What we see in the current fiber-coax strategies is fiscal timidity, justified by the usage patterns of an old-line broadcast and publishing model, not the Net.
Message 28:
Date: 10.1.95
From: <nicholas@media.mit.edu>
To: <lr@wired.com>
Subject:2020: The Fiber-Coax Legacy
In 2020, people will look back and be mighty annoyed by our profligate insistence on wiring a fiber-coax hybrid to the home rather than swallowing the cost of an all-fiber solution. They'll ask, "Why didn't our parents and grandparents plan more effectively for the future?"
As far as the American home is concerned, the phone companies have the right architecture (switched services), and the cable companies have the right bandwidth (broadband services). We need the union of these: switched broadband services. But how do we get from here to there?
No one will deny that the long-term solution is to install fiber all the way, but the benefits seem diffuse and the costs acute. In the eyes of the telcos and cable companies, the question is financial - and since the near-term balance sheets don't add up, fiber is not being laid all the way. One way around this problem is to circumvent the private market and let a telecommunications monopoly build the infrastructure, which is exactly what Telecom Italia is doing. It has declared fiber to the home as its goal and will swallow the initial cost meeting this goal in the name of national interest. This is one of the few benefits of a government-owned monopoly: Italy will have a far better multimedia telecommunications system than the United States by 2000.
A private, profit-driven company cannot always do the right thing, because the payback may be too distant and uncertain. So, rather than rush headlong into an all-fiber future, US corporations have decided to mix fiber-optic and coaxial cable. With the so-called hybrid fiber coax, fiber trunk lines are brought to a point near a group of 500 to 2,000 homes and then patched into copper coax that is shared by the entire group. Both the cable companies and the phone companies have come to this same interim solution, more or less unanimously, but from different perspectives.
Why? Running coax is US$400 per household cheaper than bringing fiber all the way. Part of that expense is coping with and switching all that fiber. Part is the differently skilled labor needed. Part is the fact that TVs will need an adapter. But none of that $400, mind you, is the cost of the fiber, which, these days, is more reliable and cheaper than copper, even including the connectors.
So, the cost difference today between hybrid and pure fiber is $400 per household. That estimate was $1,000 two years ago and will probably be $200 in a year or two. If we base our decision not to run fiber on a number that is dropping so rapidly, have we really made the right choice? If what stands between me and fiber to my home is $400, I'll raise my hand and pay my share. I bet others would too. Maybe, in staring so hard at the bottom line, we are failing to remember what's really going on here.
Face it, the argument for a fiber-coax installation is an argument for incrementalism. Friends of mine have likened the hybrid solution to cellular telephony. Once a cell gets overloaded, the argument goes, it can be broken into subcells. By extension, when the fiber-coax solution does not meet demands, the fiber beachhead can be moved forward to serve, say, 100 instead of 1,000 homes. Well, that almost works.
But the hybrid solution makes two enormous assumptions about how people will use networks. One is that a home will be happy sharing a gigabit per second with as many as 2,000 neighbors. The other is that all homes will consume more bits than they generate. Both assumptions are flawed.
Bandwidth is a complicated concept because it combines instantaneous bursts and continuous throughput. As soon as you string together homes like Christmas tree lights, you then assume that each home will receive enough fast bursts to make the overall bandwidth look and feel fast. But it won't take too long for a few simultaneous users to gobble up a gigabit per second of bandwidth, especially as audio and video become commonplace on the Net.
The second and bigger problem is symmetry. This topic is hotly contested by cable and telephone companies, who don't believe consumers want to send out as many bits as they take in. Cable companies allocate the coax spectrum with copious bandwidth flowing into the home and precious little back to the head end. But that logic fails when you reconsider the notion of a head end. Where is the head end in a true, switched broadband system?
The Net has shown itself to be a heterogeneous collection of nodes, each of which can be a source or a sink, a transmitter or a receiver. As more and more people start entrepreneurial services from their home PCs, we will need symmetrical systems, designed without a "head-end prejudice." The assumption that the average American is a couch potato involved in nothing but consuming advertiser-supported bits is wrong and, frankly, insulting.
The use of bandwidth is generational. As soon as kids find the Net alternative, they spend less time watching TV. The number of Web sites is doubling every 53 days. These will increase, not decrease, and provide the basis for a huge nano-economy when we crack the nut of e-cash.
Andy Lippman, associate director of the Media Lab, has a nice way of putting it. When people take him to task about the real need for symmetry in future communications systems, he notes that it's already built into our current ones, reserved for the head ends, not the customers. But more and more people will want to be their own head ends. Our wiring and our consumption of new media are deeply interwoven. What we see in the current fiber-coax strategies is fiscal timidity, justified by the usage patterns of an old-line broadcast model, not the Net.
There is a way to do it right, and that is to provide fiber all the way to the home. Instead of wasting time justifying half-baked ideas, let's find ways to finance the solution. The Italians already have.
Next Issue: Being Decimal
All rights reserved.