The Economics of Decentralization: Mesh Networks & Economies of Span Nicolas Negroponte recently propheized that "everything you assumed about telecommunications is about to change," because of the potential combination of P2P and WiFi (802.11). He envisions WiFi eclipsing GSM in the US (maybe) and WiFi nodes acting as hubs to pass messages between themselves, P2P style, to create a mesh network that circumvents the need for wireline metro and long-haul transport (not likely). Its a beautiful vision, reminiscent of the early ISP days, but the architectual economics are not feasible.
Every time a message hops from one node to another, that hop introduces latency, degrading service quality. Scaling such an architecture beyond a hotspot node or two in the nieghborhood isn't feasible. Its the opposite of Metcalf's law, where the value of the network increases with every node. With a WiFi Mesh, the value of the network decreases with each node.
In System Economics terms, the system exhibits diseconomies of span. With each new sequence of production, the system produces diminishing returns. More on architectual economies in a forthcoming post.
This morning, Rafe Needleman's Catch of the Day was "A Network of One's Peers." He covered two companies, Spanworks (there's that word again) and Greenpacket, two companies developing a more reasonable approach to ad hoc mesh wireless networking. By focusing on self-organizing between devices (Laptops, PDAs, Printers, etc.), they avoid scaling issues and realize economies of span.
The value of mesh networks is in their redundancy and inherent optionality. Vendors focusing on optimization logic that excercises these options have real promise. At Negroponte's scale, a vendor could create real value by combining a local WiFi Mesh with logic for multi-homing of IP Transit for optimized nieghborhood/multi-tentant unit Access. And considering the unnatural monopoly ILEC have over Access, any development in this area is beneficial.
11:46:39 AM 
|
