1On the Adoption Paths and Competition of Network Applications

Much of the growth of the Internet has been driven by new network technologies and applications. The successful adoption of theses applications greatly depend on the underlying diffusion process as governed by the system parameters and network externality effects. Developing guidelines for judicious deployment of such network applications require a study of their adoption path and behaviors. A diffusion model based on individual decision making allows us to study the mechanics of the adoption process. Prior work have shown that the commonly observed S-shaped curves for equilibrium adoption path can be observed in products where users are differentiated over the stand-alone benefits but are homogeneous in their valuation of network benefits. However, many new network applications (eg, Skype, Flickr, etc.) cannot be accurately modeled with such user utility models. In such applications, users are differentiated in their valuation of network benefits and derive less significant stand-alone benefits. In this work we identify three alternative utility functions that are more suitable for network applications, and show that the S-shaped diffusion curves can be observed as an approximation to discontinuous equilibrium adoption path even in these cases. Additionally, for each of these different utility functions, we use simulations to explore the impact the competition between two similar network applications when converters are present. We identify a range of interesting behaviors like the potentially harmful effects of converters on the individual technology and the overall market penetration levels, and the creation of instability in adoption dynamics, etc. The presence of such unexpected behaviors in the competition process therefore underlines the need for more careful policy decisions on the part of the network application providers. 1