Wednesday, November 17, 2004

Great quote from a Google guy re the new MSN Search

Today at Internet Librarian, Greg Notess hosted a panel disucssion with representatives from Yahoo, Google, and AskJeeves so they could brag about improvements and answer a few questions. At the end Greg put up a slide with the new MSN Search, bringing laughter, and asked each rep to comment. The Google person, whose name I did not catch, was introduced as a member of the Crawl team at Google. His answer was short and sweet:

“Microsoft has a history of getting a lot better when they enter competitive markets. I expect to see lots of improvements in MSN Search.”

There were a few titters in the audience among those of us who recognized a deliciously backhanded compliment.

I buttonholed the Google person to ask him why Google over-emphasizes the importance of an entry when it's on the main page of my Blogger blog, then loses track of it altogether when it falls into the archives. I think I know the answer to this: when an item is on the main page, PageRank sees the various links to my blog, some from popular sites. But when the article falls into the archive, it's no longer deep linked from anyone prominent, so PageRank disses the deep page.

Here's an example: Google for:

radioactive cat

... you won't find my article on this from the archives in the top 500 hits:

http://wigblog.blogspot.com/2003/11/our-cat-is-radioactive-self-styled.html

... but if you had done the same search when my radioactive cat posting was on my blog's home page, Google served my article as the number 1 hit. It went from the most important article on the planet related to radioactive cats, to nearly non-existent. Why?

Yes, you could argue that older items in a blog deserve a lower ranking -- after all they are older. Still, I claim this is a PageRank failure: it over-emphasized the article's importance just because it was on my blog's home page, and now it under-emphasizes it. Since Google owns Blogger, they could think of ways to fix this.

This is an example of a more fundamental flaw in PageRank: it measures the popularity of each Web page -- each URL -- and doesn't take into account that the page is part of a site whose popularity should be measured as a whole.

I gave him a business card and he said he'd get back to me. We'll see.

No comments: