How to Study a cMOOC: Part 3 of a list for #CLMOOC

So I'm closely reading Dillenbourg's 1999 introduction to collaborative learning to see what guidance it may provide for studying cMOOCs such as Rhizo14 and CLMOOC. Dillenbourg's first point about looking at collaborative learning from different scales strikes me as most helpful, but his second point about defining learning at best provides negative examples: what not to do. His third and final point about defining collaboration also seems to be a negative example of how not to approach cMOOCs. Let's see what Dillenbourg has to say.

Dillenbourg looks at collaboration from four points of view, or on four different scales:

1. situation,
2. interactions,
3. mechanisms, and
4. effects.

I like that he keeps his commitment to working with any complex system from a variety of scales. This is a key, I think, to looking at any cMOOC. But he is clearly talking about collaboration and not cooperation, an issue that Jenny Mackness opened for me and that I addressed in my last post. This suggests to me that his approach to collaboration may be another negative example for me.

Collaborative Situations - The first way to explore collaboration, Dillenbourg says, is as a situation in which agents:

1. are more or less at the same level: collaboration as a situation requires a symmetry in which all collaborating agents have similar freedom and ability to act, a similar level of knowledge, and similar status within the group. It seems that asymmetry disrupts collaboration. If you know significantly more than I, then you may try boss me, breaking the collaboration. A collaborative group, then, requires a certain homogeneity, a red flag for rhizomatic, cMOOC enthusiasts.
2. can perform the same actions: again an implied homogeneity among collaborative agents. No one is particularly more skillful than the others, though Dillenbourg does allow for some specialization of skills within the collaborative group. Too much, though, undermines the collaborative ethos, it seems.
3. have a common goal: collaboration requires a shared goal, whereas competition requires conflicting goals (this either/or approach leaves out cooperation altogether, it seems to me). Shared goals can partially assigned at the outset of collaboration, but also depends on the collaborating agents negotiating their shared goal, and in the process becoming aware of their mutual dependence.
4. work together: Dillenbourg finally mentions cooperation, but I don't think it helps much. He notes that some scholars use the terms interchangeably, while some distinguish them: cooperation is when group agents "split the work, solve sub-tasks individually and then assemble the partial results into the final output," and collaboration is when agents "do the work together." I don't think Jenny Mackness or Stephen Downes would favor this definition of cooperation.

Collaborative Interactions - The second way Dillenbourg looks at collaboration is by the nature of a group's interactions:

1. interactivity: collaborative interactions are, well, interactive, which seems intuitively obvious, but Dillenbourg defines the degree of interactivity not by frequency but by the intensity of the interactions, by the degree to which the interactions influence the peers' cognitive processes. This is an interesting approach to interactivity, but as Dillenbourg notes, devilishly difficult to measure.
2. synchronicity: collaborative interactions are synchronous says Dillenbourg. Collaborators expect their peers to "wait for [their] message and … process the message as soon as it is delivered." Asynchronous communication is out for collaboration, but I see no advantage to Dillenbourg's definition here. Linux, for instance, is a collaborative effort that seems to thrive on both synchronous and asynchronous communication. I hope Dillenbourg has dropped this.
3. negotiability: finally, collaborative interactions are negotiated rather than mandated. Negotiation implies for Dillenbourg space among the collaborators for negotiation and misunderstanding, a space for constructing shared meaning regarding the project and its execution.

Collaborative Processes - The third way Dillenbourg explores collaboration is by the mechanisms that enable group interactions and learning. He starts with those mechanisms that are common to individual learning but also occur in group learning: induction, cognitive load, self-explanation, and conflict. He then talks about those mechanisms more closely associated with group learning: internalization, appropriation, and mutual modeling.

Collaborative Effects - The final way Dillenbourg considers collaboration is through its effects on learning, usually as a measurement of individual task performance. He notes two main problems with measuring the effects of collaborative learning:

1. It is difficult to isolate in a collaborative situation, with its many contexts and interactions, the specific causes of any identifiable learning.
2. It is difficult to infer the degree of group learning from measurements of individual learning.

So what does this have to say about how the Rhizo14 auto-ethnography group, for instance, should go about looking at Rhizo14?

On the positive side, researchers will benefit from a multi-scale, multi-perspective approach to exploring any complex, multi-scale, self-organizing system such as a cMOOC. This is not to suggest that any single researcher or research effort must attempt to cover all scales and aspects of a cMOOC, but it does suggest that a research effort will benefit if it creates ample space for multiple approaches to the same system.

I am also interested in Dillenbourg's characterization of interactivity not as a quantity but as a quality. He doesn't measure the number of engagements so much as the degree to which an engagement affects a colleague's cognitive processes. This reminds me of Deleuze and Guattari's decalcomania, one of the six features of the rhizome and perhaps the feature least mentioned and discussed by others. Perhaps it is the least understood, or least impressive, but I think that is unfortunate. Decalcomania appears to me to be the process by which memes (a term coined by Richard Dawkins about the same time Deleuze and Guattari were writing A Thousand Plateaus, so perhaps unknown to them) spread through a system. It is a kind of staining. One is stained through an engagement with another and an exchange of energy, matter, information, organization, or all four. Dillenbourg rightly notes that this view of interactivity is difficult to measure and quantify, but everyone in a cMOOC experiences it: an idea emerges somewhere in the network and passes along tweets, posts, and discussions to many others in the network and outside it. Some stain, some don't. The more who stain, the more pervasive and powerful the meme and the more likely it is to spread more. There is a network power law at work here and definitely network propagation (decalcomania) which cMOOC investigators should keep in mind.

Then, Dillenbourg's take on the negotiability of collaborative interactions may hold as well for cooperative networks, but I have to think much more about this. Something wants me to frame the notion of negotiated social contracts in a different way, but I'm not ready to do it now, so I'll just pass.

Finally, Dillenbourg's thoughts about collaborative processes, the mechanisms that enable collaborative learning, seem equally relevant to cooperation as to collaboration. I don't think Dillenbourg was attempting to be exhaustive in the processes he discusses, and I see no reason why induction, cognitive load, self-explanation, conflict, internalization, appropriation, and mutual modeling would not play well in cooperative learning. Though it is possible that other mechanisms that I cannot think of just now play better in cooperative learning than in collaborative learning.

On the negative side, Dillenbourg's 1999 delineation of collaboration and cooperation seems, to me, to miss the concepts emerging in current conversations about cMOOCs. Jenny Mackness pointed to Stephen Downes' careful explanation about the differences between the two, and Dillenbourg's use of cooperation in this article does not match so well. Downes' distinction hinges on the differences between groups and networks and the role of the individual in each. In collaborative groups, individuals are subsumed under the group, becoming a part of the group, while in cooperative networks the individual is not subsumed by the collection of agents; rather, the network is an emergent property of the collection of individuals and their interactions. Dillenbourg's use of cooperation as mostly a difference in distributing the workload of the group misses most of the richness of Downes' use and affords very little help in understanding cMOOCs.

On the other hand, Dillenbourg's use of the term collaboration seems reasonably consistent with Downes' use, so I assume that they are mostly talking about the same kind of system. Thus, I take away from Dillenbourg's article some approaches not to use with cMOOCs. For instance, collaborative groups for Dillenbourg and Downes imply a certain homogeneity in the collection of individuals: similar capabilities, actions, goals, and affordances, especially similar languages and technologies. cMOOCs, on the other hand, are open to heterogeneity, and any effort to explore and map a cMOOC must account for this heterogeneity, this openness to divergent actions, aims, abilities, and affordances. For instance, lurkers have negative roles in a collaborative group and are often eliminated, but they can play a very productive role in a cooperative network, or rhizomatic community.

Dillenbourg's assertion that collaborative groups rely on synchronous communication doesn't match my understanding of what happens online in either collaborative groups or cooperative networks. I take the Linux project to be a monstrously successful collaboration, and I'm confident that group uses both synchronous and asynchronous communications to collaborate. I know that cooperative networks such as cMOOCs use both, so I see no need to limit either collaboration or cooperation to one or the other types of communication. I do see, however, a need to study how and why agents will chose one over the other and what each affords the agents in a system.

Finally, Dillenbourg's handling of collaborative effects seems hampered by its focus on local causality. Both online and f2f collaborative and cooperative systems can be explored and explained as much, perhaps more, by circular and global causalities as by local causalities. Self-organizing systems can seldom be explained by the local pushes of the one-to-one interactions among its constituent agents. Rather, one must include the global pull of the larger, emerging system as it seeks a comfortable function and form within its ecosystem. Likewise, the learning that emerges within a complex system must include circular causalities which account for the continuous flow and feedback of information, energy, and organization among the individual agents and between the emerging system and its ecosystem. To grasp a cMOOC, our field of reality must be enlarged. For instance, the Rhizo14 auto-ethnography cannot simply ask if an instructional technique employed by Dave Cormier led to a specific learning in any of the Rhizo14 participants, as we might do in a traditional classroom. Rather, the group should expand its field to explore how Rhizo14 emerged and self-organized, shaping and ordering itself around various conversational spaces such as Facebook, Twitter, P2PU, blogs, and G+. What global causes pulled Rhizo14 into this particular organization? The group should explore how one conversation fed into another conversation, reshaping both conversations in a mutually causal feedback loop. What circular causes looped Rhizo14 into poetic expressions? Did DS106 and CLMOOC feed into Rhizo14? Has Rhizo14 fed back into those systems? These are the kinds of questions that must frame any discussion of a complex, multi-scale system, I think.

How to Study a cMOOC: A list for #CLMOOC

I’ve joined CLMOOC, and the first two tasks are underway:

1. to create a how to as a way of introducing ourselves, and 
2. to create a list.

This is my attempt to do both at the same time. I propose here a list of approaches to studying cMOOCs, such as CLMOOC and Rhizo14. Many of my Rhizo14 friends are here as well, so they will recognize some of the themes I touch on here, and those who are not familiar with me will learn something about me. This series of posts that I propose to write (I already see that I can’t do this all in one post) comes from a Rhizo14 Facebook conversation that started when Maha Bali asked how Rhizo14 was different from earlier cMOOCs such as CCK008. She sparked a long conversation that is well worth reading, but along the way, someone mentioned Pierre Dillenbourg’s study of collaborative learning. I was not familiar with Dillenbourg, but he seemed to have some respect amongst the Rhizo14 group, so I moved him to the top of my reading list to see if I could learn something about how to investigate a cMOOC, which I take to be a particularly enjoyable and rewarding instance of collaborative learning.

So I read first Dillenbourg’s introductory chapter What do you mean by ‘collaborative learning’? in his book Collaborative-learning: Cognitive and Computational Approaches (1999). Keep in mind that MOOCs did not exist in 1999, so Dillenbourg cannot be held accountable for them; still, I hoped that he might provide a useful approach to studying online, collaborative courses such as Rhizo14 and CLMOOC. I am particularly interested in wrapping my head around an auto-ethnographic study that is emerging in the Rhizo14 community. In his introduction, Dillenbourg explores collaborative learning “along three dimensions: the scale of the collaborative situation (group size and time span), what is referred to as ‘learning’ and what is referred to as ‘collaboration’. This seemed promising to me, so I want to see if his experience and insights can inform the Rhizo14 auto-ethnography (AE). 

The variety of scales: Dillenbourg insists that different scales of collaborative learning require “different theoretical tools in order to grasp phenomena on various scales” and that we should not generalize the results from studying a handful of people collaborating in one location for an hour to 40 or 50 people (he seems not to have imagined hundreds or even thousands of collaborators back in 1999) collaborating in different locations over the course of a year, and vice versa. Moreover, scale is not so much a property of the object as it is “a property of the observer, who selects the most appropriate unit of analysis.” Likewise, the observer defines the agents, or “functional units”, within a collaborative learning, which can include devices and systems as well as humans.

I like that Dillenbourg begins with scale issues, as I believe that online, collaborative learning is a function of complex, multi-scale networks, or what Deleuze and Guattari call rhizomes. (Networks and rhizomes are not synonymous, but they share some characteristics, and I find it easy to shift between the terms, using networks when I want a more precise model and rhizomes when I want a more expansive metaphor.) For instance, Rhizo14 has generated an auto-ethnographic study that begins with a collection of stories from Rhizo14 participants about how and why they engaged Rhizo14. The raw material for the study, then, is a collection of discursive snapshots taken of Rhizo14 from the different angles and perspectives of each of the story tellers. At one scale, these snapshots can be aggregated and merged into a more complete image of whatever Rhizo14 is. This is something similar to what Blaise Agüera y Arcas does with Microsoft's Photosynth, which collects images, say of the Eiffel Tower, from around the Net, aggregates those images, and produces one multi-dimensional image that is far richer and more informative than any single image. Thus, the multiplicity of the network produces an image that exceeds the sum of all the parts.

But this resulting, single image is just one scale. Fortunately, the Rhizo14 auto-ethnography is conducted by a group of scholars, each of whom can explore the snapshots at different scales and from different points of view with different critical methods. For instance, Rhizo14 featured a wealth of creative responses (poems, animations, stories, etc), and I see no reason why the auto-ethnography should be limited to typical, scholarly papers. Some aspects of Rhizo14 are better and more accurately captured in poetry. Others of the group might tie the auto ethnographic stories to Facebook discussions or Twitter streams. Some might focus on a single participant’s story. Another might analyze the narrative structures of the various stories to explore how people construct their participation within a cMOOC. My point here is that there are more scales and points of view than there are researchers in Rhizo14, and we can preserve the multiplicity of Rhizo14 by looking at it from different scales and different angles.

I’m also pleased that Dillenbourg emphasizes the role of the observer in his analyses of collaborative learning. According to him, the observer defines both the scale of the network and the characteristics of the agents within the network. This positions the observer within the system observed instead of privileging the observer with some objective, godlike view outside the observed system. Thus, the Rhizo14 auto-ethnography group must recognize its position within Rhizo14 and account for its presence and interactions there as an integral part of the effects and processes that it is studying. Said another way, the auto-ethnography group must recognize its own considerable gravitational impact in the solar system that it is studying. While it is easy and natural to define the various Rhizo14 human participants as agents within the Rhizo14 system, the auto-ethnography group is still responsible for its characterizations of those agents and even more so for the configurations and groupings of those agents into Facebook, Twitter, and Google+ cohorts, bloggers, lurkers, the auto-ethnography group itself, and whatever other large-scale agents they may devise. The AE group must accept its creative, imaginative role in defining a particular Twitter stream or blog conversation, say, as an agent in Rhizo14, and though the AE group did not create Facebook, they do create Facebook as an agent in Rhizo14, in large part, by choosing to treat it as an agent, or not.

In sum, the study of a complex, online, collaborative learning system starts, as Dillenbourg suggests, by positioning the researcher within the system studied and forcing the researcher to account for her own role within and impact on the system, accepting that the system is different than it would have been had the researcher been someone else or had there been no researcher at all. This repositioning of the observer from outside the observed system to inside challenges the traditional notion about the stability of reality. If the characteristics and behaviors of the observed cMOOC depend at least in part upon the presence and imagination of the observers, then can we say anything stable and lasting about the cMOOC? Probably not. But stability is likely a problematic expectation anyway. The more promising aim of research, I think, is to say something useful about cMOOCs. We should aim to provide actionable knowledge, not absolute knowledge. The absolute, even if it exists, has always been beyond our abilities. Einstein tried to give us our last absolute when he said that nothing could exceed the speed of light, and then quantum physics came along with its spooky action at a distance to suggest that something is moving faster than light. Damn!

And even actionable knowledge may exceed necessity. Maybe what the AE group should aim for is an expansion of the field of education to include the possibility of courses such as cMOOCs. In her post Reflections on the value of MOOCs, Rhizo14 participant Tanya Lau says:

What I get out of the cMOOC experience is not necessarily practical strategies, ideas or actions that I can apply directly to my workplace (which I might get from say, an industry event, workshop, conference targeted to the field of corporate L&D that I work – or indeed, an xMOOC targeted to a domain of knowledge or skill I have a need to develop). Yet it’s something that actually has greater value than practical application: it’s the shift in mindset that results from engaging with people who are driven to continually question, experiment, explore and improve -> it’s that you start to adopt this mindset yourself too. Start to see challenges as opportunities to explore possibilities, become a little braver, make the leap from thinking about experimenting to actually doing it. No longer (as) afraid of being challenged, but open to it – inviting challenge rather than being defensive. It’s more than just being inspired. It’s inspiration + action to = change. Change in the way you think, learn and act – about life, work, learning, and yourself. It is the personal, human connections and inspiration that Clarissa speaks eloquently of in her posts on #CLMOOC and #Rhizo14.

It’s the type of engagement that most conventional courses and programs dream of achieving, and it’s the reason why I get so frustrated with the continual focus on ‘completion’ as a means to evaluate the effectiveness or value of MOOCs. It’s not about completion; it’s about engagement. And thought-provoking, behaviour-changing engagement can be triggered even through one conversation or experience – as long as it’s with the right people, at the right time, and at the right level.

This is brilliant, and I thank Tanya for saying it just so. You don’t finish a rhizome, you engage it, and when the rhizome has reterritorialized within you and you within it, then you are never free of it, even after its intensity has faded. You certainly can’t study it without becoming part of it. Like gardening, the dirt gets under your nails.

So writing about a cMOOC, defining it from within, aims not so much at giving teachers practical strategies and formulae for modifying their classroom practice as at helping them to shift their mindset and to engage “with people who are driven to continually question, experiment, explore and improve”. And if the AE group can identify a reusable practice or technique—well, that’s okay, too.
I see that I have pushed beyond what Dillenbourg was suggesting, so I have to give my usual caution when I’m discussing the work of another. I am a somewhat poor scholar in that I am much less interested in figuring out what someone else means than in figuring out what I mean. Take my reading of Dillenbourg, then, not as critique of Dillenbourg but as an exploration of how I think we might systematically map a complex, multi-scale system such as a cMOOC.

Okay, that’s Item 1 in my list about how to study a cMOOC. Items 2 & 3 soon.

The Hierarchy of Neural Complexity

Sporns says that "brain connectivity is organized on a hierarchy of scales from local circuits of neurons to modules of functional brain systems" (258). His use of the word hierarchy presents me with some problems as I have for the last few years contrasted hierarchical structures with network structures. In general, I have assumed that hierarchies were rigid, closed, traced, arboreal structures (to use terms from Deleuze and Guattari) while networks were flexible, scaleable, open, mapped, rhizomatic structures. Hierarchies admit only sanctioned, homogenous nodes within its structure and then fix them into a well-defined place with well-defined relationships to all other nodes; whereas,  networks admit heterogenous nodes within its open structure and allow nodes to develop relationships with any or all other nodes for any reason. For me, then, hierarchies and networks are not the same, and I cringe just a bit each time Sporns uses the term hierarchy to describe an aspect of neural networking.

But I think I have a resolution to my concern. I don't think Sporns is using hierarchy as I do; rather, he is describing various physical and functional layers of the brain and how they interact. As he says: "A recurrent theme in studies of collective behavior in complex networks, from epidemic to brain models, is its dependence on the network's multiscale architecture, its nested levels of clustered communities" (261). He is not so much describing a pyramidal structure as an onion structure. He's talking about layers enclosing layers enclosing layers and so on. This might be a mere quibble over visual metaphors but for his concepts of heterogenous coupling and multiscale dynamics, and some others like them. These principles prevent Sporns' neural hierarchies from calcifying into rigid hierarchies, as I have used the term. Indeed, heterogenous coupling in neurophysiology reminds me much of Deleuze and Guattari's first principle of the rhizome given in the first chapter of their book A Thousand Plateaus (1988): "any point of a rhizome can be connected to anything other, and must be. This is very different from the tree or root, which plots a point, fixes an order" (7). I think Deleuze and Guattari would be most comfortable with Sporns' concept of heterogenous coupling as very rhizomatic.

Sporns also talks about the brain's metastability, or tendency toward chaotic itinerancy:

the itinerant or roaming motion of the trajectory of a high-dimensional system among varieties of ordered states. Chaotic itineracy is found in a number of physical systems that are globally coupled, that are far from equilibrium, or that engage in turbulent flow. … It has also been observed in brain recordings … and neural network models … Over time, systems exhibiting chaotic itineracy alternate between ordered low-dimensional motion within a dynamically unstable "attractor ruin" and high-dimensional chaotic transitions. System variables are coherently coupled, their dynamics slow down during ordered motion, and they transiently lose coherence as the system trajectory rapidly moves between attractor ruins. (263, 264)

This chaotic itinerancy of neural networks with their roaming motions and trajectories and their constant transitions between coherent couplings and incoherent chaos is strongly reminiscent of another characteristic of rhizomes: the principle of asignifying rupture. Deleuze and Guattari say of asignifying ruptures:

Every rhizome contains lines of segmentarity according to which it is stratified, territorialized, organized, signified, attributed, etc., as well as lines of deterritorialization down which it constantly flees. There is a rupture in the rhizome whenever segmentary lines explode into a line of flight, but the line of flight is part of the rhizome.

Rhizomatic structures, then, deterritorialize and reterritorialize only to deterritorialize again. If I understand what Sporns is saying, then it seems that neural networks have a chaotic itinerancy that is at least a Rorschach of asignifying ruptures. Perhaps, Deleuze and Guattari's asignifying ruptures have deterritorialized and reterritorialized as chaotic itinerancy.

Seems possible. Anyway, chaotic itinerancy seems to fit nicely with D&G's whole idea about nomadology and motion in reality. Anyone know for sure?