Saturday, September 01, 2007

Google's highly dynamic indexing

For a while, my friend Prabal Dutta has been in the #1 spot on a Google search for his first name. Recently, I checked to see if he was holding his spot. Nope! Some other dude had it, instead, and Prabal had dropped to #2. I sent him a teasing email, "Dude, you lost 'Prabal!"

He wrote me back from Japan, where he is attending a conference, "I seem to have won it back," and enclosed a screenshot showing him in the first position.

I immediately checked on my end. Nope! Still some other dude in the first position.
While his location (Japan) might be at play here, it's even more likely that with him logged in, Google is blending some personalization in to the results. That itself is quite interesting, but not the most interesting discovery to come from our exchange. After we swapped screen shots, I clicked on the link to his page for whatever behavioral boost it might give to him, and went off to have a fine dinner party.

Tonight, 10 hours later, I checked again. I'll be damned. Not only is (my friend) Prabal Dutta back to #1 in my search results, but Prabal Gadh has absolutely vanished. I am not logged in for any of these searches, though I am almost certainly cookied.

The big news here is that not only is Google crawling and integrating new content incredibly rapidly (as I mentioned in my recent Scoble post) but they are re-indexing and modifying ranking incredibly rapidly, too. It's a highly dynamic index -- vastly different from the once-every-three-months crawls and refreshes of 2002, and worlds away from the crawling and indexing policies of trailing quasi-competitors like Yahoo.

It's as if they had the world's best distributed computing platform, or something.

I'll have more to say about that platform soon.

No comments: