Microsoft Search Lab

Microsoft is starting a new major lab in the bay area (in Muntain View) called Microsoft Search Lab. It appears that they are aggressively recruiting researchers for the lab. I know a well known database researcher who is moving from IBM to that lab after spending most of his illustrious career at IBM. I also know that they are courting another major database researcher.

It is no secret that Microsoft wants to do what it has done well in the past — win a catch-up game. This time they have to beat Google and Yahoo. ime will tell whether they can do that third time.

My guts tell me that there is likely a new company somewhere that will be a winner in the next round of search. Incremental approaches rarely bring a new champion — Microsoft, Yahoo, and Google are all examples of how a new company with revolutionary ideas and enormous hunger wins the game. Micrsoft is recruiting established and illustrious researchers from different places. But can they bring revolutionary ideas. My experience — and this is not a reflection on the hires for the Micrsoft Search Lab but a general experience — is that established researcher usually do not think in revolutionary ways. I don’t really think that information retrieval or database people are going to revolutionize search and produce a winner in the next round. It will be somebody who will think about search in a different way that current search people don’t. And as everybody knows, search is in its infancy so there are many interesting things likely to happen in this space.

Independent of what happens, things are really getting interesting in this space. Ask has a completely different look under Barry Diller’s guidance so they are also training for the ‘race’. Some new companies also announced their intentions to participate — at Demo2006. And I am sure that there are quite a few. I know that Marc Cuban was investor in two search companies at one time. I am waiting to see what 2006 and 2007 bring in this space.

Leave a Reply

Your email address will not be published.