search [site:sampledomain] yields 17000+ results
search [site:sampledomain keyword] yields 53 results where keyword is a word common to all the pages in the domain.
search [uniquekeyword] yields 53 results where uniquekeyword is common to all pages and unique only to the site.

This tells me that somehow Yahoo has a file of all 17000+ pages that responds to the site: command but not to the general query command.

This smells like Google’s supplemental index where a lot of crap pages are kept just for the sake of being counted in the “total” index!

Even though Yahoo does well on the “Search Engine Relevancy Challenge” by Rusty Brick, I’m curious about what you would think of many search marketing people on many different forums being given the challenge (broken down by catagory/topic) of trying to come up with ways of measuring relevancy. Danny Sullivan’s recent article: http://searchenginewatch.com/searchday/article.php/3527636 has a great challenge for search engines, but why not start out challenging many people who work with keywords and relevancy every day? After that, then the search engines could “Establish a research center, a consortium or something and a methodology that all will agree upon.” as Danny suggests. Getting all forum participants to agree on results would not need to happen, and it would surly get media attention on the more lofty goals of search engines.

Sincerely,
Bill Kelm

http://sethf.com/infothought/blog/archives/000899.html