Since it seems to be all about personalized search today, I think it's worth mentioning that Findory has been offering personalized web search results (in addition to personalized news and blog search results) since November 2004.
Comments from Findory CEO, Greg Linden, about Google's latest offering and how Findory's personalized web search results work in this blog post. He also shared the following with me via an e-mail this afternoon:
For example, Google's personalized web search might detect that you tend to be interested in computers and library search results and bias all of your searches generally toward search results about computers and library. In contrast, Findory might notice that you just looked up a movie on IMDB.com and then did another search for a movie, so it would move relevant IMDB search results up higher. Or Findory might notice that you clicked on a research paper on a related topic in a search earlier that day and surface another, related research paper in your latest search results. Findory's
technique is more fine-grained, focusing in on your mission -- what you are doing right now -- and helping you find what you need.
Also, Filangy, the impressive web tool that I've blogged about several times (and use frequently) also provides personalized results based on recency and frequency of usage of pages. Btw, the Filangy beta is now open. You no longer need an invite to check out the service. It's more than worth a look.
Only a Thought
One question I have about personalized web results is:
Does your web search history and what you've already looked at always provide a good indication as to what you want you might want to find during a future search session? I'm not so sure.
A searcher is likely to have interests, needs, and a knowledge base that change from day to day, if not from minute to minute. A tool that would allow a searcher either build a more personalized query or quickly tweak their search results based on criteria important in a specific search situation (intent, time frame, reading level, currency, type of content, features on page, etc.) might be useful. I'm thinking of something along the lines of what Yahoo offers with SmartSort and Mindset.