Friday, May 09, 2014

Networks and Systems

Let's move up one paragraph in Jon Dron's post.

We've seen already Dron's notion that connectivism should be thought of as a family of ideas. In this paragraph, Dron states more fully what he thinks that family of ideas entails:
All of this is nothing more than nit-picking if we simply accept that connectivism is a broad family of ideas and theories about how to learn in a networked society, all of which adopt a systems view, all of which recognize the distributed nature of knowledge, all of which embrace the role of mediating artefacts, all of which recognize that more is different, all of which adopt a systems perspective, all of which describe or proscribe ways to engage in this new ecology.
It's worth parsing this into individual points, because this is (I guess) connectivism from Dron's perspective:
  • all of which adopt a systems view
  • all of which recognize the distributed nature of knowledge
  • all of which embrace the role of mediating artefacts
  • all of which recognize that more is different
  • all of which adopt a systems perspective
  • all of which describe or proscribe ways to engage in this new ecology.
It's interesting that he makes the point about 'systems' twice. It's also interesting that he thinks that 'mediating artefacts' are important. Both of these points suggests that he doesn't see connectivism as a network theory at all, but as something quite different.

To use his own words, he sees at least some networks as systems. And systems are very different from networks. A clip from Wikipedia will be sufficient to make the point:
  • A system has structure, it contains parts (or components) that are directly or indirectly related to each other;
  • A system has behavior, it exhibits processes that fulfill its function or purpose;
  • A system has interconnectivity: the parts and processes are connected by structural and/or behavioral relationships.
  • A system's structure and behavior may be decomposed via subsystems and sub-processes to elementary parts and process steps.
There are many systems in our everyday experience. A car is a system. A watch is a system. The ecosystem is a system.

So - minimally, I think - Dron thinks (and Dron thinks that George Siemens thinks) that connectivism is a theory about how systems learn. Or, perhaps more minially, it is the theory that systems learn. Or something along these lines.

It's not a new idea, but it's an interesting and valuable idea. Stafford Beer, for example, was a leading and important proponent of the idea (I recall Scott Wilson being a keen reader of Beer's ideas a number of years ago). From this summary of Beer's work, we see again reflected the important properties of systems (p. 16):
  • Living organisms are notable for their ability to maintain their identity, in spite of perturbations in their environment
  • Human organizations also have this characteristic 
  • These systems are purposive  
It's also related to Marvin Minsky's concept of the Society of Mind. The idea here is that in a complex organism, each individual 'agent' has its own function, which it fulfills autonomously, and the organism as a whole is created out of these individual agents. As Push Singh summarizes, "Each agent is on the scale of a typical component of a computer program, like a simple subroutine or data structure, and as with the components of computer programs, agents can be connected and composed into larger systems called societies of agents."

But there are some important differences between the concept of systems and the concept of networks.In a system, the nature of each part matters, and it plays a specific role. In a network, the nature of the part does not matter and it does not play a specific role. Systems are governed by an order, purpose or guiding principle. Networks are not.

We can get an intuitive sense of the difference between the two concepts by thinking of a computer system, like the one you have on your desk, and a computer network, like the one that forms the internet. The system has a set of inter-related parts (and its software is similarly composed of a set of inter-related parts). It functions as a single unit and has specifically designed outcomes.

Now as Beer and others will assert, a self-organizing system (like, say, a society) will find its organizational principle or purpose inherently: it will seek to survive and thrive. But aside from this self-determined principle, it is similar in function to a computer or computer program (which is why Beer is considered one of the founding voices of cybernetics).

So, if Dron is proposing that connectivism is 'a systems view', as he states in this post, I think that Dron's view here is very different from the network view. Indeed, it is a network view only if you think systems are types of networks. But I think there is indeed an additional distinction between systems and networks:
  • in a system, the parts interact with each other but are not changed as a result of that interaction (after all, the nature, purpose and role of the part is important in systems theory)
  • in a network, the parts interact with each other by changing each other. When one part interacts with another, the other begins to do something different from what it did before
We can see that Dron again takes the former, and not the latter, perspective because he views 'mediating artefacts' as part of the system. What would a 'mediating artefacts' be? Well if we read Conole (p. 191) we see that it can be thought of as a mechanism for encoding or representing whatever is learned - thoughts, ideas, knowledge, whatever. She provides as examples things like narratives or case studies, diagrammatic or iconic presentations, or even vocabularies.

A mediating artefact is an component of a cognitivist and representationalist theory of learning; in the chapter under discussion Conole goes on to discuss the role of mediating artifacts as models or representations in the learning design process.It is arguable (and I would argue) that this too distinguishes Dron's perspective from network theory; he wants networks to include the usual components of a cognitivist theory, including models and representations.

Mediating artefacts are part of a concept of systems, indeed, a necessary part of systems that learn, because the system doesn't have the means to learn otherwise. 

It seems clear from this perspective that what Dron wants to do is to include within this 'family of ideas' aspects of a theory that is very different certainly from what I understand connectivism to be, and different certainly from what one would think a 'network' theory of learning would be, all other things being equal. Perhaps he believes that George Siemens is a proponent of systems theory, or cognitivism (and Siemens might well be, though I don't really see it as characteristic of his writing).

The rest of the paragraph is a statement of Dron's argument that connectivism should be thought of as a family of ideas. I believe that in particular Dron believes that approaches based in systems theory and cognitivism ought to be counted as part of connectivism.As I stated earlier, I don't agree.

If I don't care about getting our understanding of learning right or correct, I can ignore what I find to be flaws in the concept, and I can ignore that it actually contradicts other types of network theory, and just proceed on the basis that it is 'useful'. But even an incorrect map can be useful for a time, before it steers you over the cliff. At a certain point, we have to make a choice.

I am not sure whether Dron understands that there is a distinction between systems and networks, and is urging us to simply ignore it, or whether he doesn't understand, and is oblivious to the reasons why I reject propositions that have their origins in systems theory. In the end, it doesn't really matter.

This isn't the place to make the argument - I really just want to draw out the distinction, so people know that Dron wants to put instances of theory B under a heading intended for theory A - but in a nutshell, my objection to systems theory as a basis for learning theory is twofold:
  • first, our ascription of a 'purpose; for post parts and for the whole is mostly composed of rationalizations after the fact, and therefore not a part of the mechanism itself, and
  • second, the more something is a system, the less likely it is able to learn, because of the specificity of the parts and the roles they play (one of the defining characteristic of ecosystems is that they're fragile)

In order to support learning, you have to get away from design. Design is the enemy of learning. Networks learn precisely because they're not designed - the pieces don't play any specific role, the network itself does not have any meaning or purpose. Even survival is an accidental characteristic - it only happens that the networks we see are the networks which survived, because the ones that do not survive have long since gone away.

No comments:

Post a Comment

Your comments will be moderated. Sorry, but it's not a nice world out there.