With all the different social networking platforms around (Facebook, Twitter, LinkedIn and many many others), all exposing APIs allowing you to access the entwined web of social connections, mind the following interesting idea - PeopleRank.
I assume most of you are familiar in one way or another with Google's PageRank algorithm which is at the heart of Google’s search engine. The idea is that website pages are ranked according to their relevancy and their relevancy is determined by two main factors - the number of hyperlinks pointed to them and the quality of the hyperlinks, where quality is defined by the PageRank (a recursive definition) of the referring pages. So Google is constantly running over webpages and updates their PageRank in an iterative process.
The same can be applied over people to determine their online importance. Using social networking APIs, a person can be analyzed by these very same parameters - the number of followers to this person (that is - the number of people that consider him as a friend, not the number of friends he think he has) and the PeopleRank of the people referring to him.
Once we establish importance of people, the next question is how we use this system of ranks. Unlike PageRank, I can't see PeopleRank driving the mechanism of a search engine (not even people search engine), but there are other applications more suitable to use this database.
For example, consider viral marketing. It is a fair assumption that high-ranked people have better viral marketing abilities than low-ranked people. My hunch is that Michael Arrington and even Blonde 2.0 will get a higher rank, then say ... me. So, why not allow an advertiser or other parties who are interested in accessing a large and relevant crowd in an effective manner, to approach high-ranked people as marketing agents. A new business model may emerge, including paying money for these new marketiers in this value-chain.
That's only one example - feel free to use the comments link for other ideas for PeopleRank apps.