How to anticipate the Internet's global diffusion

Tracking the Internet's global diffusion is a daunting but increasingly important task, especially for network capacity planners. A 1997 report found that Internet consists of more than 16 million registered host computers. Conceived initially as a demonstration project for the U.S. government, the Internet today aggregates traffic from a vastly wider set of constituencies. And commercial use now accounts for 58% of Internet traffic, far exceeding the network's original purposes in research and education. But consumers of Internet-related services are regularly frustrated by slow response time, inaccessible online services and breakdowns leading to services being unavailable. These problems can follow inadequate capacity planning caused by ignorance of the Internet growth process. Understanding Internet growth patterns involves assessing alternative models for the Internet diffusion process. The internal influence model is one class of models used to study the diffusion phenomenon. These models are informed by the diffusion of innovation and various economic theories and assume that nonadopters of an innovation are increasingly likely to imitate adopters over time. Empirical studies have used these models to investigate the diffusion of such communication systems as residential telephones. Internet governing bodies, task forces, like the Internet Engineering Task Force and commercial services also use such models to study Internet growth. The article explored two commonly used internal influence models-Logistic and Gompertz-both of which assume variable growth rates whereby the growth rate first increases and then decreases over time to achieve a finite subscriber saturation level. It is unclear to researchers and planners whether internal influence models adequately convey the Internet diffusion process. There appears to be some conflict between assumptions underlying the models and the context of the Internet diffusion phenomenon