Social Network Analysis of Blogosphere, part 2

I sent out the link to New York magazine’s article on Social Network Diagram of the Top 50 Bloggers to the Socnet mailing list, which generated a lot of interesting discussion.

Elijah Write pointed to this older picture of the blogosphere:
http://www.blogninja.com/hicss05.blogconv.pdf

Valdis Krebs pointed to these links:

Stan Wasserman’s comments on this topic in his blog “Centrality”…
http://www.centralityjournal.com/archives/what_is_a_technorati_ranking.html

Eszter Hargittai’s analysis of political bloggers [she followed actual conversations]
http://crookedtimber.org/2005/05/25/cross-ideological-conversations-among-bloggers/

Lada Adamic’s study on political blogs
http://www.hpl.hp.com/research/idl/papers/politicalblogs/

Jan Schmidt wrote:

“I agree that Blogroll links are not as well an indicator of actual blogging practices than links in postings and comments. Just to give an indication of blogrolling practices, here are a couple of findings from a large-scale (N=5.247) survey of the german-speaking blogosphere we’ve conducted in October 2005 (alas, not published in english yet; some more
info: http://www.bamberg-gewinnt.de/wordpress/archives/348).

34% state they modify their blogroll once a month or more often. Frequency of blogroll update correlates with age of weblog, with younger weblogs updating the blogroll more often. Both findings indicate that a blogroll gets build primarily in the beginning of one’s blogging activitates, while authors build their networks within the blogosphere, but is a less reliable indicator for ties among “older” bloggers.

Valdis Krebs asked if the data in the article I linked to ( http://newyorkmetro.com/news/media/15972/ ) included blogrolls. The data is from Technorati which does analyze blogrolls, and all other data on a blog’s home page. As an example of this, see Technorati’s vulnerability to manipulation by blogroll spam (see http://datamining.typepad.com/data_mining/2005/08/technorati_hit_.html )

A Socnet member mailed me off-list to observe that he thought that the most-linked-to blogs are not the same as the most-read blogs. I completely agree. For example, the article in New York magazine points out that Fleshbot (a blog about people not wearing clothing), has relatively few incoming links but 300,000 page views per day.

Also, blogs that attract bloggers as readers will attract more incoming links. A blog like Gawker or Brain Food ( www.Teten.com/blog ), which attracts a less technologically sophisticated readership than Engadget or BoingBoing, will generate relatively fewer incoming links.

ADDED:

Bill Ives wrote:

Technorati looks at all the links that are on the “front page” of a blog. This includes both the posts that are still visible and the blog roll. This means that they are in a rolling time window and links are no longer counted when the post falls of the front page of a blog. If you type link:url (url being the url of the blog or web site) into Google you will get all the currently active links, generally a much bigger number. For example, for my blog the Google search reveals ten times as many links as Technorati.

Google makes their APIs available for free for researchers (or anyone) and I have used them with a colleague, Peter Gloor, to look at social networking within blogs using his software. I have also heard that Google has better search capabilities for links than Technorati but I cannot confirm this.