My current company (thehotlist.com) has been working on what you asked in these two paragraphs for two years.
"Facebook began by testing Instant Personalization with a handful of popular existing sites. But the interesting question for us is: what new things could be built using it that couldn't even have existed before? What new things could you do if you already knew users and their friends the moment they arrived?
One obvious advantage is that it will decrease the friction of trying something new. That will make it easier to launch dramatically novel things. The problem with dramatic innovations is that users often don't understand what you've made till they try it. Now you'll just be able to show them."
Would you like to give me some feedback if you visited thehotlist.com?
And if anyone visits thehotlist.com and has questions to ask me about the experience in using Facebook as main data source, feel free to ask me.
Eugene Jen