This document summarizes and compares three different types of networks: social network analysis, sociotechnical networks, and organizational networks. Social network analysis focuses on relationships between homogeneous human nodes. Sociotechnical networks consider relationships between heterogeneous human and non-human nodes. Organizational networks examine relationships and lines of authority between people within a phenomenon. The document outlines the key differences between each network type in terms of their nodes, links, analysis approach, and whether the network is viewed as a frame of analysis or phenomenon.
2. “When we talk about ‘networks’
are we talking about
technologies, or societal
organizations, or both? Or are we
simply projecting the latest
metaphor into any and every kind
of social relationship we can
see?” — Milton Mueller
5. (Social) Network
analysis
Nodes: can be anything, as long as they
are a specific type of thing
Links/edges: Can be any relationship, as
long as it’s a specific type of relationship
Analysis: Typically quantitative
The network is: a frame of analysis
6. Sociotechnical
networks
Nodes: can be anything, often of any
category
Links/edges: Can be any relationship, as
long as participants interpret it as a
relationship
Analysis: Typically qualitative
The network is: a frame of analysis
8. The three networks, compared
Social Network
Analysis
Sociotechnical
Networks
Networked
Organizations
The network is... Analysis Analysis Phenomenon
Nodes are... Homogeneous
(human beings)
Heterogeneous
(human and
nonhuman actants)
Homogeneous
(human beings)
Links are ... Homogeneous
(communicative
connections)
Heterogeneous
(rhetorical and
cognitive
associations)
Heterogeneous
(communicative
connections and lines
of authority)
Data collection
methods are ...
Quantitative Qualitative Quantitative and
qualitative
Intro self
Title: Like the beginning of a joke. Because the term “network” has become more and more confusing—”Who’s on First” level of confusing. At least three things are meant by the term “network” in the social sciences. Since it’s the same word, readers are not always sure what lineage is being referenced. But people such as Arquilla, Barabasi, Burt, Castells, Latour, and Engestrom are not all talking about the same thing.
These three understandings have developed from distinct traditions, involve distinct concepts and theories, and produce distinct analyses. They’re different, but they’re sometimes used synonymously, and that leads to some undesirable incoherence.
What I want to do today, based on Ch. 3 of my book All Edge, is to disentangle these three sorts of networks. Let’s talk about what each does, what they’re good for, and how we might—cautiously—use them together.
Photo credit: Marlon E, https://www.flickr.com/photos/sjsharktank/6031423074
Mueller, in Networks and States. Good question. For “network” to mean something, it has to be more than a metaphor, it has to be an analytical concept with principles for application. Is that what it is? Maybe we could ask some of the scholars who are using the concept.
We could ask a lot of people—here are just a few that I cited in All Edge. And if we did, we would get at least three distinct answers.
And it’s okay to have three distinct answers as long as they’re analytically distinct and as long as we know which one is operant. We need to disentangle them. Because networks are always getting tangled.
So what are these three networks?
Photo credit: Akash Kataruka, https://www.flickr.com/photos/akash_k/140910730
Social network analysis (SNA) involves exploring direct, explicit connections among human beings, typically communication connections. (Network analysis could be applied to anything; social network analysis specifically takes human beings as nodes.) As the name implies, SNA is a sort of analysis rather than a description of a distinct phenomenon. SNA typically involves quantifying these connections; its data collection tends to focus on quantifiable data such as surveys [12; 20; 27; 51], interviews [19; 31; 51]; text analysis [24; 37; 42]; and case studies [51]. SNA excels at finding detecting weak and strong ties [27; 31]; identifying structural holes, or points at which people in a network don’t directly connect [12]; and brokers, people who serve to connect others across these structural holes [12]. Beyond academic studies, popular books also describe SNA (e.g., Barabasi; Gladwell).
Ex: Kevin Bacon game. Six degrees of separation. From Granovetter.
Barabasi, Burt, Granovetter, Krackhardt, Small.
Notice that SNA does not describe a specific kind of organization, phenomenon, or technology: Any set of connections among people can be mapped as a social network; these links tend to be underdefined, often focusing on explicit communication (conversations, phone calls, email) or strength of relationships as self-reported on a Likert scale. SNA is also not well equipped to examine such connections in terms of rhetoric, argument, or persuasion, since its basic analysis is quantitative rather than qualitative and focused on identifiable connections rather than evaluating their impact.
Examples:
How, and how frequently, individuals contact each other in an organization.
Who people use as contacts when searching for jobs.
How phones are connected via phone lines.
How scientific articles are cited by other scientific articles
The latter: Explored in Latour 1986. Articles are nodes, citations are links. But Latour became more interested in a second type of network.
Whereas SNA takes human beings as nodes, sociotechnical networks take other things as nodes as well: artifacts, activities, and beliefs. As the name implies, sociotechnical networks describe social and technical dimensions of a given activity.
Ex: Patrick gets in his car. It doesn’t start. He then sees the car as a collection of parts-- he “opens” the “black box.”
The “black box” isn’t just nonhuman parts, either.
Ex: English or metric socket set? (norms)
Ex: Seatbelt light and beep (morality)
In an analysis based on sociotechnical networks, the emphasis is on the qualitative, social relationships of a system. Any node can be understood as a network in its own right.
The implementation varies widely. For instance, in actor-network theory, humans and nonhumans are treated equally as heterogeneous nodes [32]. In activity theory, humans act (individually or collectively) as agents, mediated by artifacts [30]; the entire object-oriented activity is the node.
Sociotechnical networks are analytical constructs that can be applied to any sociotechnical system: that is, like SNA, sociotechnical networks are a frame of analysis rather than a distinct phenomenon. But unlike SNA, sociotechnical networks rarely involve quantification. For instance, since anything can be an actor (node) in actor-network theory, no firm principle exists that could unambiguously ground a quantified analysis. Humans and nonhumans, concrete and abstract, an atom and a galaxy, each can conceivably be a node in the same network. Links are underdefined relations. And any node contains its own links—any actor is itself a network. Thus a mathematical approach is no longer productive because the nodes and links are no longer stable.
Thus studies involving sociotechnical networks tend to involve qualitative data collection methods: observations, interviews, and text analyses that focus on persuasion, interpretation, mediation, and alliances.
Again, sociotechnical networks are an analytical construct: they do not describe a specific kind of organization or phenomenon. Any observable set of connections among humans and nonhumans can be analyzed as a sociotechnical network.
Latour, Callon, Law, and Mol use this concept of networks. But it also gets picked up by activity theorists such as Engestrom and Nardi. In activity theory, an activity is relatively defined according to a cyclical objective and the apparatus of people, rules, and tools that help to achieve it. An activity network involves relations between activities, which are again relatively defined.
When we talk about activity systems, though, we are talking about collections of people doing things. ...
In contrast to the other sorts of networks we have discussed, networked organizations are phenomena. That is, the network is not a frame of analysis, but rather a distinct organizational form that we can contrast to other organizational forms such as tribes, hierarchies, and markets. In networked organizations, human beings constitute the nodes, but the links are lines of alliance and authority. Networked organizations are studied using both quantitative and qualitative methods.
When we look at the work of Castells, Child, Boisot, Powell, or the work of warfare theorists such as Arquilla and Ronfeldt, we are talking about organizational networks. When someone describes al Qaeda as a “network,” they are using this concept. They are saying: al Qaeda is configured in a way that is identifiably different from a hierarchy, market, or clan.
Remember: From the perspective of social network analysis and sociotechnical networks, any configuration of human beings can be examined as a network, because the network is the analytical approach. But from the perspective of organizational networks, the network is a phenomenon. If al Qaeda is (structured as) a network, that has implications for how people interact, how power and authority are asserted, and how people construct arguments that will be persuasive to that network. Organizational networks function differently from organizations with different architectures.
That’s especially interesting to us because al Qaeda is not the only entity to take on a networked organizational form. As I discuss in All Edge, networked organizations have become far more common due to a number of factors, including automation, new information and communication technologies, and the resulting drop in communication costs. Even in hierarchical organizations, the associational links across the hierarchy are becoming more common; mutual adjustment is becoming easier to initiate and sustain. And many of the functions that once resided within hierarchical organizations are migrating to temporary networks of specialists. (outsourcing noncore functions.)
Ex: graphic design
1970s: in an ad agency, in a specific unit
1990s: crossfunctional teams
2010s: outsourced to a freelancer with laptop and home office
So when we compare these different understandings of networks along these specific dimensions, suddenly they look very different. They really are different things, good for different applications.
But: Once we separate them analytically, we can apply different types of networks judiciously—if we know what we’re doing.
Ex: SNA and org networks: Is the organizational network functioning efficiently and effectively? Where are the structural holes?
This combination is sometimes used in warfare analysis, especially in understanding irregular warfare. See Ronfeldt et al on the Zapatista movement. (Discuss: Marxist rebels; NGOs; villagers living in communities.)
Photo credit: Jose Villa, https://en.wikipedia.org/wiki/Zapatista_Army_of_National_Liberation#/media/File:Zap1.jpg
Ex: Sociotechnical networks and organizational networks: How are new actors inducted into the organizational network? How do different organizations interact and modify each other’s mutual goals?
This is what I tried to do in All Edge. (Especially: once freelancers take over noncore functions, how do they scale up and down to take on different jobs?)
Other combinations are possible, and maybe even a three-network analysis would be possible—but tricky. The important thing is to remember that when these three networks walk into a bar, they really are three different things. They do different things, they describe different kinds of relationships.
Methodologically and theoretically -- Don’t let them get tangled up!
Photo credit: Marlon E, https://www.flickr.com/photos/sjsharktank/6031423074