10/11   Does anyone know the secret algorithm behind http://www.google.com It is
        quite good. I'm very impressed. I want to know HOW they did it.
        \_ a bunch of shell scripts -- awk, sed, grep, and pipe.
        \_ My understanding is it's a popularity based results engine.  The
           more people who choose a particular result for a particular query
           the  higher that result will be displayed for similar future
           queries.  I don't work at google but we did a similar thing at the
           search engine company I did work for.
           \-maybe PCI. there are a lot of ways to do this kind of thing.
            there are linear algebra approaches and statistical/baysian
            approaches, depending on problem size and nature of prob. --psb
           \_ I've read a couple of articles that have mentioned it, which
              said that they base their scoring on how many other pages
              link to a particular page, rather than number of times it's
              chosen on their site.  http://www.google.com/why_use.html
              seems to support this.  -niloc
        \_ number of links to a page determine importance. they sort according
           to relevance and important. i know what i'm talking about. -ali
            \_ relevance is only determined by what people actually choose
              out of the links returned from the search. the second factor
              is accuracy, which is the "drift" from relevance. accuracy is
              the perennial problem, since almost all search engines start
              suffering around 5 to 10% of the first links offered.
              \_ Ali is correct.  The stanford prof of the grad students who
                 developed google comercially came to Soda a few weeks ago
                 and said exactly that - it's click throughs and links to
                 that determine ranking -jones
            \_ It has to be more complex than this or Yahoo would show
               up as the #1 hit for every query.  #2 would be Microsoft.
               They *must* take into account the query itself (seems
               obvious, no?) in some way before doing a most-linked sort
               on the results.  So, no, I don't think you know what you're
               talking about.  Are there any CSUA'ers on their architecture
               design, engineering or database staff?  If so, please come
               forward.  Ali having had coffee with someone's secretary at
                                        \_ was this after the mindblasting sex?
               google doesn't impress.
               \_ Eat your words, blasphemer!  The only person I trust more
                  than ali is bh.
        \_ I think they use a variant of the clusterfuck algorithm.
        \_ Algorithm, Heuristic, BAH! They don't interest me and are trolls.
           \_ You are a faggoty bitch.