Rory Cellan-Jones

Kosmix - searching for optimism

  • Rory Cellan-Jones
  • 9 Dec 08, 08:57 GMT

What's the definition of optimism in these troubled times for the hi-tech economy? How about starting a business which plans to take on Google from a base in Mountain View, the search giant's home town? Or maybe ploughing $20m into that business as an investor when everyone else is hiding under the bed covers?

But that's the story of Kosmix, a new search business that has announced today that it has won $20 million in funding from investors who include Time Warner.

The co-founder of this apparently mad venture dropped into London a week or so back, and on meeting him I was almost convinced that he might just have a chance of success.

In order to see this content you need to have both Javascript enabled and Flash installed. Visit BBC Webwise for full instructions. If you're reading via RSS, you'll need to visit the blog to access this content.

For one thing, Anand Rajaraman's idea does look quite compelling. He is very keen to stress that he isn't taking on the behemoth of search head-on. "Google is the best place to go if you're looking for a needle in a haystack - if you're looking for one single website or piece of information. But Kosmix is about exploring a whole topic." In the bar of the London hotel where he was staying he showed me an example. A search for Stonehenge grouped on one page pictures of the site, documents and discussions about its history, but also details of accommodation in Bath, should you want to stay there during your visit.

Later, I searched for "internet censorship" and found a useful collection of articles, videos and news relating to this topic. In this case, the results were not so different from what a Google search threw up, but they were organised in a more accessible manner.

The idea sounds a bit like the various "semantic search" companies that have popped up recently - from True Knowledge to Semantifind - but Anand Rajaraman said his firm had not gone down this path. "We deliver on the promise of semantic search without doing it that way." Instead, he explained, the Kosmix work involves "categorisation", crawling the web, extracting huge amounts of data, then using statistical analysis to figure out clever ways of extracting meaning from it. Such as working out that someone interested in Stonehenge might combine their visit with a trip to Bath.

The other impressive thing about Mr Rajamaran and his co-founder Venky Harinarayan is that they have a track record of turning their computer science expertise in to large piles of cash. They are friends who dropped out of Stanford in 1996, as part of the team behind Junglee - one of the web's firs comparison shopping software. That was sold to Amazon in 1998 for stock worth $280 million. Since then the two Indian-born entrepreneurs have used their share of that windfall to back other Silicon Valley start-ups, before setting to work to create Kosmix.

Still, even with a solid team and what appears to be impressive technology the odds must be against Kosmix making it big. Plenty of others have tried and failed to convince the world there is a better way to search. But there is no denting Anand Rajamaran's optimism: "It's a very good time to build a business," he told me. "There's not so much noise out there, and it's a great time to hire people." And when I suggested that Kosmix was entering a crowded market, he had a swift response: "It's getting less crowded every day."


  • Comment number 1.

    Hmmm...I've tried a few searches. Each search returned a nice selection of information...but not about the exact item I was searching for.

    It looks like you will have to be very specific in your search terms in order to find this site useful.

  • Comment number 2.

    I found the page quite cluttered and basically a repeat of Yahoo's homepage. The search engine just returns results from Google and most of the other result sections weren't relevant to my searches. If I want to search for videos I go to a video site like youtube, if I want images, I use google images. I can't see how this site does anything differently. Oh, and it's slower than google.

  • Comment number 3.

    It would appear to just be a hash of Wikipedia, Yahoo and YouTube rolled into one website. Like previous posters, if I want any of those, I'd go straight to source rather than use an middleman. Another website I'll not visit again...

  • Comment number 4.

    Not my cup of tea. Any search engine that doesn't fit in 1024x768 isn't focusing on its job. From the article it seems that it's main selling point is that if you search for one thing it also brings up links to information it thinks you might also like. In other words, it gives you links to stuff you never searched for. Since when was that a good thing?

    Frankly Googles model is going to be extremely hard to beat. The page is simple, uncluttered, fast to load and execute your search, and it returns decent results. I admit that it works best with a well engineered search query, so the only real improvement I can think of is some conversion between "granny speak" and a proper search query to improve Googles chances of finding what you actually want. The advantage of that kind of business is Google may actually pay for your know how.

    How they got $20 mil I'll never know. Must be based purely on the lads name rather than any strength in the business model.

    Give us proper media searching (e.g. "picture of Dallas skyline at night" bringing up appropriate images based on their content rather than keywords attached to them), give us searching in foreign language sites (e.g. I seach for "dalmations inherited diseases" and it brings up not only appropriate English language sites but translated French sites about the issue too), give us searching by website class (e.g. specify that your search results should only be returned from media sites, or from consumer advice sites etc) but gluing together a few queries and returning their results in a page littered with trash isn't what people need right now. Ok, it's not what I need right now.

  • Comment number 5.

    i tried a search and it was not very good

    as a veteran of the web industry (since 1994) i've seen and tried all the search services

    this one isn't worth using

    it's more of an aggregate of content, something that pageflakes or netvibes, or my yahoo, igoogle does way better

  • Comment number 6.

    These are dotcom bubble-gummers, but with a difference. Their past lives have included astounding successes as the (sale of) - a site with no success, but a concept worth a lot when in the right hands. So, yes, I wouldn't count on them to make the greatest search engine - i would expect them to make inroads into search that either microsoft or google would want to buyout their IP. Wait and watch....

  • Comment number 7.

    Sorry, but I really didn't like the way this search engine worked. For a start if I just stumbled across this

  • Comment number 8.

    Sorry, but I really didn't like this search engine. For a start without having read this article I wouldn't have even known that the page was a search engine by default, a small search box at the top with what is basically their homepage below makes it less than obvious what the function of the page actually is (I have never liked this about Yahoo either). For an as yet small and unknown search engine, this just doesn't work.

    To my eye the results are cluttered, unclear and far too many assumptions are made about what the user might or might not want to see. Also far too few results are shown, both in terms of numbers of results and the range of results shown (entire sections of the results page query only a single site).

    As it stands I can't see this making any headway what so ever into the search market.

  • Comment number 9.

    Hey guys, the valley startups are not merely into the familiar B2C (business to consumer), more often than not, they are into B2B (business to business).

    This company, like many others, has a business case in the B2B market - if they get their tech innovation right.

    Making a successful B2C business case - especially for a startup - is extremely challenging. Both startups and their investors should be crazy to only target a B2C success.

    So, startups take the B2B route (of lesser resistance) and selloff at the right price. They even go by a fancy name - "serial entrepreneur".

    Journalists, especially not the ones in the valley circuit, are not endowed with the eye for spotting this difference. Bear with them.


The BBC is not responsible for the content of external internet sites