have too much when we don't need it or too little when we do. In addition, we scarcely understand the trade-off between bandwidth and intelligence.
If computer companies were the only players in our wired lives, we would experience a greater tendency to compute (apply intelligence) at the periphery of the network rather than shipping bits back and forth in wholesale fashion. The computer culture has learned from human interface research that the most supreme form of interaction is the lack of it. Less is more.
Fire Hose Providers
Telephone and cable companies have a different view. It is in their interests to ship as many bits as they can. Look at the fax machine, a perfect example of channel capacity (albeit limited to 9,600 baud today), which allows us to ship pages in exactly the wrong way. Had the world been saddled with the 110-baud rate of the Teletype (requiring about 20 to 30 minutes for transmission of a one-page facsimile), ASCII and page description languages (PDLs) would have prospered, thereby avoiding the extraordinary nosedive of computer-readable information in the past decade. I actually have heard a sophisticated computer scientist suggest the facsimile storage of books, newspapers, and magazines for shipment via gigabit pipes. This suggests an ignorance of the value of computer readability and an allergy to hard problems such as intelligent PDLs.
The Phone Companies Believe Their Own Arguments
Judge Greene made a terrible mistake when he barred the Regional Bell Operating Companies (RBOCs) from entering the information and entertainment industries. It has taken almost ten years to correct this error. Ironically, the RBOC lobbyists used a gratuitous but effective argument to get into the game. They claimed that unless they became content providers, they could not justify the enormous cost of a new infrastructure (read: fiber).
The argument worked. But now some of the telephone companies are forgetting just how specious it was: We don't know what to do with that bandwidth. We are staring at a $60 billion installed telephone plant of copper and fiber that offers enormous untapped opportunity. Worse, the Clinton administration is buying the wholesale need for, and provision of, bandwidth to maintain a major competitive edge without recognizing what Mother Nature and commercial imperatives already provide. More bits per second is not an intrinsic good. In fact, more bandwidth can have the deleterious effect of swamping people and of allowing machines at the periphery to be dumb.
Two Paper Cups and a String
I am fond of using the example of a wink as a form of massive data compression in human-to-human communication between intimate friends. In effect, this is one bit transmitted through the ether that could require at least 100,000 bits to explain to a third person. At that compression ratio we could transmit more than ten channels of NTSC television over a 300-baud modem.
There is a tendency to think of the trade-off between bandwidth and intelligence as merely a matter of computer cycles in the transceiver. But the transceiver should also contain knowledge of the signal. A simple example: Store all the static video information from, say, 50 movies on a CD-ROM (by itself a useless disc) then later, on demand, use ISDN to squirt 64 Kbits into this memory to reconstitute any one of these movies by delivering only the motion or other inbetweening data.
Nature's Role in Copper Versus Fiber
Few people know how good copper twisted pair is. Asymmetrical Digital Subscriber Loop (ADSL-1) can provide 1.544 Mbits per second into, and 64 Kbits per second out of, 75 percent of American and 80 percent of Canadian homes. ADSL-2 runs above 3 Mbits per second and ADSL-3, above 6 Mbits per second. ADSL-1 is fine for VCR-quality video.
Which would you prefer: 500 channels from which you can choose one, or one channel that can be switched to any source on the network?
It is absolutely true that fiber delivers thousands, in fact, millions of times more bandwidth. Frankly, we don't really know the limits of fiber. In addition, fiber now costs less than copper - when lines are updated, fiber will be used, with or without a need for bandwidth. Therefore, fiber will come into being automatically through the forces of common sense and Mother Nature.
Is That Soon Enough?
Dates like the year 2005 or 2010 are frequently heard estimates of when fiber will pervade the world, given appropriate investments and incentives. However, without any new incentives, telephone companies update three to five percent of their existing infrastructures each year. Some cable companies are proposing updating 80 percent of their plant in less than five years.
But here is the punch line: Why are we worrying about billions of bits per second into the home when we haven't used 1.5 to 6 million bits per second creatively? Yes, I will need those billions when I watch holographic television or expect a can of spinach to be teleported into my home. But in the meantime?
Dear telephone companies, now that your argument prevailed, please take advantage of your installed base of copper twisted pair, which can provide so much more than you are telling people - including video on demand, which is really in demand.
Next Issue: The Small Vision of the Set-top Box
[Back to the Index of WIRED Articles | Back to Nicholas Negroponte's Home Page | Back to Media Lab Home Page]
[Previous | Next]
[Copyright 1993, WIRED Ventures Ltd. All Rights Reserved. Issue 1.03 June 1993.]