Microsoft Research Report on IM usage
Microsoft Research just released a very interesting research report "Planetary-Scale Views on an Instant-Messaging Network" on the usage of Instant Messenger Usage. The anonymized study analyzed 30 billion conversations among 240 million people, from which a communication graph with 130 million nodes and 1.3 billion undirected edges has been constructed. The data for the analyzed dataset was gathered within 30 days in June 2006 and includes three main data sets:
- user demographic information
- time and user stamped events describing the presence of a particular user
- communication session logs
which have been used to conduct the analysis.
Next to usage and population statistics the study reports on the social network that is formed by the conversations. This is a quite unique approach and in contrast to several other studies, which in most cases used the buddy list to compute the social network in instant messenger networks. What is interesting in the study is that the degree distribution doesn't follow a power-law distribution although it is heavily tailed. A strong cutoff parameter and a low power-low exponent has been found, which indicates a high variance in the degree distribution. Another interesting finding is that the average shortest-path length is 6.6. This validates the famous Milgram "small-world" experiment about the "6 degrees of separation".
What has surprised me was that the average number of contacts in the buddy's list is 50. I know that I am a heavy user and my msn buddy list only includes 45 contacts. Maybe my perception is wrong and I am not a heavy IM user, but I still find that number very high.
Nevertheless does the study offer a lot more interesting data and findings including:
- Levels of activity
- Demographic characteristics of the users
- Communication characteristics
- Communication by age
- Communication by gender
- World geography and communication
- Communication among countries
- Communication and geographical distance
- Homophily of communication
- Network cores
- Strength of the ties
This study shows once again, that there is so many data out there that can be gathered and analyzed. We can find so many interesting facts just by analyzing the data that is already available to make sense of our current world and understand it. This is what Hans Rosling is promoting in his vivid TED talks. His gapminder.org joint effort with google shows how powerful the combination of an intuitive visualization interface and the analysis of available data can be.
Reference
