WP Twitter Widget News

Follow me on Twitter

Archives

Categories
«  
  »

Future of search may not be about indexes and algorithms

At their conference yesterday Yahoo! talked about how they are moving from a web of pages to a web of objects – search engine land has a good report on the details here, but Head of Yahoo! Labs Prabhakar Raghavan sums it up thus:

We’re moving toward surfacing real-world objects rather than documents

and search engine land added:

Raghavan added that Yahoo is not going to be concerned about index size going forward. Rather Yahoo will be building these composite bundles of structured data. As a practical matter, these web objects are manifested in the form of multi-media content and images (Shortcuts). The broader objective is to provide more context and “answers” to minimize links and clicking back and forth.

In many ways there is nothing new in this editorialised view of search – Yahoo! has long believed that the best way to answer a query is for them to second guess the underlying intent and serve an answer – a tradition which includes the original Ask Jeeves, Mahalo, Wolfram Alpha and TrueKnowledge amongst it’s advocates.  On the other side of the debate of course there is Google and I guess now Twitter with their algorithmic approach.

Interestingly – whilst Yahoo! are talking about objects Google is talking about going more realtime, even indexing the whole web every second, as well as a host of other search initiatives.

Yesterday I wrote about how sometimes companies can lose touch with why customers love their products and I am wondering if there might be something similar going on with Google here.  Their mission of ‘organising the world’s information’ and algorithmic DNA may open them up to competition for a whole host of queries that are looking for simple answers to simple questions.  These users might not care if their search engine only works with 10% of the web, being happy to sacrifice exhaustiveness for speed and an easy UI in an 80-20 trade off.  (Saying that Google might be vulnerable is of course not to say that Yahoo! or Microsoft are any smarter – these companies have their own problems.)

Reflecting on the trajectory of the search market before I wrote this post it occurred to me that we might be witnessing a fragmentation.  We already have YouTube for video, Twitter for realtime, Kayak etc. for travel and dedicated map search sites – maybe we will now see the general query splinter between those who value Google’s approach and those looking for a quick answer.

May 20th, 2009 | Category: Google, Microsoft, Search, Yahoo!
  • I do see Google dropping the ball increasingly and being unable to break free of their previous mindset to explore new ways of doing things is a significant cause of this.

    I can even see a case where search as we know it fades away.
  • Daniel
    I agree with Simoncast. Though originally Google was brilliant, they continue building more on their original foundation rather than improving the foundation noticeably or as drastically as Yahoo is. I expect this will be their downfall when they fall (though for now I still am very much a Google fan).

    I'd just like to chime in about Wolfram, though, since there's so much hype around it and much of it is untrue. The biggest mistake is people calling it a search engine. It does not search the web for its data.

    Says CNN, “Well first of all its not a search engine, it’s a computational knowledge engine. Roughly translated, that means that this site, Wolfram Alpha figures out results for you based on its own databases, not by scanning the web. In some cases the data’s pretty extensive. Mathematical information, there’s physics searches you can do, historical data.” (as quoted at newsy.com)

    But I think your fundamental argument, that web search may be completely changing and possibly fragmenting, is insightful.
  • Thanks for the comment. Your point re Wolfram is well made raises an interesting question: do we even need to be searching the web to find answers to many popular queries?
  • Thanks for the comment. Your point re Wolfram is well made raises an interesting question: do we even need to be searching the web to find answers to many popular queries?
  • Thanks for the comment. Your point re Wolfram is well made raises an interesting question: do we even need to be searching the web to find answers to many popular queries?
  • yahoo the one se i like but still cant bit google



    Melayu Boleh
  • I agree; although actually what you need is better algorithms -- if not for indexing then for natural language processing, for pattern matching in data, semantics and learning user behaviour.

    Rummbles technology produces a trust network based on your tastes - as you say, people want answers FAST, especially on mobile - and the browsing concept behind good, of the user having to dig for information, breaks on mobile devices.
  • Good point Andrew - different types of algorithm will come to the fore
blog comments powered by Disqus