Pietro Panzarasa sent me this paper. From the abstract:

The online community represents a prototypical example of a complex evolving network in which connections between users are established over time by online messages. Results indicate that the network topology is compact, densely connected, with a higher clustering than would be expected by chance, and dominated by a minority of users that preside over a disproportionately large amount of connections. The network follows a power-law scaling behaviour that is governed by preferential linking mechanisms. Connections between users are not established at random, but are more likely to originate from users that have already created many connections, and to be directed towards users that have already received many connections. The effects of preferential linking on interpersonal dynamics, however, are mediated by the degree to which users share the same demographic characteristics and affiliation to the same groups. We discuss the implications of the findings for research on evolving social networks. We also describe the managerial implications of our results for a host of applications, from information diffusion to the security and robustness of electronic information systems.

This reminds me a lot of the paper by Gueorgi Kossinets and Duncan Watts on the college email network. Beyond that, I like the graphs, but for once I think they’re too small! (Table 1 should be a graph too. Do we really need to know that some parameter is estimated at 0.2021???) Finally, I like that they combine empirical analysis with some sophisticated modeling (although I’m not up to digesting the model right now…).