battle of the networks

May 15, 2006

5. Battle of the Networks

Companies have sought to exploit network effects since W. Brian Arthur dubbed them the competitive linchpin for information-age business. Many have used technology to tie together critical masses of customers and the most or best suppliers and so have gained an edge. But now enough companies derive competitive advantage from their networks that they are coming up against one another. That means we must learn a whole new set of principles: not how companies compete against networks but rather how networks compete against networks.

Companies that introduce new networked products or define the standards by which networked products interact can quickly dominate a market. That strategy is known as “lockout,” and it’s tough to beat. Companies seek lockout not just through product design but also through an advantageous arrangement of buyers or sellers, through ingenious feedback or feed-forward loops within supply chains, or through the exploitation of technology-enhanced social interactions within markets (think eBay and Friendster).

We now have techniques for evaluating some characteristics of networks, such as the distance between nodes, diffusion dynamics, and connectivity patterns. But we know almost nothing about how networks compete against each other. And since most of us think in linear, nonnetworked terms, our intuition provides little help.

One approach to studying this new dynamic is to redesign the boards of games like Battleship, checkers, and Go into complex networks and observe how players compete. These games are traditionally played on grids, which are very regular networks (nodes and links are evenly distributed across the board). The redesigned boards are modeled on the Internet and other real-world competitive networks whose link and node distributions are irregular because of “rich-get-richer” connection schemes identical to lockout in business. The boards comprise a small number of very well-connected nodes, a medium number of moderately connected nodes, and a large number of sparsely connected nodes. This connection pattern is a primary source of adaptation—and complexity—in networks.

Consider Go, an ancient Chinese game in which players capture stones and occupy territory. We found that when a Go board was redesigned for greater complexity, competitors could not visualize even basic patterns of play without such mathematical tools as a “connectivity matrix” (a map of who links to whom) at hand. (See the exhibit “The New Rules of the Game.”)

 The New Rules of the Game
Once armed with such tools, however, players began to invent entirely new strategies, even though the basic rules of play remained unchanged. For example, one classic Go strategy is to occupy territory (nodes and links) with large contiguous masses of stones. Occupying the nodes with the most links achieves this goal quickly, so the smart thing to do is seize those nodes first. Players using the redesigned board soon found that with this strategy, a first-mover advantage heavily influenced the outcome, and the winner was determined within several moves. After repeated play, however, participants discovered several ways to counter that advantage. For example, they distributed small clusters of stones around the board so they could keep their options open until much later and prevent competitors from guessing the specifics of their strategies. Players could win by rapidly amassing their stone clusters into a large group at the appropriate time, in effect unlocking the lockout achieved by the first mover.

Such research has practical business applications. Consider, for example, the supply chains of competing companies. Suppose Company A operates an innovative vendor network that rearranges inputs to its production process according to the latest market data. Company B might build a similar network and compete with A on the basis of network efficiency, lower cost of inputs, or better market data. But it might be locked out because A has already climbed the learning curve (and perhaps invested its enhanced profits to further improve its innovative process). B might be able to overcome A’s growing advantage with heroic efforts in traditional competitive competencies (for example, by recapitalizing plant production, tightening profit margins, or slashing transportation costs). Or it could go network a network, for example, by examining the structure of its emerging vendor network for undetected strengths, such as connections within vendor clusters that are even more advantageous than those in A’s value chain.

Of course, once B has unlocked A’s lockout, it will have to continually reexamine and, when warranted, reconfigure its network to fend off attacks by others. Close attention to competitive dynamics is the key to long-term survival in networked competition.

Jeff Cares ( is a military futurist who consults to the Office of the Secretary of Defense and the international defense community. He is the author of Distributed Networked Operations: Foundations of Network Centric Warfare (Alidade Press, 2005).

%d bloggers like this: