Pump It Up! Social Pumps and Information Diffusion in Social Networks

• Activity Number: 485
• Type: Invited
• Date/Time: Wednesday, August 4, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics and Marketing
• Author(s): Andrew Stephen*+ and Jacob Goldenberg and Yaniv Dover
• Companies: INSEAD and Hebrew University of Jerusalem and Hebrew University of Jerusalem
• Keywords: social networks ; online networks ; complexity

Abstract

We examine whether characteristics of the people who insert information into social networks affect diffusion, specifically considering both a person's connectivity (e.g., number of friends) and activity (e.g., frequency of "talking"). Past literature has only examined connectivity (e.g., work on "hubs"). We use formal mathematical models, simulations, an experiment, and statistical analysis of data from Twitter. We find that the extent to which a piece of information (e.g., a URL) introduced by a person into a social network (e.g., Twitter) spreads is positively affected by that person's activity rate (e.g., average rate of posting messages) but not by their connectivity (e.g., number of friends). While information introduced by a "hub" has a large audience, they do not necessarily retransmit it. Information introduced by hyper-active "pumps" does seem to get passed on.