The currently issue of Science magazine is entirely dedicated to networks and network science. The issue is packed with interesting articles, and is certain must-read for anyone studying or working with networks. The editors of Science have done well in capturing the breadth of disciplines and interests studying networks. One article that I will not cover in detail but recommend to all readers is Carter Butt’s “Revisiting the Foundation of Network Analysis,” where he discusses what is, and more appropriately, what is not network analysis, and how the science got here.
The article discussing network analysis and national security, in fact, is an excellent example of the wide audience for this topic; however, the thesis of the piece was rather disappointing. In “Counterterrorism’s New Tool: ‘Metanetwork’ Analysis“, we we hear from a veritable who’s-who in the national security/network analysis space. Starting with those on the technology front at Palantir Technologies (the same software we used from Project Grey Goose), to well respected practitioners in academia, business and government such as Marc Sageman, Valdis Krebs and Kathleen Carley, among many others. The article discusses where networks have helped, but also possibly hurt U.S. couterterrorism efforts, which made its focus on so called metanetwork analysis confusing.
In short, metanetworks are simple multiple layers of networks; that is, in any given space there will be a layer of social structure as well as physical (roads and waterways), infrastructure (power and communication), exchange (financial), etc. Metanetwork analysis attempts to examine this complex system as a whole in order to examine how activity on one layer can affect the others, and vice a versa. In theory, this is very appealing, however, in practice this method fails in two major ways.
First, in order for this methodology to have a significant impact, one must have a relatively complete view of all of the network layers, e.g., have really good data. This is often easily done for the non-social layers, however when dealing with counterterrorism the data is undoubtedly poor. In this case, one will most certainly under of over estimate the effect of changes in certain layers due to this incomplete perspective. A brief exchange between Sageman and Carley in the article gets at this point:
But Sageman is skeptical that military progress in Iraq can be chalked up to network analysis. “I’m not convinced [metanetworks] have helped at all,” he says. “An easier explanation [for the drop in sniper attacks] might be the tribal uprising” against the insurgency in Iraq. “There’s no way to know, and that’s a big problem with this field in general.” Carley counters that Sageman “doesn’t understand the methods.”
While it may be true that Sageman does not understand the method completely, he certainly understand network analysis, which brings me to my second problem with metanetworks: how is this different from thorough and methodical traditional network analysis? Part of the claimed strength of metanetworks is accounting for the context of social networks within a complex system, however, any good network analysis should always be accounting for this context. This can be done analytically, by adding attributes to actors and edges, or qualitatively by consulting area and subject-matter experts relevant to a given study.
Unfortunately, we never get from this article how metanetwork analysis is different, rather just a tertiary exploration of its benefits. Finally, I would like to not that there is a small insert in this issue about the ‘mosaic philosophy,’ entitled “Investigating Networks: The Dark Side.” I have covered this absurd interpretation of network theory is the past, but would direct readers to this discussion after reading its coverage in Science.
Photo: AAAS
Automatically Generated Related posts:
[...] Conway has a review of the July 2009 issue of Science dedicated to Network Science [via Zero Intelligence Agents]: The currently issue of Science magazine is entirely dedicated to networks and network science. [...]