Hapgood

I gave a presentation at InstructureCon last week on the distributed flip, and in particular, how it related to MOOCs. I thought the session went extremely well, and when the video is available a couple weeks from now I’ll post it. But I wanted to get the main argument down here. I flew through it a bit faster in the presentation (done in a record 13 minutes), but the thrust of the argument is the same.

One note here — I use the term MOOC here to denote xMOOCs. I used to always rigorously distinguish xMOOCs and cMOOCs, but I’ve found general audiences know MOOCs as CourdacityX stuff, and sometimes you just have to refer to that phenomenon under the terms they know. When I’ve used xMOOC, people have gotten far too hung up on the term. 

The distributed flip is a way of approaching flipped classroom design. It’s worth noting that I’m using a fairly broad definition of flipped classroom here. This is not the simple “homework in reverse” model. Rather, it is the idea that *some* amount of “low-level” activities done inside class (lecturing, quizzing, etc) can be moved outside of class using technology, and the reclaimed time can be used for high-impact activities – project-based learning, peer instruction, guided inquiry, etc.

I think it’s also important to note that I don’t think the flipped classroom is particularly new. Instead, it represents a way of teaching that has existed for centuries in many disciplines. Maybe longer. Effective teachers have always realized that face-to-face time was a precious resource that is usually most effectively used on integrative tasks, not on lower-level skill development. The student of Spanish memorizes words at home, not in the classroom, and uses classroom time to deal with more difficult issues of fluency and conversational application. In the humanities, students are expected to come into class with a basic comprehension of a text, one which the professor will push on through Socratic questioning. The sciences, a bit of a latecomer here, now push teachers, via schemes like Mazur’s Peer Instruction, to use classroom time to address conceptual hurdles identified in student pre-class work.

So it’s not really new. I don’t care, because the research shows that the principles the term stands for work, which is what actually matters. I think an argument can be made that technology allows us to move more activities out of the classroom than was possible previously, but even if that wasn’t true, I’d still push for the flip, because it works.

So how best to think about flipped classroom design? I find that Robert Talbert’s idea of a “cognitive cutoff” is a helpful way to think about prepping a flipped classroom. According to Talbert, most understandings worth attaining involve some level of hierarchal learning goals. Here, in this slide, Talbert points out the target outcome is fluid real-world application of the Chain Rule in Calculus. If Talbert wanted to practice that high-level skill in class, then students are going to have to come into class with a number of outcomes already achieved. They will have to know what the Chain Rule is, be able to identify composite functions, and do some basic application and analysis. It’s the price of admission to the high-impact stuff.

That line that separates the objectives that students need to come into class having accomplished from the objectives that they will practice in class is what Talbert calls the “Cognitive Cutoff”.

What I find interesting about the cognitive cutoff from an educational materials perspective is this. If your cognitive cutoff is low – somewhere just over remembering – your educational materials issue is pretty simple, and you might be able to deal with it by recording some mini-lectures and combining them with textbook readings. In this sort of scenario it’s completely possible that a faculty member could do this by themselves, though admittedly at an increased workload.

But what if you want to push that cognitive cutoff to where Talbert is pushing it? And what if you want to develop a rich set of activities and materials that support the students in reaching that cutoff before class?

I’d propose that to do that in a really effective way requires more work than any one single faculty member can do, and that when you get into designing online components that test application, understanding, and analysis that you may also be pulling from skills that are not traditionally those of faculty (e.g. instructional design). You have to get beyond the cottage industry model of course design.

So enter the Distributed Flip. The idea here is that some amount of design of flip materials is done centrally by a group of people, either as a company, consortium, or loose network of individuals. Those high quality materials are then distributed among many instructors who work to localize and modify them for use in their own flipped classrooms.

If this doesn’t sound revolutionary, it’s because it’s a trend that has been around for quite some time now. If you think about that “cognitive cutoff”, you’ll realize that digital materials have been attempting to cut into those higher order objectives for a while. So, for example, many publisher resources used to concern themselves mainly with introducing students to the material. But over the past decade or so publishers have developed online resources that claim to push that cutoff point higher, to get the students up to higher level objectives before they come to your class. Pearson, in fact, no longer calls itself a textbook company — it’s a “digital learning and services company.” In the non-profit arena, Carnegie Mellon’s OLI announced its initiative to build shared materials for blended community college courses in 2009, a process similar to that which NCAT, the National Center for Academic Transformation, has been doing since 1999 (long before the term “flipped classroom” was invented).

Niche efforts also abound. As just one small example, the American Association of State Colleges and Universities wants universities to experiment more with a Global Challenges-based curriculum. But there aren’t really good texts to support this. So they got together 10 universities, pooled their teaching materials, and wrote their own set of digital resources. Those materials are hosted on an LMS run by the New York Times called Epsilen, which hosts both the the teaching resources and student online activity around those resources. This was put together (the planning stages at least) before the xMOOC craze.

A broader and more recent effort (and one which I am particularly excited about) is the Kaleidoscope Project, where multiple institutions collaborate to build “Open Course Frameworks” which can be used by any institution. In efforts like Kaleidoscope and Global Challenges, the teaching community around the resources can as valuable as the resources themselves.

Which brings me to an important point: I’m skeptical about the hype around educational technology. But I’m very bullish on the possible impact that well-crafted materials, cross-institutional collaboration, and robust communities of practice could have on education. When it comes to educational materials, the move away from single author resources towards resources informed by our collective educational experience is a good thing, increasing the impact of the best ideas we have on how to educate and helping us move past the “it worked for me as a student” school of educational design.

And if it moves us also to talk more about what works in our classrooms and what doesn’t; if it starts connecting faculty, and making practice visible — so much the better.

So what about MOOCs? How do they fit into this world of the distributed flip?

Like many people, when the first Coursera and Udacity MOOCs were published, my first thought was whether these could be used as distributed flip materials. When you are looking for materials to help people flip classes, you have the sort of single-mindedness that a zombie has looking for brains. And MOOCs looked like brains to me.

Of course, it shortly turned out that MOOCs weren’t really open, and that Coursera, edX, and Udacity all explicitly prohibited reuse of their materials without permission. And permission involved being in a pilot, paying some money, or doing other things we can’t know about because of non-disclosure agreements. (Even more openness, apparently…).

But say you want to pony up the money and use these materials – do MOOCs make good distributed flip materials?

There’s two answers to that. One is simple and uninteresting, and one is complex and fascinating.

Here’s the simple answer: of course they could make good materials. If they are good videos, good online activities, good forum discussion questions – well, then – yeah, sure. Good integrated materials are what you want, right?

But that’s kind of a dumb answer. We’ve really said nothing there about MOOCs per se. We’ve said good materials make good materials.

Here’s the more complex question: Do MOOCs offer anything substantial to the distributed flips above and beyond materials? The answer to that is going to depend on affordances of the global cohort associated with MOOCs. Because the thing which separates MOOCs from other digital resources efforts is the “massive” cohort bit – you’ve got a large, synchronized group of students going through the lessons at a group at the same time. Remove the cohort, and it’s just digital resources.

So I was talking with Amy Collier and Helen Chen at Stanford about this question (Amy, by the way, is the person that coined the term “distributed flip”) and we decided to do something novel. We decided to talk to a variety of people that were doing distributed flips with MOOCs, and find out what their experience was. Our thought at the time was that the relationship of the large cohort of the MOOC and the local cohort of the class might be really productive.

But that’s not what we found. What we found when we talked to people actually doing this was that in most cases the large and small cohort weren’t even really sync’ed up.

This is kind of an odd thing, but it makes sense when you think it through. The obvious reasons are that differences in schedule matter. Start dates and end dates are different. You have semester vs. quarter issues. You might have different breaks, different mid-term reporting requirements, snow days. Maybe throwing the students straight into the MOOC, before getting project teams set up is not the best thing, so you want to introduce that on Week Three. The project you are doing might require students get through the material faster. You might be on a four-credit system. Maybe you’re working around guest speakers.

That’s the obvious thing. The less obvious thing was this – professors, quite rightly, can’t be designing activities on a week-to-week basis as the materials come out. We had one professor who initially tried to do that, until she fell behind a week because her students needed more time on a particular topic (yet another reason why sync’ing is hard). She expressed that falling behind was a blessing in disguise, because now she had more time to prep her class activities that would be based on the MOOC.

Professors in a distributed flip are responsible for the most important part of the experience – the higher order activities that occur in real time. In order to structure those activities properly, they need to be able to look at the digital materials as a whole while designing the class. They need to be able to make timeline adjustments around their in-class projects and activities. Running classes in sync with the global cohort reduces their ability to do that, which is why it is unsurprising that so many professors choose to use the MOOCs primarily as digital materials, without paying too much attention to the central MOOC’s schedule.

In fact, if you look at all these stories in the press about this or that MOOC being used to flip a class at Random State University you’ll find that Random State University isn’t sync’ing up either. I’ve looked at a lot of stories around this trend, and I’ve yet to find one where syncing with the global cohort was a priority. In fact, many of the offerings are impossible to sync up with by design. That edX Michael Sandel course that San Jose State was arguing about? It runs for 20 weeks, which far exceeds any known semester length. The Circuits and Electronics course we heard about as the gold standard of “MOOC impact” starts mid-March and runs for 14 weeks, which accommodates neither a traditional quarter or semester schedule. This is why, if you read carefully in these articles, you’ll find that San Jose didn’t use a MOOC, they “used material” from a MOOC.

In other words, there’s no MOOC in these MOOC applications…just a collection of digital resources. The “Massive” part of the course is not even used.

The other thing we found, or at least heard from the professors, is that the students didn’t really use the social features of the MOOC much. This isn’t surprising – if you are not in sync with the global cohort, it significantly reduces the sorts of conversations you might have. But even when we looked at the backend data for the one course that was (nearly) in sync we found that the Median Student visited the forums twice over the entire course period.

Twice. That’s two visits, not two posts. For the vast majority of students, that’s just going to the forums and looking. And one of those times was probably in the initial course kickoff to see what the forums were, or say hello.

It might be tempting to see this as a failure of either the MOOC design or of the local teaching. But digging down into what the students were doing we came to see this disengagement as a feature, not a bug. These students in flips were working on project teams, or participating in local learning communities. In such systems you want the students to rely on their local teammates for help and answers. In fact, the extent to which those students bond to their local classmates is one of the highest predictors of college success.

So the students *were* conversing. They were just doing it in the local instance. And that’s good. That’s what we want.

So back to our original (complex) question. Are MOOCs, in practice, adding much to the distributed flip toolbox? My sense (and not necessarily the sense of people I worked with on this) is that the features that make a MOOC a MOOC don’t really add much to the blended classroom.

So what, you might say. There are some features you don’t tend to use in blended scenarios. It might be tempting to see this the way you see cable TV – I have 100 channels, I only watch five, but it’s not like those channels are hurting anything.

Except that’s not really the case here. In this case the large sync’ed cohort you aren’t using still has an impact on the product. Because that sync’ed cohort requires a centralized version of the course. And a lot of the issues we heard from distributed flip educators – inability to localize, lack of direct access to data – are products of that centralization. If we were getting lots of benefits from centralization, we’d lump this stuff. But it seems like we aren’t, or at least weren’t in these specific instances.

Now it looks like, in the recent news stories about Coursera and its deal with a number of state systems, that Coursera is going to decentralize these offerings a bit. It’s still unclear what the structure of these recent initiatives is, but I’m guessing it is going to trend toward local control and instances if they are going to become a vendor of services to educational institutions. I imagine that edX and Udacity will follow suit – the future of the distributed flip just involves less centralization than the current MOOC structure entails. That’s not to say that either centralization of local control *wins* — just that we’re going to find a better balance.

And the story about MOOCs is not all bad, either. One thing we found is that the professors really appreciated the chance to be a student in the MOOC, to see how the course was put together. You can imagine, in fact, a sort of “milkweed” model here, where many faculty take an initial centralized MOOC as students (along with self-learners), and then disperse to the corners of the earth to run their own local version of that course using the materials. You’d have to have the MOOC providers adjust their copyright policies, but such an experience could be transformative for faculty, building on the success of modelling as an educational transformation strategy, and potentially jump-starting robust communities of practice.

So that’s the argument. We can learn a lot from the excitement that MOOCs have generated among instructors looking to flip their classroom. The distributed flip is not going away, and something in MOOC-space is definitely going right here. But to fully realize the potential of MOOCs we have to see the tensions inherent in MOOC design, and understand how some of the cohort-based affordances of the MOOC are bought at the expense of the local experience. A marriage of traditional OER concerns with some of the visible practice inherent in the MOOC could have, in my opinion, significant impact of the quality of teaching (and the cost of quality teaching), and such a marriage is possible if we can open up an honest conversation about what real-world reuse looks like, and what could make it better.

Thanks for reading/listening. Now go out and make the world more awesome.