danah boyd has a very smart post (as she so often does), looking at how male and female bloggers use links and blogrolls on their blogs. Insipired by a conversation at the Blogher conference where someone asserted “Women don’t network”, danah took a look at 500 random blogs tracked by Technorati, as well as a subset of the top 100 blogs on Technorati, coded them by gender (based on details the authors revealed about themselves) and looked for patterns in linking by gender.
One conclusion: political blogs tend to link a lot, while personal blogs link very little. And men appear more disposed to maintain political blogs than women, while women are more disposed to maintain personal blogs. While this might lead to a conclusion that men link and women don’t, danah points out that users of LiveJournal (a huge segment of the blogging population and a community that’s 75/25 female to male) maintain large “friends pages”, which are a special (and more effective) sort of blogroll, as they act as both blogroll and aggregator.
danah goes on to observe that a link, in a blogging context, is an amazingly oversimplified representation of a social relationship. In a given day, I might link to the post of someone I’ve never met, but who said something interesting; to something by a business colleage, something by my wife, something by someone I actively dislike but who said something interesting… a simple analysis of links declares that I’m “linked” to all these people, but can’t distinguish between strong and weak ties. (The gang at Technorati and others have been trying to get people to code relationships into their link tags, but the idea doesn’t seem to have caught on that well…) danah observes that, generally speaking, men are good at making lots of weak ties (like the stereotypical blogger?) and women are better at making a smaller number of strong ties.
Reading through danah’s observations, I’m struck by a problem I face every time I try to do quantitative analysis of blogger behavior – we’ve got one word – “blog” – which points to a vast range of technologies and behaviors. It’s very hard to generalize about “blogs” when that term includes everything from personal diaries with an audience of none, or intimate friends, to ad-supported groupwritten magazines with audiences of hundreds of thousands. When I beat bloggers up for infrequently pointing to international news stories, I don’t mean to slam people maintaining personal journals – I don’t talk very much about Africa on my livejournal either… but it’s very hard to find data sets segmented into political blogs, technology blogs, personal blogs, bridgeblogs, etc. If anyone is doing good work in automatically partitioning random blogs by function or audience, I’d love to know about it, as it would help some of the projects I’m currently coding a great deal.