The Philosophy of Kosmix

Kevin Maney writes: Over lunch in San Francisco, I talked with Anand Rajaraman, co-founder of the new search site Kosmix. Rajaraman isn't just some wide-eyed newcomer thinking he can ride into town and take on Google: In 1996, he founded Junglee, which was bought by Amazon.com, which in turn made Rajaraman its director of technology. He's also a founding partner of Cambrian Ventures and a sometimes-professor at Stanford.

To some degree, I've had this same conversation a half-dozen times with other search company CEOs. Google is about searching, not finding, they say. Google dumps results on your laptop with all the care of a truck dumping gravel on your driveway. People want a search site that better understands what they're looking for, and then organizes results in a helpful way. Ask.com told me this. Cha-Cha told me this. Kosmix is joining the parade.

Kosmix might be the breakthrough, but I don't know enough to tell you that. The company just got another $20 million in funding from investors led by Time Warner, and any company that can get $20 million out of anybody in this atmosphere deserves to be taken seriously. The technology behind Kosmix seems to take the idea of search to a whole new level. The site often seems to understand not just what you're looking for, but where best to look for it. Rajaraman shows me, for instance, that if you type in "cake," Kosmix knows to search Martha Stewart videos. (Although, I notice that if you happen to be looking for the band Cake, you're out of luck.) In the end, Kosmix gives you kind of a Web research report on any given topic. At its best, the Kosmix results look like something an assistant might pull together if you asked for an overview about, say, Stonehenge, or The Rubinoos.

Clearly, there's a tempting opening in search that's luring people like Rajaraman. In the model called the Fidelity Swap that I've been researching for a book, Google is the ultimate high-convenience search offering. It is ubiquitous, easy, fast and bare-bones. And since users aren't invested in Google search in any way (compared to the ways they're invested in Facebook or Twitter), it looks as though all any newcomer should have to do is create a better, more satisfying search experience -- and traffic would rush in.

So tempting -- and apparently so hard to do, since it hasn't yet happened. It will be interesting to watch Kosmix.


. □