Tuesday, March 16, 2004

previous entry | main | next entry | TrackBack (1)

Thanks, but no thanks

Via Glenn Reynolds, I see another ranking of blogger influence. This one claims to rank order "[t]he most influential reporters and bloggers on the web."

The good news -- I come in at #15. Wow -- this and the Library of Congress in less than 24 hours!

The bad news -- According to this ranking system, David Brooks comes in at #20, Tom Friedman comes in at #40, David Broder at #57, and George Will at #172. Fareed Zakaria is not among the top 200.

In other words, I'm fairly certain that the methodology used to compile this list is horses--t. [What if you're wrong?--ed. Then I'll magnanimously offer to trade places with Tom, Fareed, George, or either David -- because I'm that kind of guy.]

UPDATE: After informing my lovely wife Erika of this ranking page, she queried, "I didn't know your Mom had a web site."

ANOTHER UPDATE: Kudos to Philippe Lourier for responding to semi-constructive criticism and taking the responses in stride.

posted by Dan on 03.16.04 at 01:14 AM


If George F. Will starts sprinkling in some Salma Hayek cheesecake will he improve his ranking?

And Friedman has too much CIA spies and not enough Spy Kids 3D.

Fun and smart beats smug and smart every day. Top notch DWD, top notch!

posted by: joejoejoe on 03.16.04 at 01:14 AM [permalink]

Sadly, counting unique references to posts, blogs, and articles is not a good way to rank people. It's as if the importance of an Internet site was based on how many links it received, without any rating of how important each of those links is.

The right way to do this is to take the link graph and apply PageRank to it. In PageRank (which treats the link graph as a Markov chain), more important sites have links that count for more. Thus, if I linked to a post by, say, Dan Drezner, some of that influence would percolate through Dan Drezner to the person he linked to.

Technically, this would involve ranking posts, rather than authors. But it's a simple matter to get author scores from post scores: just add up all the posts by an author.

I'd do this myself if I weren't so busy. :)

posted by: Bob McGrew on 03.16.04 at 01:14 AM [permalink]


There are objectively verfiable degrees of potency of animal excrement. Trust me. I handled a lawsuit involving the corrosive effects of various types of animal excrement on metal hog pens. A summary judgment motion in this case involved declarations by expert witnesses with impressive credentials on this particular subject.

They agreed that chicken**** had the least corrosive effect (on metal sheds) of the various types of commonly edible livestock. The most potent was unquestionably bull****.

As I recall, the case name was Agresti v. American Hog something or other, before the Superior Court of California in and for the County of Stanislaus.

posted by: Tom Holsinger on 03.16.04 at 01:14 AM [permalink]

what abut make a more global list also blogger from europe austila and singapore and canda? the web isnt just a usa thing even it looks like amrican are forgetiong it

posted by: erte on 03.16.04 at 01:14 AM [permalink]

Since when are Zakaria and Friedman bloggers?

posted by: Dragutin on 03.16.04 at 01:14 AM [permalink]

any ranking where "the Jeff Corwin Experience" comes in as the 38th most influential reporter or blog has to be excrement, although I leave what type of excrement to the experts...

posted by: mike d on 03.16.04 at 01:14 AM [permalink]

How can this be right? Atrios comes in way down at 15?

posted by: paul goyette on 03.16.04 at 01:14 AM [permalink]

Since when are Zakaria and Friedman bloggers?

Ever since Bill Safire taught them how to use HTML. (c:

posted by: uh_clem on 03.16.04 at 01:14 AM [permalink]

I didn't see Kaus anywhere, either. --sw

posted by: scott on 03.16.04 at 01:14 AM [permalink]

No one has ever answered this question to my satisfaction: “What if the electricity is turned off?”

posted by: Michael on 03.16.04 at 01:14 AM [permalink]

Dan, I don't see any contradictions. Since Friedman et al. use major newspaper outlets primarily, their blogs may be way underutilized. It doesn't mean that the methodology is wrong. Learn to take a compliment.

posted by: ch2 on 03.16.04 at 01:14 AM [permalink]

Thanks for the link and the comments. I've posted another list using a different methodology - Bob;s pagerank idea is intriguing but I wonder if there;s isn't a cyclical problem with it unless you can rank a site's authority using some other metric than the number of links. Counting links is admittedly crude but what is an authoritative way to measure authority? Traffic? Btw, we don't count multiple links to post from a single blog. Any suggestions welcome!

posted by: Philippe Lourier on 03.16.04 at 01:14 AM [permalink]

Philippe - the idea isn't cyclical. As I said, it's the same idea used in Google's PageRank algorithm.

The idea is that you start off with everyone being equal, then adjust importance scores to make those who are linked more have higher importance scores. Then you adjust the scores to make those links more important, and then repeat. The whole thing eventually converges to a fixed point(which is independent of your initial importance scores.)

If you've got a technical background, the PageRank is basically just the first eigenvector of the link matrix.

posted by: Bob McGrew on 03.16.04 at 01:14 AM [permalink]

Bob - Hmm I had a quick look at Brin & Page - I wonder if the fact that online media (NY Times ...) doesn't point back to blogs skew the results -? I'm looking into this - I'll continue this conversation w/ you by email - Thanks

posted by: Philippe Lourier on 03.16.04 at 01:14 AM [permalink]

Philippe - didn't realize you were the BlogRunner author. Obviously you have a technical background. There's also a few parameters you can tweak to get the rankings correct - I'm writing a paper on a variant of PageRank right now.

I'll be happy to help you out with what I know of that methodology.

posted by: Bob McGrew on 03.16.04 at 01:14 AM [permalink]

None of these blog ranking programs can be considered the definitive ranking tool. But it's difficult to rank blogs; as some have pointed out, do you go by links? Give more points to an Instalink than a Drezner link, and fewer still to a Yourish link?

I know that Blogstreet attempts to do that with its BIQ.

No matter which way you look at it, you're going to get a giant clash of the egos. And I speak as a blogger with an ego big enough to be content with my position on Blogrunner.


posted by: Meryl Yourish on 03.16.04 at 01:14 AM [permalink]

Post a Comment:


Email Address:



Remember your info?