Eric Schmidt on blogs while speaking in England: "The average blog has one reader: the blogger". Damn, I'm caught out. Update: Interesting response to Eric's talk here from Micah Sifry – Eric was giving a speech to the annual meeting of the conservative party in Britain, Micah covers all…

Eric Schmidt on blogs while speaking in England: “The average blog has one reader: the blogger”. Damn, I’m caught out.

Update: Interesting response to Eric’s talk here from Micah Sifry – Eric was giving a speech to the annual meeting of the conservative party in Britain, Micah covers all things digital democracy…

## Author: John Battelle

A founder of NewCo (current CEO), sovrn (Chair), Federated Media, Web 2 Summit, The Industry Standard, Wired. Author, investor, board member (Acxiom, Sovrn, NewCo), bike rider, yoga practitioner.
View all posts by John Battelle

Funny.

5,491 subscribers in bloglines is not half bad.

I know it’s partly a joke, but I wonder what’s the best way to find the “average” blog? How do you find the middle of a power law curve? I’d imagine the tail of the blogosphere is longer than most, skewing the typical average quite a bit.

Odd perspective from a math guy like Dr.Eric. A single reader would be the *lowest number of readers possible* and the average is going to be considerably higher, esp as Google increasingly (and correctly) indexes blogs above other types of sites.

But this demonstrates Google’s achilles heel. Failing to see that *community* is in the process of overtaking *raw data* as the key online focus, and interest, of an “average” web user.

Adam: I had the same question. Did some looking around. Apparently the mean of a Zipf distribution is H_{N,s-1}/H_{N,s}, where N is the number of blogs on the web, s is a real value, greater than zero, which characterizes the shape of the curve (how quickly the head of the distribution drops off into the long tail), and H_{N,s} is a generalized harmonic number, i.e. the sum for i = 1 to N of 1/(i^s).

So if someone from Technorati or Google some other blog search engine could tell us how many blogs there are (N), and what parameter (s) was the empirical best fit for the slope of the curve, we could figure out how many readers the average blog would have.

Actually, I wonder if someone has published statistics on the web, if we could use these web statistics as an approximation of blog statistics. What is the average number of views a page on the web gets? Given the amount of machine-generated web pages out there, I am almost certain that there is a long tail of web pages that have zero views.. the creator of the page probably hasn’t even looked at it.