Business Week: Microsoft’s Search Moves Analyzed

July 1, 2008

Catherine Holahan’s “Microsoft’s Plan B for Search” popped into my news reader this morning. The interesting essay–almost a business school-type write up–appears on the Business Week Web site here. I think the story will appear in some form in the hard copy magazine, but I read the online version this morning, July 1, 2008.

Ms. Holahan looks at the alleged Powerset buy out by Microsoft. The “Plan B” is acquiring additional search technologies in the aftermath of Redmond’s failed Yahoo deal. Her analysis is closely-reasoned, so it is difficult for me to summarize the argument.

I did find one point of particular interest; that is:

Rather than focus on creating one consumer-facing site capable of answering any query, like Google has, Microsoft has split its search engine into specific categories—a comparison-shopping engine, Microsoft Live Cashback; a travel search engine, Farecast; and a health-specific search engine, health.live.com. Today, semantic search engines do best with such category-specific searches, which help them to scan a smaller set of pages in detail. Scanning the entire Web in that much detail is difficult to do quickly.

Business Week has done a good job of explaining that Microsoft has a more fractionalized approach to search than Google. Keep in mind, however, that Google is not a single piece of digital cloth. There are different search mechanisms in operation at Google; specifically, the search system used for Google Base differs from the search system used for the search box on Google.com. The Google Search Appliance is also moving in its own direction as well.

In general, I applaud Ms. Holahan for identifying the difference initiatives within Microsoft. She has also identified two other interesting “semantic engines”. The first is Hakia, a company that offers a “Compare Hakia” function here and the Berggi Search for mobile devices other than the BlackBerry or iPhone, which limits the market somewhat. Hakia is working to generate the type of buzz that Powerset’s team found so effortless. She also mentions Expert System, based in Modena, Italy, and founded in 1989. The firm has beefed up its US presence with a new president and a more focused public relations campaign. You can learn more about Expert System here. Expert System has gained some traction for its software componetns in the mobile search market and has a lower profile in North America than Italy.

Observations

  1. The buzz about semantic search is gaining pitch and volume. My view is that semantic search is not an end in itself; it is a component of a search system. Vendors of semantic search are likely to find warmer welcomes as utilities or refinement functions within larger constellations of information retrieval methods. I guess I don’t buy the notion of “semantic search”.
  2. The key difference between Google and Microsoft boils down to the fact that Google has been working on its infrastructure for a decade. Without a honking big super computer, semantic technology is tough to implement on [a] large amounts of content and [b] content that changes frequently. The well known problem of updating indexes becomes quite challenging.
  3. Fragmented search is not necessarily a bad thing. But when there are many different search systems, costs become a problem quickly. Each system requires its own technology, engineers, and infrastructure. Google–while not homogeneous–avoided the “pushcart full of junk” approach taken by Yahoo. Microsoft, with its purchase of Fast Search & Transfer, may be unconsciously following the Yahoo model. Google’s approach of greater, not less, search homogeneity is the lower cost path. I was surprised Business Week’s B-school analysis of Microsoft’s Plan B ignored cost as a factor. Cost is a very big deal in search, which is the reason search vendors crash and burn. There’s no money to buy fuel.

Agree? Disagree? Use the comments section of this Web log to inform me of my intellectual short comings.

Stephen Arnold, July 1, 2008

Comments

Comments are closed.

  • Archives

  • Recent Posts

  • Meta