Reading :: The Exploit

The Exploit: A Theory of Networks
By Alexander R. Galloway and Eugene Thacker

Note: I reviewed the Kindle version, so I don't include page numbers.


The Exploit: A Theory of Networks seeks to develop an ontology of networks. The authors attempt to build this ontology from at least three very different traditions of networks - but unfortunately, rather than putting these traditions into dialogue, they conflate the traditions' definitions of networks. Since these definitions and their applications are so different, The Exploit becomes incoherent in places. In my judgment, it does not reach its goal of developing an ontology of networks.

To understand what went wrong, let's look at three distinct ways that the term network is used in the literature. (See Milton Mueller's excellent book Networks and States for more on how the term is used in different ways.)

In network analysis, a network is a set of nodes (units) connected by links or edges. Since this is a form of analysis, anything can be considered a node - an academic paper (linked via citations), a set of phones (linked by wires and switches), a group of people (linked by communication lines), you name it. As long as the units are consistent and the links are identifiable, you can perform a network analysis. That doesn't mean that the phenomenon you're studying is a network in an objective sense. For instance, you could study a hierarchical organization via a network analysis - and many people do - even though the hierarchical organization is not considered to be a network per se. Network analysis typically involves mathematical tools. (See Barabasi.)

In contrast, networked organizations are a phenomenon rather than a method of analysis. In this tradition, networks are a distinct organizational form, demonstrably and qualitatively different from other organizational forms such as tribes, hierarchical institutions, and markets. In networked organizations, control and discretion are typically decentralized (pushed to the individual level of the organization), although command may still be centralized. Networked organizations are often studied qualitatively. As a phenomenon, networked organizations are networked organizations no matter what analysis you apply. (For instance, al Qaeda is considered a networked organization; but the US Army in World War II was an entirely different kind of organization, a hierarchical one.) For examples, see Ronfeldt; Arquilla & Ronfeldt; Castells; and Toffler, who Ronfeldt cites as a strong influence.

A third construct, sociotechnical networks, is again an analytical frame rather than a phenomenon per se. In this tradition, theorists analyze the social and technical as an inseparable unit made up of varying other units (which could include humans, animals, tools, activities, concepts, etc.). One early example is Bateson's example of the blind-man-with-cane, in which he invites us to consider how the blind man taps his cane to navigate. By himself, he can't; with the cane, he can feel the curb, even though he's not touching it. The system of man-cane-curb is a sensory system in a way that the individual elements are not. This networked understanding of sociotechnical systems is used in somewhat different ways by Deleuze & Guattari, Latour, Engestrom, and others; I wrote a book about it. 

So: at least three kinds of networks and thus three different understandings of what the term network means. Two are analyses, one is a phenomenon. Two are largely qualitative, one is quantitative. Two involve fixed units (in social systems, these are people); one doesn't. They're bound together by the metaphor of network but little more.

Unfortunately, The Exploit doesn't acknowledge or even seem to realize the deep differences among these. I actually gasped out loud when I realized that they had read Arquilla and Ronfeldt's work on networked organizations, but missed how A&R carefully separated this phenomenon from network analysis. Indeed, they soon conflate Arquilla and Ronfeldt's networked organizations with Barabasi's work on "heterogeneous systems, be they terrorism, AIDS, or the Internet." And a couple of paragraphs later, they ask: "what is the principle of political organization that stitches a network together?" - a question that is unanswerable as long as the senses of network are still conflated. Incredibly, late in the book, they claim that Arquilla & Ronfeldt's notion of netwar is grounded in graph theory - despite the fact that A&R don't employ mathematical analysis!

The authors similarly deploy terms from network analysis in ways that seem inconsistent with it. For instance, in discussing SARS and bioterrorism, they claim, "Such skirmishes highlight an important point in our understanding of networks: That the networks of emerging infectious diseases and bioterrorism are networks composed of several subnetworks" such as the genome, epidemiological networks, knowledge networks, and pharmaceutical networks. But they don't explain the units and links of these subnetworks, nor how they qualify as subnetworks of a larger network (as opposed to, say, interfering networks). 

The authors do seem to hew closer to a sociotechnical understanding of networks in some places, particularly when they discuss Deleuze, Negri, and others in this line. Shockingly, although the authors mention Latour near the end of the book, they actually don't mention actor-network theory, nor do they correctly represent Latour's discussion of symmetry in their discussion of it: 

Yet despite Latour's emphasis on the way that human actors are influenced by nonhuman actants, there is a sense in which the nonhuman is still anthropomorphized, a sense in which the circumference of the human is simply expanded.

The principle of symmetry - the stance in which Latour applies the same terms and concepts to humans and nonhumans - the reason why he is constantly having to explain that he doesn't actually have conversation with lampposts - and the authors think that they just might possibly detect a hint of anthropomorphism. And that seems like as good a place as any to conclude this review.