purple motes

The growth of the Internet has emphasized functional rather than structural aspects of networking. The end-to-end principle, the concept of “the Internet,” and widespread concern about “bandwidth of connections to the Internet” push into the background ownership interfaces between networks and the geographic structure of interconnection. One result is that opportunities to innovate at the edges conflict with network “pipe” innovation, i.e. the paradox of the best network.

Major industry trends have major implications for network geography. Municipal networks, such as wi-fi networks or open-access municipal fibre optic networks, are a rapidly developing form of network infrastructure. From a regional or national perspective, municipal networks make cities important elements of network structure. If you just understand these networks to be providing Internet connectivity, you miss that they connect residents of a city to other residents of that city in a distinctive network organization.

Network geography significantly affects the cost of providing network services. If communication bandwidth cost is distance-senstitive, then local caching reduces the cost of distributing content. That effect is particulary important for high-bandwidth content such as video. Even if you believe that bandwidth costs will rapidly go to zero irrespective of physical distance, transaction costs associated with providing services are likely to remain distance-sensitive. At any given degree of infrastructure ownership consolidation, the number of ownership interfaces are likely to increase with distance. Ownership interfaces are a source of transaction costs. In addition, customer behavior and customer service have local components. Local knowledge allows a service provider to respond better to (local) customers’ needs.

The economic geography of internetworking is starting to attract more attention. In an interesting presentation at the recent Firstmile conference, Mike Hrybyk discussed BCnet transit exchanges in British Columbia (if you’re wondering, that’s in Canada). These transit exchanges provide a low-transaction-cost environment for the exchange of network services, including peering of local users and user purchasing of network services from a variety of carrier suppliers. Research and educational institutions seeking to foster local network development and to experiment with innovative networks have led the development of these transit exchanges.

The Internet is wonderful. Future forms of internetworking can be even better. Recognizing cities as important structures for internetworking can help to make the Internet better.