**Eigenfactor** – not just how many, but who they are.
**Betweenness** – I have a friend who seems to know somebody every where you go – he sits between lots of social groups. Without these people, the network collapses.
**Homophily** – people tend to hang around with people who are most like them. Matt feels most at home with people like the audience here – and we probably share many interest. That creates a huge feedback loop.
You can start to build algorithms based on who people know – for example, look at a network of politicians, and based on their relationships, work out which party they are.
**Susceptibility** – could we start to make guesses about what people are susceptible to, based on their networks? What can we give them the last push towards?
**Why Viral Isn’t** – viral marketing isn’t viral. If you haven’t thought about it, the exponential/viral model seems reasonable. However, “viral” spreading of media is more like hiccough attacks – lots of little peaks. Social networks are clumpy. Distinct groups of friends, often tenuously linked. So – if you send an e-mail to everyone at work, the pressure to pass it on beyond that is low, because people outside are different network groups with different interests. On average, people spread it to less than one other person. So, it tails off quickly.
BUT retweeting is a social act. It’s a nice thing to do to retweet something. There are complex social emotions behind it. And those are almost unmeasurable.