After this NetSci 2006 gig, 17 capstones in 11-1/2 hours this Friday is going to be a breeze. Unless there’s math.
I almost skipped the early morning session, or part of it, for fear of the same. Glad I didn’t, as Gary Robins gave a talk on modeling social network structures to account for both attitudes (personal beliefs) and norms (perception of what others believe). He showed, albeit through just one small study, significance in both norms and attitudes in the construction of social ties where it is not significant when ignoring the norms. This is important because it suggests that the structure of a network relies on independent factors working in combination. It was also very exciting when Gary based his theory on work by Connell that relates gender to hierarchies of power. I’m very interested to look that up and see how it matches up with RCT at the Stone Center.
With a few exceptions (see below), this proved to be the day with the least relevancy for me. It was plagued with some long-windedness and tech problems, and of course … math. Plus, I spent $40 for “social events” that apparently everyone can attend. The latter is great, but I feel like I’m part of a social experiment to test the tragedy of the commons, or something. I’m not going to the outing tomorrow night, since I feel a need to spell Amy in reading the boys to sleep.
The other highlights from Tuesday:
- Dynamics of viral marketing / Lada Adamic (Michigan) — Another great talk by Lada. This one was about work conducted while still at HP Labs about viral marketing campaigns for book and DVD sellers (she didn’t say the name of the online merchant, and asked us not to guess aloud who it might be … wink, wink). This is viral marketing only in that it was a low-cost effort to distribute information about products, but the results were not shaped like the spread of a virus. In fact, the findings suggested that it really didn’t work as expected. While the ROI may be just fine for this Company Who Shall Not Be Named, the data showed that multiple recommendations seemed to lessen the effectiveness of the referral message in making a sale, and price was the biggest incentive for personal reward. The most interesting thing for me was that small, tight-knit and well-organized communities are more susceptible to viral marketing campaigns. (Oh, by the way … The Nile is one of the two longest rivers. What’s the other one?)
- Evolution of transportation networks / David Levinson (Minnesota) — In trying to answer the question of whether roads are an emergent property, David used an agent-based model to predict the development of traffic infrastructure. It was only a 20-minute talk, so I think I need to look at more of the detail, but it was neat to see this kind of modeling done here. There are more than a half dozen factors that went into this kind of system (economic growth, land use, travel behavior, maintenance/expansion, network revenue, investment, and induced supply and demand). In the end, the conclusion was that a hierarchy does form that places faster roads at the center of the system with medium-speed roads helping to form a loop and slow roads reaching into the local neighborhoods.
- Measuring node importance, impact on consumers / Paul Dwyer (Texas A&M) — Another hit for my research interests, this study looked at a virtual community as two separate but related social networks: authors and information. The actual purpose was business-driven, examining the impact Word of Mouth (WoM) has in purchasing, but I was excited to see someone talk about relationships between social capital and knowledge capital (the people and what they write). People who start discussions come from all over the demographic field, but people who answer queries tend to have higher social capital. In fact, knowledge is the most valued property.
- Fruitfly data tools / Memo Dalkilic (Indiana) — Although I skipped all of the other bio talks Tuesday afternoon, I stepped back into the room to see Memo talk. Always fun. It was a very informatical presentation about massive, confusing amounts of gene data and needing to present it in an easy-to-understand, easy-to-access manner.
Huge all-star lineup tomorrow. Albert-László Barabási … Jeffrey Johnson … and a second chance to understand Carter Butts (who seems completely brilliant and asks lots of questions, mostly about math). There are a bunch of IU talks scattered throughout the day, too.
