Remembrances of Things Past

By: Scott Bradner

Long, long ago, in a hype not far enough away, some advocates of nuclear power claimed that it would be "too cheap to measure, " the proverbial free lunch. There turned out to be a few implementation details between the promise and the reality.

In the same vein, here is a quote from a thread on the com-priv mailing list last week. "the Spectrum and Bandwidth is, or will be, so abundant it will almost be more expensive to charge for it than to give it away." The poster went on to quote George Gilder's "Life After Television". "The limitations of the air, and even of the coaxial copper wires that carry cable television, have given way to the unlimited bandwidth of lasers and glass. Two gigabytes a second can be transmitted in a second [sic] down fiber-optic lines. Fiber optic glass wires the width of a human hair could potentially bear billions or even trillions of characters of information per second."

Now I will admit to not having read this work so I do not know to what degree the quote has been taken out of context, but it does seem a bit of a stretch to go from 'fiber optics can carry a lot of data real fast' to imply that the cost to the user for the transport of data will tend towards zero. There is this little detail of installation cost. I heard an estimate a while back that it would cost more than $200 billion to wire every house in the U.S. with fiber, and that includes the effect of the economies of scale. (How come people talk about 'wiring' with fiber? I guess its like talking about 'dialing' a telephone.)

So lets see, $200 billion divided by about 100 million homes means about $2000 per house. Amortize that over some number of years but add in the cost of money and maintenance and it might come out to $120 per year or $10 per month. Hey, that ain't too bad. But wait a minute, not everyone will be using the fiber so lets do a fudge factor for the idle fibers that still have to be paid for and push it up to $12; well how about $13 per month?

Still not too bad. Oops, I forgot that someone might want to actually use it so I guess we need some electronics to go along with the fiber. In my experience this costs somewhere around half of the cost of the fiber itself, with maintenance and management another half, so we will have to double the rate to about $26 per month. OK, I can live with that. Darn!

This is the U.S., we can't amortize over 20 years, Wall Street would never stand for it -- and what if some direct broadcast satellite technology came along and obsoletted all of this investment? We better cut that to about 7 years. Humm, now what does that do to the cost. The $12 becomes at least three times that. Come to think of it, you can't amortize the equipment over more than 3 or 4 years at the outside, so we will have to up the rate or return on that part. (pause for some hand waving). Gee, that's now up to $70 or $75 per month for the connectivity. Now we have to add the cost of the services that the user subscribes to. I've read that the average

Cable TV subscriber gets 2.5 or so 'Premium' channels (I put Premium in quotes because that is very much in the mind of the beholder.) That gives a final cost of $90 to $100 per month. Gee that's a bit steep I guess everyone would not want to pay this level of tariff so the penetration would be less which would reduce the economies of scale so the individual cost would go up. Do I see a Catch-22 here? Not exact calculations but close enough for column work. So much for the free bits.

The thought was fun while it lasted. It is always nice to step back to the fairy tales of ones childhood. Now that reminds me of ATM, but that is a story for another night.

disclaimer: Although Harvard knows full well from recent experience that there is a substantial cost to fiber-based network installation, the above are all my miscalculations.