tive in tearing apart a pure scale-free network (such as the Dark Web, with a high degree-betweenness-rank-or-der correlation, 0.70) in which hubs function simultaneously as bridges connecting different parts of the network.

conclusion

Dark networks (such as those involving terrorists and criminal narcotics traffickers) are hidden from nonparticipants yet could have a devastating effect on our social order and economy. Understanding their topology yields greater insight into the nature of clandestine organizations and could help develop effective disruptive strategies. However, obtaining reliable data about dark networks is extremely difficult, so our understanding of them remains largely hypothetical. To the best of our knowledge, the data sets we explore here, though subject to limitations, are the first to allow for statistical analysis of the topologies of dark networks.

We found that the covert networks we studied share many common topological properties with other types of networks. Their efficiency in terms of communication and information flow and commands can be tied to their small-world structures, which are characterized by short average path length and a high clustering coefficient. In addition, we found that due to their small-world properties, dark networks are more vulnerable to attack on their bridges that connect different communities within them than to attacks on their hubs. This finding may give authorities insight for intelligence and security purposes.

Another interesting finding about the three elicited human networks we studied is that their substantially high clustering coefficients (not always present in other empirical networks) are difficult to regenerate based on only known network effects (such as preferential attachment and small-world effects). Other mechanisms (such as recruitment) may also play an important role in network evolution. Other research has found that alternative mechanisms (such as highly optimized tolerance) may govern the evolution of many complex systems in environments characterized by high

risk and uncertainty. ³ Our future research will focus on the effects of such alternative mechanisms on network topology. In addition, our findings are

to the best of our all based on a static view of the net- _{works we studied; that is, we did not} knowledge, the data consider a large variety of dynamics _{sets we explore} that might have taken place in the evo- _{lution of the networks, so evolution} here, though subject study is definitely in our plans for fu- to limitations, are ^{ture research.} Please also note that care is needed the first to allow for when interpreting these findings. Because dark networks are covert and

statistical analysis largely unknown, hidden links may ^{of the topologies be missing in the elicited networks.} These links may play a critical role in of dark networks. maintaining the function of the covert organizations. As a result, one must be extremely cautious when a decision is to be made to disrupt them.

References

1. Albert, R. and Barabási, A.-L. Statistical mechanics of complex networks. Reviews of Modern Physics 74, 1 (Jan. 2002), 47–97.

2. Barabási, A.-L., Jeong, H., Zéda, Z., Ravasz, E., Schubert, A., and Vicsek, T. Evolution of the social network of scientific collaborations. Physica A, 311, 3–4 (Aug. 2002), 590–614.

3. Carlson, J.M. and Doyle, J. Highly optimized tolerance: A mechanism for power laws in designed systems. Physical Review E60, 2 (Aug. 1999), 1412–1427.

4. Crucitti, P., Latora, V., Marchiori, M., and Rapisarda, A.
Efficiency of scale-free networks: Error and attack
tolerance. Physica A 320 (Mar. 2003), 622–642.

5. Holme, P., Kim, B.J., Yoon, C.N., and Han, S.K. Attack vulnerability of complex networks. Physical Review E 65, Article No. 056109 (May 2002), 1–14.

6. Levitt, S.D. and Dubner, S.J. Freakonomics: A Rogue Economist Explores the Hidden Side of Everything. William Morrow, New York, 2005.

7. Liben-Nowell, D. and Kleinberg, J. The link prediction problem for social networks. Journal of the American Society for Information Science and Technology 58, 7 (May 2007), 1019–1031.

8. Newman, M.E.J. Mixing patterns in networks. Physical Review E 67, 2, Article No. 026126 (Feb. 2003), 1–13.

9. Sageman, M. Understanding Terror Networks. University of Pennsylvania Press, Philadelphia, PA, 2004.

10. Wasserman, S. and Faust, K. Social Network Analysis: Methods and Applications. Cambridge University Press, Cambridge, U. K., 1994.

11. Watts, D.J. and Strogatz, S.H. Collective dynamics of ‘small-world’ networks. Nature 393, 6684 (June 1998), 440–442.

12. Xu, J. and Chen, H. Untangling criminal networks:

A case study. In Proceedings of the First NSF/NIJ Symposium on Intelligence and Security Informatics (Tucson, AZ, June 2–3, 2003). Springer, Berlin, Germany, 2003, 232–248.

Jennifer Xu ( [e-mail]) is an assistant professor of computer information systems in Bentley College, Waltham, MA.

hsinchun Chen ( [e-mail]) is McClelland Endowed Professor in the Department of Management Information Systems and head of the Artificial Intelligence Lab at the University of Arizona, Tucson, AZ.

© 2008 ACM 0001-0782/08/1000 $5.00