I just wrote a newsletter article on how to get 'pure' - that is, un-personalized, anonymous - search results from the major search engines. For more on unwanted personalization as well as the uses of Scroogle, DuckDuckGo, Ixquick and StartPage, head over to here.
"Pure" search results is as elusive as the "truth". They are all subject to biases and points or view, whether they're personalized or not.
I didn't read the newsletter article, but I'd be curious to know what you mean by "unwanted" personalization, which would seem to imply that there is some personalization that is "wanted". If I search for "pizza" and the search engine, knowing where I'm located, gives me local establishments, I'm assuming that would be wanted. At what point, however, do you determine that customizing a set of results to the perceived personal preferences of the individual is unwanted?
Personalization is just another ranking factor layered on top of the many, many other ranking factors that go into producing a set of results. When you're dealing with thousands and sometimes millions of results in the result set, I'm not sure that's necessarily a bad thing.
Posted by: Mark Cramer | August 15, 2011 at 09:19 AM
You'll see more about the discussion in the article I linked to here. For researchers, personalization is often not helpful. Just the fact that prior search activity is used as a ranking criterion is a problem, since we often work on several projects at a time, on wildly divergent topics.
Posted by: Mary Ellen Bates | August 15, 2011 at 10:48 AM