Internet Services: A Market for Bandwidth or Communication?

The author is aware of the various studies that have been done considering the pricing for bandwidth, but he asserts that the customers in the market are not buying bandwidth, but information content.

That's not to say that bandwidth is unimportant, as it impacts the quality of the product, but this is too far removed from the customer for them to consider it directly: the cable television subscriber whose programming is choppy is irate because he's not able to see the program, not because his provided did not provide him with adequate bandwidth (though that may be the root cause).

Given that federal funding for the Internet is being (at the time) phased out, the cost of building and maintaining the network must be borne by users, who will pay directly for the service they receive, and demand a quality of service that makes them feel it's worth the price - or they will move to another provider, or cease to use the service at all.

Multi-Part Tariff: Access, Capacity, Usage, and Congestion

The author presents the demand curve for the Internet, expressing that consumers as a whole will consume greater quantity (capacity) of service at lower prices, and that as price increases, the quantity demanded will likewise decrease. From the supply side, the marginal revenue per unit of capacity supplied decreases as the quantity increase.

One problem that is encountered is in the fluctuations in usage: a supplier cannot run his network at 100% capacity all the time - he must "sell" only a portion of his capacity to have excess capacity for peak use periods. Much ado is made about this, but the bottom line is that the cost of supply remains fixed: there is a certain level of "excess" capacity that all suppliers will have - and if every supplier restricts his sales proportionally, then the differences among suppliers (and to quantity supplied) should be largely moot.

The author draws a comparison to passengers getting "bumped" from airline flights, and how airlines must plan to minimize this occurrence if they wish to retain customers against competitors who will set their booking systems to be more accurate in terms of capacity usage, and industry standards and best practices arise such that all suppliers use the same equations.

To price or allocate bandwidth, the author suggests a multi-part tariff that considers four characteristics"

  1. Access - The times at which the user is connected to the system.
  2. Capacity - The amount of data that the system can move (regardless of actual usage levels)
  3. Usage - A per-unit charge for the data a user sends through the system
  4. Priority - The charge for giving the user's traffic priority in times when the network is congested

Liability for Transmission Costs

Since transmission takes place over multiple networks, the fee paid by customers should be divided among the suppliers, in accordance with the proportion of the network they provided for any given transmission. This would be weighted by the capacity of the lines as well, such that if a signal travelled equal distances over T1 lines and T3 lines, the owner of the T3 lines would receive proportionally higher compensation for the use of his network capacity.

How Are Networks Similar or Different?

Meatspace analogies exist for data networks: you can consider the flow of electricity across the nation's grid of power lines, or the flow of passengers among the airports. In all cases, there is a "network" that consists of a number of arcs (cables from one point to another) and nodes (points at which data moves from one arc to another).

There is also the matte of capacity and storage: capacity on a network is fundamentally different than inventory of physical goods: it must be used when available. It may be possible, for example, to have a storage reservoir on a network of oil pipes to generate a reserve of fluid when demand is low, but it cannot be pumped out at greater than the outbound pipe's capacity at a later time. Also, not all goods on a network can be warehoused: electricity must be used, or it is lost.

The author compares the capacity of different networks: mail, electricity, data, telephone, transportation, and oil, but he's already made his point.


The author concludes that pricing on the internet should be geared to both the value of the transmission and the value of the data received (though he's written far more about the former than the latter)