WolframAlpha and fact search

By May 11, 2009Search

There has been a lot of chat in the last couple of weeks about the new search engine WolframAlpha which is due to launch this week, including a detailed article this morning in the New York Times.  As the founder (Mr Wolfram) says, he won’t compete head on with Google (never a good idea…), but rather than help you find web pages it tells you the answers to fact based questions, lets call this fact search.  For example you might ask it “which is taller, the Eiffel Tower or the Seattle needle?”.  From a technology perspective then it has two things – natural language processing technology to decipher the questions and a huge database of facts+/ways of inferring them to provide the answers.  This is not unlike the UK startup TrueKnowledge, which has generated some hype of its own.

For me fact search would be a useful service – in the past week I have tried and failed on Google with two questions that these companies should be able to help with 1. Why are so many UK pubs called ‘The Cross Keys’, and 2. What is the breakdown of UK GDP by sector?  Further, I suspect that once services like these become commonplace we would start to reformulate many of our Google queries into question form.

I also think that Google is unlikely to try and offer an equivalent service any time soon, as it would be hard for them to do so except as a separate offering which would be complicated for them to promote.  (Although what they can do, and are doing, is to pick off the most common fact based queries and answer them right there in the results page – e.g. a search for ‘weather London’ produces a three day forecast at the top of the results page.)

On the downside, as my friend Shak once pointed out to me, it is unclear how well these sort of queries will monetise, as they don’t contain the same sort of buying intent as much of the things we search for on Google.

  • brian_gilbert

    As I understand it so far Wolfram Alpha adds to the huge body of consistent logic in Mathematica a new body of consistent data. Looks like the nemesis of specious arguments.

  • Agree with Shak's view on difficulty to monetize based upon intent to buy, however, W|A offers the ability to buy data, which is a huge business now (called research and analysis). So, there's definitely a biz model in there. The other big problem that W|A will highlight is that there are big gaps in consistent data that is available. W|A can build a big competitive advantage if it ties up exclusive rights to difficult to find, yet valuable datasets with revshare deals.

  • Agree with Shak's view on difficulty to monetize based upon intent to buy, however, W|A offers the ability to buy data, which is a huge business now (called research and analysis). So, there's definitely a biz model in there. The other big problem that W|A will highlight is that there are big gaps in consistent data that is available. W|A can build a big competitive advantage if it ties up exclusive rights to difficult to find, yet valuable datasets with revshare deals.