From: marina@sapir.us   
      
   On Jul 20, 10:31 pm, Ted Dunning  wrote:   
   > This doesn't actually work because the measurements of number items   
   > above and below are themselves random variables.  This means that this   
   > procedure will give you a ranking, but it will give you neither an   
   > estimate of the quality of the ranking, a set of alternative   
   > orderings, nor any intuition about which question to ask next.   
   >   
      
   This works exactly because they are random variables. U-statistics is   
   a statistics, it is designed to work with random variables.   
      
   There were no  questions about quality of ranking or alternative   
   rankings. The question was - how to order? U-statistics is a standar   
   answer. There is lots of literature about it and its properties.   
      
   There was no question about "which question to ask next". It does not   
   appear to be a primary concern.   
      
   [ comp.ai is moderated ... your article may take a while to appear. ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)