The New York Times research labs do a lot of interesting things with technology, such as that discussed in Nick Bilton's book about living in the future. Sig posted a link to their Project Cascade, which is a visualization tool that looks at connecting tweets over time.
The short video does the best job of explaining what they are up to, but the essence is looking at links in tweets (both original URL and shortened) to see how ideas propagate. They are showing us a network of actions that people take.
This reminds me of the work of Lilia Efimova and Anjo Anjewierden (and others) have done in building connections between blogs and bloggers, based on the linking information embedded in those blogs. Twitter is interesting because it is a richer body of links, but then it is also poorer in that the surrounding context is by definition much less clear.
An element that is missing in these analyses is the work that happen under the surface. I often find my reading habits and topics crossing boundaries - even the sources themselves jump from one topic to another. And then the way I read them also changes, based on other ideas that are floating through my personal space at the time. This means that links I tweet or blog posts I write are influenced by many ideas: not just the specific links I add to the ether. Those networks are much harder to visualize.