Blog metrics

Posted by | September 30, 2003 | technology | No Comments

Doc points out that there are five bloggers born every minute. I guess if bloggers outnumber suckers five to one, they can’t all be bad.

Keeping up with tracking blogs growing at this rate sounds like a daunting task, however, I suspect it is easier than it at first seems.

The popularity as measured by the number of inbound links tends to follow a power law distribution.

Perhaps the measure of how prolific a weblog is also follows a power law. Our analysis at Moreover tended to show that the majority of weblogs are not updated even once a week.

The two standard metrics for information retrieval are relevance and retrieval, i.e. what percentage of all the good stuff you get back. In tracking weblogs popularity and prolificacy are the equivalent of relevance and retrieval.

I’d love to see Dave Sifry’s indispensable Technorati include a blog popularity and freshness rank.

Popularity rank would be pretty much what is there already, but instead of the rank being based upon the raw number of links, the links would be weighted in turn based upon the number of inbound links from each site providing a link. As Dave pointed out this is just page rank applied to blogs.

The freshness rank could be based on the last month and grouped into specific timescales updated on average at least once every: month, week, 5 days, 3, days, 2 days, day, multiple times per day. In this way a weblog aggregator for example could easily group monthly weblogs into a single feed.