The Economics of Layered Networks
The Internet began as the marriage of two government-owned networks operated by DARPA and the NSF, but beginning in 1993, private companies were invited to link their private networks to the backbone, and even to develop their own backbone networks that were "layered" with the public Internet.
Economics of Pricing under Differing Market Structures
General economic theory tends to have a good grasp on monopoly and perfect competition, but struggles to provide a consistent conclusion about equilibrium pricing structure in mixed markets.
Competition at Various Layers
Fundamentally, the cost of supplying internet access tends to be high: developing large dark-fiber networks is the demesne of a few, large companies, so the general principles of oligarchy would apply.
However, the cost of access from the MAE to the "curb" is considerably less, resulting in greater competition among local and residential providers, which more resemble a system of perfect competition, but one in which a significant portion of their costs are uniform, as they are supplied by the oligarchy.
As a result, some degree of equilibrium has been achieved, in that the unit cost has begun to approach the marginal cost of service provision for connections of various speeds: the price of modem connection is virtually the same from every provider. As a result, competition is fierce, and the market is highly uncertainty.
Term contracts have added an element of shared risk: in exchange for agreeing to a term of service, the customer receives a discounted rate for a fixed period of time. This reduces the risk to business of the customer leaving for another provider should the equilibrium price fall, but reduces risk to the customer of paying higher prices should the equilibrium price rise.
Especially in the case of the smaller ISPs, the term contract poses the highest degree of risk: they must pay a fixed price to their supplier, meanwhile depending on the market to set the price they may charge to their consumer.
At the time this book was published (1998), the authors suggested that the interplay of these factors will result in diminished competition - smaller competitors, without the financial resources to survive the squeeze in pricing, will drop out, resulting in a few large companies remaining.
EN: Ten years later, I'm not sure if it worked out quite that way: while the mom-and-pop ISPs with a T-1 line and a bank of modems in their garage have died out, there are still a handful of choices in each local market. What's more, it's not the same choices in every local market, nationwide, and the desire to have a single set of prices to be nationally competitive (and avoid customer outrage) has prevented separate cartels from forming in each local market, it does seem to have set a benchmark price, nationwide, that is inelastic to demand in the various local markets.