[Tech] Applying Social Network Analysis to weblogs
Scale-free networks: people do not link randomly Link density and network-buiding as a "natural" process: some people simply have fewer links - not a problem. Power Law Distributions & Hubs: "A-List Bloggers"/ "Dead" Weblogs/Those with cat-pictures
Governing Principles of networks: • rich get richer • Achilles heel: if a hub is removed, the community might collapse • Diffusion: Hubs play a crucial role in influencing the community (Hubs, like Schockwellenreiter or Dave Winer) • Growing difference: The more nodes, the bigger the difference between "rich" and "poor" nodes
Methodology - how are the networks/[personal] connections measured: - Incoming / Outgoing Links - Connections between persons determinded by Comments 40%, Blogroll 40%, Stories 20%
Incoming links • Average: 6 Links • Median: 2 Links • Power Law • 20% have no links • 60% have less than 3 links
On twoday.net, users have a different linking philosophy - many weblogs do not consider links to be important.
Community of approx. 600 blogs is held together by just 10-12 blogs - very close "clique" with personal contacts at the heart of the community - hard to become part of this core community. This holds the network together, but it's also a cause of churn if users try too hard to get "accepted" by these "A-list" bloggers, but don't.
Power law seems impervious to "typical" portal strategies for diffusion. How much of this is related to personality, and how much to content? What makes an A-List blogger?
Visualisation using touch graph
Community pressure by A-List blogger heavily influences churn -- the visuals look like those from physicists related to gravity around large bodies in space. How does this impact extraordinarily active, but tiny communities of bloggers? Would you see the same effects in communities of bloggers who are all friends and blog exclusively with one another?
Conclusions • every node can be reached through 6 connections
Q: roland: Password protected blogs - has any analysis been done on these? (e.g. his sister got interested when Roland posted baby pics).
A: 20-6 no data 19% have password protected blogs A: ??? no data - only 10% of community is password protected
Q: blogs & media coverage? A: ??? -> had good media coverage in Austria when they set up the service. This likely leads to new bloggers and resulting churn. A: 20-6: article in times, gained more than 1000 bloggers, only 10 survive (UK)
Q: MB: in 1996 - was worried about the collapse of the web to 1 or 2 sites; similar concern with A List bloggers. This never happened. why?
A:20-6 : it did happen in online auctions and google. (power laws and 80:2o rule are out there) ◦ Will this happen not with A-List bloggers, but by those who offer blogging
Suw: welsh (30 blogs, half dead) -- how do you encourage a reluctant community to start using blogs (could be a perfect medium for minority languages/ minority groups) -- so how do we encourage them to (a) start them (b) continue using them?
A: MS: no real recipe for success, best thing is if those who are enthusiastic about it tell others about it (use "contagion") - but it boils down to whether people feel they have something to say or not.
NL: word of mouth among bloggers and their friends ◦ What about usability? Or the ability to "blog" from a spontaneous impulse? Do minority groups have different needs that traditional blogging doesn't satisfy?
LE: most important: not functionalities, but what it enables for the community
MS: Blog survey: several types of bloggers
neighbourhood is important - few, if any outward links strength in seclusion?
NL: communities should not be based on using the same tools
SG: This seems to be a German phenomenon, in the UK this is not the case
- there are cross-tool coimmunities.
MS: tries to be part of the community where he currently is - maybe that's representative - isn't wanting to know the whole blogosphere like wanting to know every person in the world?