Statistics about blogs are all the rage at the moment.
Steve Rubel says it is time that we had standards so trends can be gauged and advertisers can be appropriately gouged. He is right, but his evidence for the existence of a problem is flawed.
He complains that a study in November 2004 put readership at 58% while a
different more recent study put it at 20%.
Actually, the BBC 2004 report says readership is up by 58%, taking it to 27%
of online Americans. The article says 120 million Americans were online in
2004, which is about 40% of the population. That means that study was
saying 11% of the population had read a blog.
Even without measurement standards, the increase from 11% “have read a blog”
to 20% read blogs “frequently” or at least “occasionally” seems fairly
consistent and believable over the space of a year or so.
One of the challenges for any sort of automated measuring is going to be spam detection. Here is an example of a real blog that has been left unattended for a while. Detecting pure link-farms seems to be a hard enough problem in search. Removing spam noise from a site that contains a mixture of real and spam comments has to be harder.