Megan Conklin, Assistant Professor, Elon University
Track: Emerging Topics
Date: Wednesday, July 28
Time: 10:45am - 11:30am
"Power laws" are increasingly used to describe an organizing feature of networks as diverse as infectious disease transmission, the World Wide Web, airport hubs-and-spokes, actors in movies, and corporate boards of directors. These "scale-free" networks behave according to certain principles: (1) a small number of nodes control a large percentage of links, and (2) evolution of the network is dependent upon complex "preferential growth" models.
Determining why the nodes in such a network "prefer" to link to one another is a key to understanding scale-free networks. A "rich get richer" hypothesis implies that nodes prefer to link to well-connected nodes. A "winner take all" hypothesis compares nodes in scale-free networks to particles in a Bose-Einstein condensate: some very "fit" nodes will grab all links, leaving nothing for other nodes.
Do open source development networks evolve according to "rich get richer" or "winner take all" models? Are new links (developers) attracted to the largest, oldest, or fittest existing nodes (project teams)? Is there a mutual selection process between developers and teams, and does this challenge existing notions of "preferential growth" in scale-free networks?
This talk presents the data collection methodology and analysis for a network of open source projects. Data describing the membership of over 15,000 Sourceforge development project teams was collected and analyzed for scale-free characteristics. This talk introduces power laws in open source development, similarities to other known power law networks, and the implications of power laws for predicting certain events in the evolution of open source development networks.
Download presentation file