…and eight times more difficult to cancel.

There is not a particularly close relationship between internet connection speed and cost. Most of the money to install a subscriber loop cable, say, about 95%, goes to the installation, not to the cable, and it doesn’t much matter what kind of cable. What does matter is that a lineworker is up on a telephone pole, and if he falls, he may be badly hurt. Again, the capacity of an optical fiber is so great, and a cable can contain many optical fibers, that, short of an undersea cable, the capacity is practically unlimited. The backbone, of course, is trivial compared to the subscriber loop. Google will have little or no difficulty in immediately offering ten gigabits for its previous price, just to say Nyaa, Nyaa, Nyaa.

As I understand it, Comcast is proposing to charge about $8000 for a two-year contract, which involves building a new subscriber loop, and Comcast expects to recover the capital cost in that time. Comcast’s quoted rate for new installations is now twenty dollars per foot. On that basis, $8000 would work out to 400 feet. That doesn’t sound an unreasonable distance for a subscriber loop.

The point of unfairness is that, having paid his $8000, the subscriber does not get ownership of the 400 foot cable, and cannot use it with some other service provider. It’s a case of “heads, I win, tails, you lose.” The customer takes the financial risk of a new installation, but does not get the right to use it for free once it is paid for. In my family, we once spent nearly that much money for a single desktop computer, a PC AT back in 1984? when it first came out. The thing was powerful enough to run LISP and PROLOG, and could be used for artificial intelligence research. The results of the research were mostly negative, in that a lot of algorithms turned out not to converge. But once we had had bought the machine, IBM did not go on charging so much a month.

There simply isn’t a satisfactory use-case for a Gigabit as against, say, 20-50 megabits, and if you go about it the right way, you can get 20-50 megabits out of existing copper infrastructure, which is already in the ground, and which is already paid for. That said, there is no assurance that a Comcast subscriber would be willing to go on paying a premium price beyond the contract period, or that a subsequent occupant would be interested, or that neighbors would also want to adopt Gigabit. Comcast wants to make money from its existing network, not to build a whole new network on a speculative basis.

There have been a couple of cases in the press involving IT workers who wanted to telecommute from places far out in the country, and could not get sufficiently fast internet access. These cases are a bit suspect. No one can examine and manipulate, say, fifty gigabytes directly. You have to have software tools, and these software tools can be designed to operate by remote control, on a computer located in such a way as to have fast internet access. The remote-control connection involves manageable quantities of data. The first megabit is the one which matters the most, and after that, successive levels of diminishing returns set in. The criticisms we should be making about telephone and cable companies should have to do with the terms on which they provide (or fail to provide) the first megabit. What we should be complaining about is that the first thirty or forty dollars a month, spent on telephone or cable, provides essentially nothing in the way of internet access.