Welcome to Microsoft Pri0: That's Microspeak for top priority, and that's the news and observations you'll find here from Seattle Times reporter Sharon Chan.
July 28, 2008 12:48 PM
Posted by Benjamin J. Romano
A regular refrain from Microsoft these days is that it's "early days" in Internet search. The status quo model -- 10 blue links on the left, paid-for ads on the right and the top -- "is ripe for innovation," CEO Steve Ballmer said at Microsoft's Financial Analyst Meeting last week. Today, we see the launch of a new search engine that promises to disrupt the model. But it's not from Microsoft. Or Google.
The buzz today is around Cuil, pronounced "cool." Cuil, whose founders include former Googlers, is boasting:
-- An index of 120 billion Web pages, triple any other search engine;
-- Results organized magazine-style, separated by subject;
-- Page relevancy determined by content rather than popularity;
--Complete privacy protection. "Cuil does not keep any personally identifiable information on users or their search histories," the company says.
I'm here at Microsoft's Faculty Summit, where computer science researchers from academia are trading ideas with the big brains of Microsoft Research. And you don't need to look far to understand where Ballmer's coming from when he says he's confident there's lots of room to innovate in search. (Incidentally, as he told the analysts last week, "if it's not ripe for innovation, we shouldn't be doing what we're doing. We will not be able to be very successful by only doing what the market leader does.")
During a wide-ranging session on artificial intelligence research this morning, Eric Horovitz of Microsoft Research mentioned ongoing work in "personalized Web search." While the effort is not brand new, it is certainly relevant.
The "personalized" model is designed to improve relevancy of search results while greatly improving user privacy. Instead of one's search query history being stored on the search provider's servers -- where that valuable information is "owned and explored by company X" -- the history would live on the searcher's own computer.
The user makes a normal Internet search query and his local PC would use the stored history to personalize the results based on recent queries, documents stored on the hard drive and other information. In this way, no information beyond the initial query is shared with the search provider. But the results page would contain a box of "personalized" -- and presumably more relevant -- results.
"This is a hot topic area of focus for us right now," Horovitz said.