I put up a little project on github last night, twitter mutual friends. This is the page that should serve as the reference point for it.
I started up the effort while trying to make sense of the "follow or don't follow" decision when someone follows you. Certainly following everyone blindly is a bad mistake, but ignoring new followers is a bad mistake too, and sometimes you just need to prune back the following stream to leave room for new voices.
The existing Twitter feedback for these is very limited, based almost entirely on numerical counts. I thought I could do better by examining what kind of mutual relationships already existed between me and other people and to see if I could characterize folks based on who else they followed to know at least which part of my life they might have found me by.
Thus, the twitter mutual friends effort. The simplest version of this downloads two friend lists, and looks for overlap. The list of overlapping ids is walked through, noting who the overlapping people are, what their last twitter was, how many people they in turn were following, and the like.
You could look either at friends, followers, or mutual friend/follow pairs; you can derive things from the text of recent tweets from mutual friends, or their follow counts, or other second derivatives of mutuality or reciprocity. At some point there's a nice visual of all of this, as well as some good textual analysis ("trending words in our mutual followers"). But as a start this is just an effort to get beyond simple counting and into network views of twitter.