How do I do a people search?

If natural language search had Watson-like accuracy, would people change how they search?

• One of the criticisms of natural language search is that people don't search that way. It takes too much processing power to solve a use case that people don't have. Top Google searches are usually two to three words and not very complex. If writing more complex queries provided high precision, would it change search behavior? For context, see http://searchengineland.com/could-google-play-jeopardy-like-ibm-watson-65038/ and http://blog.agrawals.org/2011/02/16/watson-vs-google/

• Answer:

  Shashwat wrote "if making a natural language search engine would be that easy, don't you think it would have been done already".  An important point to remember here, however, is that Watson had an enormous amount of hardware to rapidly answer questions from one single Alex Trebec.  Google can't possibly devote that much hardware per user to searching for content.  IBM is making rapid progress on further optimizing and scaling Watson, to bring down the total computational cost per query (and is, as always, continuing to improve the hardware, too).  However, I think it will be a very long time before something like Watson is feasible in an advertising-based business.  From a business model perspective, Watson is much better fit for organizations (such as hospitals) that have a relatively constrained user base and can afford to invest a substantial amount per user to making those users more effective. You can read much more about the challenges involved in scaling out Watson in "Making Watson fast" by Epstein et al.: http://ieeexplore.ieee.org/xpl/tocresult.jsp?isnumber=6177717