On Connectivism and Scale

Image: Sui Fai John Mak - https://suifaijohnmak.files.wordpress.com/2011/03/bubblus_pln_ple_rev1.png
Martin Weller makes some excellent points in his recent post on connectivism and scale, and they merit a short response.

Weller is writing in response to a specific statement I made in a recent post on another topic, which I'll quote here:
One of the major objectives of our original MOOCs was to enable MOOC participants to create interaction and facilitation for each other. This is because there is no system in the world where a 1:30 instructor:student ratio will scale to provide open and equitable access.
In my view, this model worked very well.
Weller's criticism follows two general lines - I may overstate them a bit for clarity, but the reasoning is fairly clear:
  • First, it has not been shown that a connectivism model of an online course will scale to massive numbers of students (like, say, 30,000 or more), and certainly not for inexperienced and lower-level students.
  • Second, it has not been shown that a connectivist approach is the only way a course can scale to such numbers while preserving openness and equity of access.
As Weller says,
I’m uncomfortable with this over-reaching of connectivism. Open universities across the world have been operating large scale, open, equitable learning for decades. As I’ve bored you all with on many occasions, I chaired a course with 12,000 online students (many more than the CCK courses). We operated a model with around 600 part-time tutors on a ratio of around 1:20. 
So let's address these arguments.

Here's what we know directly from the connectivist course experience. We know that it can scale to a number in the thousands - the largest course I know of, Change 11, had about 3,000 participants. It had three facilitators - myself, George Siemens, and Dave Cormier. So we achieved a 1:1000 ratio of facilitators to participants.

It is possible that this is the upper limit for a connectivist course - I have never been popular enough to try it at larger scale. :p On the other hand, it is also possible to offer the same course ten times to reach 30,000 people, still maintaining the 1:1000 ratio. In other words, the question of "will it scale" is not a question of absolute number per course. The question of scale is a measure of the cost of reaching such numbers of people.

A second aspect of the question of whether connectivism will scale addresses the quality of the learning experience. My assertion was, "In my view, this model worked very well." This is not exactly a data-driven empirically supported assessment (and in fairness to me, it wasn't framed as one).

But the point may certainly be made that the 1:1000 ratio may be achieved only in very special circumstances. Weller points to one: "Most of the successful learners in the cMOOCs were already experienced learners. But for a level 1 undergrad, open entry course, this is palpably not the case." And, "for many learners their issues are not with content (even when they appear to be), but with confidence, identity and other skills. A person experienced in the support role can identify this."

This is a fair criticism. It is certainly arguable (and I would agree with the assessment) that a certain degree of literacy is needed in order to be successful in a connectivist course - not merely the traditional literacies, which would be required to be successful in any course, but also matters related to confidence, identity and skills. This was something we tried to address directly with Critical Literacies course, where we asked, could a connectivist model be used to bootstrap those literacies. Our results (the course was offered by Rita Kop and myself) were definitely inconclusive, and Kop's publication on the course reflects that.

A second question that could be asked of a connectivist course is whether they can support the claim of having achieved certain educational outcomes. This is the sort of question that is directly relevant to something like a certification program. One aspect of this question concerns completion rates - are the people who started the course actually finishing the course. This is a question that has been studied a lot, and overall, completion rates in MOOCs are not great. Most such studies, however, have focused on non-connectivist MOOCs.

With respect to a connectivist MOOC, the issue is more complex. First, completion isn't an objective of a connectivist MOOC. In the past I've compared one to a newspaper - the idea isn't to read everything cover-to-cover, but rather, to pick and choose those aspects that are relevant. Second, there are no content-based learning objectives in a connectivist MOOC. Even were we discussing, say, open content licensing, there would not be a specific body of content against which we would test students for having learned.

Having said that, the question can be asked, can a cMOOC be successful where the objective is to achieve a certification or demonstrate mastery of a body of knowledge. I would argue (but not claim to have established empirically) that it can. The key here is, achieving the certification or mastery of a body of knowledge would need to be the student's objective. The question then resolves to: can a student, in a connective environment where these subjects are being discussed and where the relevant resources are being made freely available, achieve mastery?

I argue that such a student could, and more, that it happens all the time, though not exclusively in MOOCs, and (maybe) not solely by means of a MOOC. My assertion is based on the observation that a great deal of learning does take place in connective environments on the world wide web, that these have scaled to large numbers, and that often they do not require any institutional or instructional support. In this regard I cite websites like (the old and now-defunct) WebReference, GitHub, Stack Overflow, CodePen, etc. Now, has this argument been conclusively made? Probably not. But I believe that there is a strong prima facie case here.

Let's turn to the other side of the question: whether connectivism is the only way a course can scale to such numbers while preserving openness and equity of access.

A lot depends on what you mean by 'scale'. Because to a certain degree, anything can scale if you throw enough money at it. Personal one-on-one tutoring, arguably the gold standard for learning, would scale to reach everyone if we spent enough money on it. It certainly scales well enough to reach all of the elite, where it forms the basis of private primary education and for the Oxbridge experience. But of course the sheer cost of scaling 1:1 tutoring to everybody means that for any practical meaning of the term, 1:1 tutoring does not scale.

That's why I would argue that you can't say 'our courses scale to 12,000 online students' and 'support is not cheap' in the same argument. Sure. With enough resources, you can teach a lot of people. But the cost of teaching a lot of people puts an upper limit on what you can do. And while massive open universities like the OU and IGNOU and others certainly teach on a massive scale, they also hit hard limits, and are vulnerable in the face of those limits.

The cost imposed on students make this clear. The OU Fees and Funding page states, "Our fee for a 30-credit module is £1,506, and for a 60-credit module it's £3,012." And "More than 70% of students studying with us in England take out a tuition fee loan to fund their studies." Now I would be the first to agree that the Open University has perfomed a valuable service, that the MOOCs offered by the OU have been an improvement on that service, and that it is therefore more open and accessible than the traditional university (especially in the U.K.).

But it's still expensive, the facilitator:participant ratio is still low, and so there are limits. This is why I argue in favour of connectivist-style approaches.

So we turn, then, towhat I think is the main argument offered by Weller in this post:
If connectivism is to be as broadly applicable across domains and levels as its advocates seem to want, then support needs to arise from somewhere. Most of the successful learners in the cMOOCs were already experienced learners. But for a level 1 undergrad, open entry course, this is palpably not the case. It seems to me to underestimate the value of support to assume this could all be accommodated by other learners. The sort of support required for new, often unsure learners requires experience and expertise, and to suggest the network will accomplish this diminishes the value of the knowledge in my view.  

And I'll respond briefly as follows:

First, it seems to me astonishing that students who have already completed some 12 years of schooling - as is the case for most students entering university-level studies - would be considered "new, often unsure learners." In a way (and I know this is a bit unfair) it's like saying "our new students are illiterate, and that's why it costs so much to teach them."

Now in the case of actual illiteracy (or innumeracy, of other deficiencies) we do not assign the remediation to all students - rather, we set aside special classes that enable these students to get up to speed. Once they read at college-level, then they are ready to be successful at college-level classes.

Clearly, it would make a lot of sense for lower level schools to focus on critical literacies, and that's why I was (and still am) interested in the subject. But treating every student in a class as a novice learner even so seems like overkill. Why is it necessary? I submit that there are two reasons: first, because the student has to pay so much money, to simply fail students who are unprepared amounts to malpractice. And second, since the model is based on a uniform curriculum for all students, if you provide this support for one student in the class, you are essentially bound to provide it to all students in the class.

Second, "to require other learners to take on all of that burden (of offering support to other learners) is to potentially create a lot of hidden labour. Someone still pays, but we just don’t see it." In this sense, the criticism is a bit like criticisms of ride-sharing, where it's not the case that a person is getting a ride for free; someone is still paying for the gas. And (arguably) the ride isn't as good as the ride you would get from a proper taxi, and the very idea of ride-sharing as a means of addressing access to transportation devalues taxis. (Maybe overstated? We can quibble.)

And maybe this points to a fundamental difference between connectivist and other approaches. We don't want education to be a client-server industry. (By 'we' I mean 'me'). We are saying explicitly, "part of learning is teaching." As in, for example, "part of being a fully-trained physicist is working in the wider physicist community, both learning from others, and teaching others." The whole idea of a connectivist education is that it isn't supposed to be a dog-eat-dog world where those who can pay are able to obtain specialized tutoring services.

And if I have to be blunt about it: the cost of educational labour is what makes it so expensive, and what makes access to it overall inequitable and limited. The open universities are able to be more equitable and open by minimizing the cost of that labour - hiring inexpensive tutors, limiting contact hours, using mass-produced course materials rather than bespoke lectures. The wider university system is doing the same by employing underpaid sessional (adjunct) instructors. All levels draw on their governments for large sums of money. And most of them charge students significant tuitions.

The connectivist approach is to do whatever can be done to help students perform this labour themselves. This is made possible by creating and sustaining learning networks, where participants create and share learning resources and support with each other. It is probably true that they cannot perform all of this labour themselves. But it is arguable that they could do more of it - a lot more of it - than they do now. This is what I believe is shown by connectivism and the connectivist MOOCs.

I freely admit that connectivism does not prove absolutes. It isn't infinitely scalable. It may not be the only possible approach. There may be some types of learning it does not support. It might not even be appropriate for a majority of learners. These limits are all to be defined empirically, and it may be a long time before we know what these limits are.

But for me, the key thing about connectivism is the directionality. It defines success in terms of learners being able to manage and sustain their own learning.

Weller defends tuition saying that "at least that is an explicit, if you like, honest, charge." But it's not honest. Much of the reason it costs $500 is not that it costs $500 to provide that education, but that $500 is what people are willing to pay for what has been made an artificially scarce resource. The cost could be way below $500 if provision were made to enable the people who are able to learn on their own, to learn on their own, and focus whatever expenditures need to be made on those places, and those people, where this cannot be done.

We've done this in so many other fields that it seems incredible the point even needs to be argued for with respect to education. Transportation is accessible to everyone, but at the cost of requiring people to do their own driving, and providing public transportation for those who can't. Low-cost food is available to everyone, but at the cost of requiring that people prepare their own food, rather than providing ready-made meals to everyone, using restaurants and fast foods when needed for convenience. Global communication is available to everyone, but at the cost of having to buy and run your own computer, web server, or whatever. None of these is infinitely scalable, none of these provides perfect access, but they are all better than chauffeurs and taxis, personal chefs and ready-made means, or couriers and post offices.

That's the argument. Carefully stated with clear limits on what is claimed by connectivism, and what is not.



Popular Posts