« What is a designer doing writing about economics? | Main | Live »

Interlude on Networks

Interlude on Networks

Network. What is a network? Other than the hot terminology of the moment of course. It is a word that gets tossed around a lot, and rarely well understood. I am certainly guilty of tossing it around a lot in these texts, so with some luck at least I try to help you understand it the way I understand it. Or understand them. A network, what is it? what are they?

First lets go the boring and safe route and see what the Oxford English Dictionary has to say. It has 13 different definitions just for the use of the word as a noun, and of those 13 a full nine of them are actually relevant to our use

-A piece of work having the form or construction of a net; an arrangement or structure with intersecting lines and interstices resembling those of a net.
- A chain or system of interconnected immaterial things.
- Any netlike or complex system or collection of interrelated things, as topographical features, lines of transportation, or telecommunications routes (esp. telephone lines).
- Broadcasting. A broadcasting system consisting of a series of transmitters able to be linked together to carry the same programme; a group of radio or television stations linked by such a system; (chiefly U.S.) a large (esp. nationwide) broadcasting company which produces programmes to be relayed to affiliated local stations. Also (occas.): a nationwide broadcasting channel.
- Computing. A system of interconnected computers. Freq. attrib. local area network, wide area network: see the first element.
- An interconnected group or chain of retailers, businesses, or other organizations.
- An interconnected group of people; an organization; spec. a group of people having certain connections (freq. as a result of attending a particular school or university) which may be exploited to gain preferment, information, etc., esp. for professional advantage.
- Math. A graph, esp. a digraph, in which each edge has associated with it a non-negative number (its capacity).
- Math. A diagrammatic representation of interconnected events, processes, etc., used in the planning of complex projects or sequences of operations.

Ok, maybe you can get a sense of why network is such a tricky term to pin down. It can mean just about anything: "A chain or system of interconnected immaterial things", "Any netlike or complex system or collection of interrelated things", "A diagrammatic representation of interconnected events, processes". But a few key features do stand out in that long list of definitions. "Inter" is key, purality is key, a network is never about one thing but about multiple things that are connected. Lets make it really simple, a network is connected things. Lines and nodes. Things and the things that connect the things.

Still pretty much anything can be a network. No that is wrong pretty much anythings can be a network. One thing can never be a network unless that thing can be broken apart. And if anythings can be a network how is a network useful to us?

Well it's useful because so many things are networks, or at least hold the potential to be networks. It's useful not because of the things though, the world is full of useful tools for addressing and understanding things. It's useful because of the inter, the lines, the connections between the things. The world is not full of useful tools for addressing and understanding what connects things together, and a network is important because it marks a start.

If you look back over that list of OED definitions perhaps you will understand just why we can not just throw the term network around casually (although of course we do anyway). A computer network is not the same as a broadcast network, the internet is not NBC, a LAN is not the BBC. Nor is a "system of interconnected immaterial things" the same as a "diagrammatic representation of interconnected events". But that is where it starts to get more complex, for the math always has the potential to also offer an explanation or way towards greater insight into any other network. Yet having potential is not the same as a guarantee. If anything could hold this collection of definitions together, if anything could make these things a network, it is the math. But that is a wildly unproven proposition, we must be wary of thinking that one network might behave like the next just because of the math (although of course we do anyway).

Perhaps the most fundamental question we can ask ourselves about a network is, is it real? Some networks are, others are closer to imaginary, and distinguishing the two is often difficult, sometimes impossible and sometimes a very distinct possibility. A network can be said to be real if we can accurately identify the mechanisms by which all nodes connect to each other. We can call these networks protocological networks or generative networks because it is through protocols embedded in the nodes that the network is generated. Computer networks are the classic example of this, although the postal system or the hub and spokes system of a late 20th century airline are close to as valid examples. Computer networks are idea still though, because the protocols can usually be run entirely by machines and thus enforced with a degree of rigor that mere humans can not always maintain. A renegade postman for instance can opt to trash a load of mail and spend the day at the ballpark, while a digital router is far more likely to stick to the protocol in doing its job.

Most networks are not protocological at all, or at least have yet to be confirmed to be protocological but instead are what could be called traced networks. These networks can not be said to be real, although they certainly can be suspected of being so. In a traced network there is no proof that the connections are actually relevant other than the word of the observer. In other words what ties all the elements together is not necessarily inherent in the elements themselves, but instead has been applied externally. This does not mean that the network in question is not actually cohesive, just that we can not prove that it is. Yet a network by definition transforms a multitude of things into one, a bunch of nodes become one singular network. If that cohesion can not be explained through protocol, through the nodes themselves, than it must be assumed to have been vested there by the identifier.

If one is identifying a traced network, perhaps you can say they are capturing a network, then it is critical that the suspected source of cohesion be identified. There are at least two very valid reasons for vesting a collection of things with the unifying identifier of a network. One is the belief that there actually is a protocol, but that it just has not been discovered yet. Another is that there is no actual cohesion to the network, the network is not real at all but instead an artificial construct used explicitly to collapse a complex set of objects and connections into one neat construct. While this can be an incredibly useful intellectual tool, it is absolutely essential to remember that such a network has no agency of any sort. The artificial network itself can not be used to explain anything, but rather any explanitory powers it possesses must come from somewhere, some mechanism, within the network.

The original networks were quite literally large workings of nets, and their home territory was the fishing villages and trading ports along the ocean. That the first real global networks to develop where in large part also tied to the sea via these ports is most likely a coincidence, although one wonders if the merchants saw echoes of their crisscrossed routes and exchanges in the netting holding down their cargoes. If their home was the sea then the home of the major protocological networks must have been the air, although early communications technology, from complex messenger systems to the telegraph wires where rooted to the ground. But the technology driven worlds of radio and television really forced protocols up into the atmosphere and with satellites beyond. The computer networks of course started on the ground too, but over the past few years are reaching for that wireless space as well, it is perhaps no accident that Apple brands its wireless technolgy "airport".

The traced network really found it's first home in sociology, a discipline, less concerned with the earth, sea and air and more perhaps with the ether. On might suspect their interest in networks originated as an attempt to figure out the mysterious workings of that mythical construct of theirs "society". That might sound like a critique, but it is not for some of the journey's they have produced have been truly fantastic, Howard Becker's immensely enjoyable Art Worlds springs to mind first. Perhaps more importantly boldness of sociologists in jumping into the concept of a network has produced a wealth of insights and techniques from which to build on.

The finest articulation of the traced network concept is Bruno Latour's (as a clear leader of a loose group) Actor Network Theory or ANT. Latour is by no means an uncontroversial figure among sociologists to the point where it might just be best to not look at him as a sociologist at all, an idea he seems to flirt with then discard in his latest and most accessible work Reassembling the Social. Indeed it is the terms that he brings in and then opts to discard that are most potent and telling in that book. Sociology Latour argues should no longer be about "society", a concept, I might add, that is so broad and amorphous that it is better suited for explaining away then for actually explaining. Instead he proposes that sociology should be about associations, before as an aside wishing that he could use the rather glorious term "associology". Instead he concludes "alas, the historical name is 'actor-network-theory', a name that is so awkward, so confusing, so meaningless that it deserves to be kept." With a bit of research into the context though one see that he really means something more along the lines of "I've tried to kill this dumb fucking name for two decades now, and it won't go a way so I better just embrace it again".

As one might gather from the name that historical Actor Network Theory name, associology is where the social sciences have really begun to grapple with the slippery concepts of networks. Not just any networks though but very explicitly just traced networks. I'll let Latour round off the details here: "Network is a concept, not a thing out there. It is a tool to help describe something, not what is being described. It has the same relationship with the topic at hand as a perspective grid to a traditional single point perspective painting: drawn first, the lines might allow one to project a three- dimensional object onto a flat piece of linen; but they are not what is to be painted, only what has allowed the painter to give the impression of depth before they are erased." (Latour 2005, p131)

To distinguish these traced networks from the real, or in Latour's terminology "technical networks", Latour proposed and then immediately discards the simultaneously awkward and elegant "worknet". Well perhaps it is more awkward than anything else, and in the end I doubt it will survive. Yet it does compress the sense that these networks identified by these sociologists and associologists and as networks become more trendy by other disciplines, and other less disciplined people (like me!), are not real at all, but in actuality the products of people doing work in order to cast a net around some set of things. In other words worknets/traced networks are apparatuses of capture. Ways to take the amorphous and hard to pin down collections and trap them into one thing, into one network.

Ultimately though this capture holds it's own trap, for if an actually mechanism of cohesion does not exist for the network, how does one know where the network ends? And how does one know if what is left out is just the necessary holes in the net or in fact whole collections of nodes? The question is really one of cohesion, the tracer is out to turn a collection of things into one network and as such must provide some element of cohesion to transform many into one. The simple answer is to just pick one thing and generate the entire network out of connections going into and out of the thing. In an early work Latour himself chose Louis Pasteur the great French scientist. But this approach is clearly problematic, in that each connection tends to lead towards more things, things that each have their own potential networks. At some point the lines must be drawn and the work turned in to a publisher, or else the associologist themselves might perish. And it is exactly at this point that Latour finds his point of cohesion. What ties this network together should not be tied to any given point in the network, but instead is the book itself!**

Now it is pretty easy to frame that concept as critique, to frame it as a cop out, the associologist (and sociologist too) is out there trying to trace a network and it goes on forever so he just bounds it nice and conveniently where the book ends. But in fact it should not be seen as a critique at all but instead an act of radical liberation. For the associologist in action is at high risk of trapping themselves into the network of their own devising. By seeing the book itself as the final point of cohesion suddenly the are free to stop pretending to be social scientists and instead become social novelists. And if one stop to realize that sociologists are interested in gaining a fuller understanding of the world and that for centuries this sort of understanding has always been best communicated not via academic publishing but via literature.

None of this means that associologists should just go out and make everything up. No, just as good novelists often do extraordinary amounts of research, the associologist also must head out to the field and discover. A good novel after must possess novelty, just as a good work of academic research must produce some new. This means in effect that the associologists actually possess a tremendous unfair advantage over traditional novelists. The associologist after all is connected to an extensive apparatus designed for the discovery of the novel, the poor novelist needs to go out and do all the work on their own. That massive advantage is quickly tossed out the window when one considers that becoming a sociologist requires an extraordinary indoctrination into some techniques of bad writing. Latour, for all his attempts to the contrary, has not done anything to change this situation, expect perhaps to accelerate it, and unfortunately I would probably be just as guilty were I actually to be a sociologist. Of course a little acceleration can go a long way too, perhaps sociology's problems might actually stem from being too adequate as writers. In fields where the writing is even worse, economics being the case to point although physics and biology probably fit as well, it often gets bad enough that someone actually tries to make it better.

While early economics was blessed with a series of writers capable of writing decent prose, it movement deeper and deeper into the obscurities of math was met by a remarkable series of translators, writers capable of transforming the dense concepts of the field into crisp public prose. The latest example the ingenious Freakonomics tag team of an economist with a professional journalist, represents something new though, an attempt by economics profession to hijack sociology's turf. The battlelines were laid a few decades ago, when Margaret Thatcher, channeling Friedrich Hayek declared "there is no such thing as society. There are individual men and women."* When the Freakonomics team, channeling Gary Becker, says "incentives matter" this is ultimately their point, for neoclassical economics wants nothing more than to explain away society as the actions of selfish or hedonistic individuals. Out of all the sociologists it is Latour who provides the strongest strike back against this move, perhaps channeling Thatcher his argument can be boiled down to: there is no such thing as society, only networks.*** But it is now left to the associologists, if any even exist, to tell that story.

*She actually also added "and there are families" to the end of that, she was of course a diplomat. It also can be seen as a foot in the door towards a whole world of other institutions, a critical point elsewhere in this work.

** There is a very clear and interesting parallel with Deleuze and Guattari's overused figure of the rhizome, which they actually explicitly (if not particularly clearly) introduce as being a book. Interestingly Latour, who is familiar with their work does not seem to catch or acknowledge this, although he does occasionally give nods towards the rhizome terminology.

*** Latour does not say it quite that implicitly, one wonders if paraphrasing Thatcher would be too much even for his rather developed taste for the intellectual shock. His actual words are a bit more like this: "Conversely, when our definition of the social is retraced, the common definition of the social has to vanish first. It's hard to see a more extreme contrast: it is either a society or a network." (2005 p.131)

Post a comment

(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)