Enterprise Communicationsbell

CATV Posed No Threat to Telephony

by Colin Berkshire

The value of pairs of copper wires to the homes is minimal, and is declining. I see nothing on the horizon to change this long-term trend.

I think it has been shown that dedicated copper pairs from a home to a central office (or concentrator) cannot be pushed to speeds which will ever again be competitive. That is, nobody has any idea of how to push 100 megabits, 250 megabits, or a gigabit through a single pair of wires for any meaningful distance.

The model which seems to be the correct one is the cable-TV model where you run a high bandwidth cable into the home that can easily provide gigabit speeds. This cable serves a group of homes up to a “node” and then the nodes are connected to a high capacity switch via fiber. That is, coax for the last mile, and fiber back to the switching center. Coax trunk lines can host enough fibers to serve enough nodes that each node can be relatively small. Best of all, there are no dedicated pairs and no dedicated outside plant.

Back at Bell Labs in about 1980 we studied the cable-TV wiring model to see if it posed a threat to the Bell System. A lot of smart minds went into this study. We all concluded that cable would not be a threat to the Bell System for a number of reasons, most of which turned out to be wrong:

  1. We surmised that the cost of a broadband adapter would forever exceed the cost of a line card in a central office. At this time, broadband adapter cards were fiercely expensive, often costing $500 to $1,000 while Central Office line cards cost about $1,000 to serve 8~16 lines. Thus, the Central Office model supposedly had a cost advantage. We never foresaw the cost coming down to about $80 while Central Office line cards would actually increase in cost.
  2. We concluded that the bandwidth on a coaxial cable could not handle the traffic of many users. At 64 kilobits per second, and with coax carrying about 2 megabits, a coax cable would cap out at 15 phone calls and thus only be able to serve about 100 homes. Then, it became fiercely expensive to haul traffic back to a central office. We saw cable as only serving the last ¼ mile, which was not where the cost really was. We failed to forsee cable increasing to gigabit speeds, or that gigabit adapters could be build for a hundred dollars. Gigabit speeds dramatically and profoundly changed the economics of cable. It meant that a single coaxial cable could serve thousands of users, not just 100.
  3. We incorrectly believed that cable could not handle 2-way traffic in any practical way. At the time, all cable systems were strictly broadcast. Protocols such as CSMACD and later frequency agile modems were inconceivable. And, they were viewed as so complex that they could never be cost effective.
  4. We knew that the problem wasn’t really in the last ¼ mile. The problem was from this last ¼ mile to the central office, where cables of thousands of pairs were required. A cable with 1200 or 2400 pair of wires (and even 4800 pairs) were the norm. And, such cables were required to physically get traffic to a Central Office. Cable-TV had no way to install such cables.

What we profoundly missed was what left the regional operating companies in a distant second place in what is today’s communications market.

We failed to realize how inexpensively broadband modems could be made. We failed to realize how much bandwidth could be scaled up on cable (1,000 fold) while on copper the scaling was limited (100 fold.) And, we failed to realize that with high speed packet switching (AKA TCPIP) coaxial cable could be converted to fiber-to-the-node. Finally, we failed to realize that packet switching would be so inexpensive and cheap once it was committed to silicon that central offices would be unnecessary.

We were so locked into our thinking about how data switching would evolve that we couldn’t forsee the architecture that is common today in Cable-TV: gigabit coax to the homes on a shared coax buss, then nodes that would transfer the traffic to fiber for compact backhauling, and finally, packet switching that could inter-switch all of the fiber traffic in a compact, inexpensive switch that didn’t have the logic of a central office (such as providing dial tone or call waiting.)

In hindsight we really messed up. We failed to see a vastly superior architecture that was evolving and that is what we know today as the cable-TV (Internet) industry typified by Comcast, etc.