The MOOC Misstep and the Open Education Infrastructure

The following is a pre-print of an essay set to appear in Bonk et al.’s forthcoming book MOOCs and Open Education around the World. It may undergo some additional editing before publication. Unlike the rest of the content on opencontent.org, this article is published under the Creative Commons Attribution ShareAlike license v4.0, as per my contract with Routledge. This essay remixes some material that was previously published on opencontent.org.

In this piece I briefly explore the damage done to the idea of “open” by MOOCs, advocate for a return to a strengthened idea of “open,” and describe an open education infrastructure on which the future of educational innovation depends.

MOOCs: One Step Forward, Two Steps Back for Open Education

MOOCs, as popularized by Udacity and Coursera, have done more harm to the cause of open education than anything else in the history of the movement. They have inflicted this harm by promoting and popularizing an abjectly impoverished understanding of the word “open.” To fully appreciate the damage they have imposed requires that I lightly sketch some historical context.

The openness of the Open University of the UK, first established in 1969 and admitting its first student in 1971, was an incredible innovation in its time. In this context, the adjective “open” described an enlightened policy of allowing essentially anyone to enroll in courses at the university – regardless of their prior academic achievement. For universities, which are typically characterized in metaphor as being comprised of towers, silos, and walled gardens, this opening of the gates to anyone and everyone represented an unprecedented leap forward in the history of higher education. For decades, “open” in the context of education primarily meant “open entry.”

Fast-forward 30 years. In 2001 MIT announced its OpenCourseWare initiative, providing additional meaning to the term “open” in the higher education context. MIT OCW would make the materials used in teaching its on campus courses available to the public, for free, under an “open license.” This open license provided individuals and organizations with a broad range of copyright-related permissions: anyone was free to make copies of the materials, make changes or improvements to the materials, and to redistribute them (in their original or modified forms) to others. All these permissions were granted without any payment or additional copyright clearance hurdles.

While there are dozens of universities around the world that have adopted an open entry policy, in the decade from 2001 – 2010 open education was dominated by individuals, organizations, and schools pursuing the idea of open in terms of open licensing. Hundreds of universities around the globe maintain opencourseware programs. The open access movement, which found its voice in the 2002 Budapest Open Access initiative, works to apply open licenses to scholarly articles and other research outputs. Core learning technology infrastructure, including Learning Management Systems, Financial Management Systems, and Student Information Systems are created and published under open licenses (e.g., Canvas, Moodle, Sakai, Kuali). Individuals have begun contributing significantly to the growing collection of openly licensed educational materials, like Sal Khan who founded the Khan Academy. Organizations like the William and Flora Hewlett Foundation are pouring hundreds of millions of dollars into supporting an idea of open education grounded in the idea of open licensing. In fact, the Hewlett Foundation’s definition of “open educational resources” is the most widely cited:

OER are teaching, learning, and research resources that reside in the public domain or have been released under an intellectual property license that permits their free use and re-purposing by others. Open educational resources include full courses, course materials, modules, textbooks, streaming videos, tests, software, and any other tools, materials, or techniques used to support access to knowledge (Hewlett, 2014).

According to Creative Commons (2014), there were over 400 million openly licensed creative works published online as of 2010, and many of these can be used in support of learning.

Why is the conceptualization of “open” as “open licensing” so interesting, so crucial, and such an advance over the simple notion of open entry? In describing the power of open source software enabled by open licensing, Eric Raymond (2000) wrote, “Any tool should be useful in the expected way, but a truly great tool lends itself to uses you never expected.” Those never expected uses are possible because of the broad, free permissions granted by open licensing. Adam Thierer (2014) has described a principle he calls “permissionless innovation.” I have summarized the idea by saying that “openness facilitates the unexpected” (Wiley, 2013). However you characterize it, the need to ask for permission and pay for permission makes experimentation more costly. Increasing the cost of experimentation guarantees that less experimentation will happen. Less experimentation means, by definition, less discovery and innovation.

Imagine you’re planning to experiment with a new educational model. Now imagine two ways this experiment could be conducted. In the first model, you pay exorbitant fees to temporarily license (never own) digital content from Pearson, and you pay equivalent fees to temporarily license (never own) Blackboard to host and deliver the content. In a second model, you utilize freely available open educational resources delivered from inside a free, open source learning management system. The first experiment cannot occur without raising venture capital or other significant funding. The second experiment can be run with almost no funding whatsoever. If we wish to democratize innovation, as von Hippel (2005) has described it, we would do well to support and protect our ability to engage in the second model of experimentation. Open licenses provide and protect exactly that sort of experimental space.

Which brings us back to MOOCs. The horrific corruption perpetrated by the Udacity, Coursera, and other copycat MOOCs is to pretend that the last forty years never happened. Their modus operandi has been to copy and paste the 1969 idea of open entry into online courses in 2014. The primary fallout of the brief, blindingly brilliant popularity of MOOCs was to persuade many people that, in the educational context, “open” means open entry to courses which are not only completely and fully copyrighted, but whose Terms of Use are more restrictive than that of the BBC or New York Times. For example:

You may not take any Online Course offered by Coursera or use any Statement of Accomplishment as part of any tuition-based or for-credit certification or program for any college, university, or other academic institution without the express written permission from Coursera. Such use of an Online Course or Statement of Accomplishment is a violation of these Terms of Use.

The idea that someone, somewhere believes that open education means “open entry to fully copyrighted courses with draconian terms of use” is beyond tragic. Consequently, after a decade of progress has been reversed by MOOCs, advocates of open education once again find ourselves fighting uphill to establish and advance the idea of “open.” The open we envision provides just as much access to educational opportunity as the 1960s vision championed by MOOCs, while simultaneously enabling a culture of democratized, permissionless innovation in education.

An “Open” Worth the Name

How, then, should we talk about “open?” What strengthened conception of open will promote both access and innovation? I believe we must ground our open thinking in the idea of open licenses. Specifically, we should advocate for open in the language of the 5Rs. “Open” should be used as an adjective to describe any copyrightable work that is licensed in a manner that provides users with free and perpetual permission to engage in the 5R activities:

1. Retain – the right to make, own, and control copies of the work (e.g., download, duplicate, store, and manage)
2. Reuse – the right to use the work in a wide range of ways (e.g., in a class, in a study group, on a website, in a video)
3. Revise – the right to adapt, adjust, modify, or alter the work itself (e.g., translate it into another language)
4. Remix – the right to combine the original or revised work with other open works to create something new (e.g., incorporate the work into a mashup)
5. Redistribute – the right to share copies of the original work, your revisions, or your remixes with others (e.g., give a copy of the work to a friend)

How would a change in the operational definition of “open” affect the large MOOC providers? If MOOC providers changed from “open means open entry” to “open means open licenses” what would the impact be? Specifically, if the videos, assessment, and other content in a Coursera or Udacity MOOC were openly licensed would it reduce the “massive” access that people around the world have to the courses? No. In fact, it would drastically expand the access enjoyed by people around the world, as learners everywhere would be free to download, translate, and redistribute the MOOC content. MOOCs could become part of the innovation conversation.

Despite an incredible lift-off thrust comprised of hype and investment, MOOCs have failed to achieve escape velocity. Weighed down by a strange 1960s-meets-the-internet philosophy, MOOCs have started to fall back to earth under the pull of registration requirements, start dates and end dates, fees charged for credentials, and draconian terms of use. It reminds me of the old joke, “What do you call a MOOC where you have to register, wait for the start date in order to begin, get locked out of the class after the end date, have no permission to copy or reuse the course materials, and have to pay to get a credential?” “An online class.”

Despite all the hyperbole, it has become clear that MOOCs are nothing more than traditional online courses enhanced by open entry, and not the innovation so many had hoped for. Worse than that, because of their retrograde approach to “open,” MOOCs are guaranteed to be left by the wayside as future educational innovation happens because it is simply too expensive to run a meaningful number of experiments in the MOOC context.

Where will the experiments that define the future of teaching and learning be conducted, then? Many of them will be conducted on top of what I call the open education infrastructure.

Content As Infrastructure

The Wikipedia entry on infrastructure (Wikipedia, 2014) begins:

Infrastructure refers to the basic physical and organizational structures needed for the operation of a society or enterprise, or the services and facilities necessary for an economy to function. It can be generally defined as the set of interconnected structural elements that provide a framework supporting an entire structure of development…

The term typically refers to the technical structures that support a society, such as roads, bridges, water supply, sewers, electrical grids, telecommunications, and so forth, and can be defined as “the physical components of interrelated systems providing commodities and services essential to enable, sustain, or enhance societal living conditions.” Viewed functionally, infrastructure facilitates the production of goods and services.

What would constitute an education infrastructure? I don’t mean a technological infrastructure, like Learning Management Systems. I mean to ask, what types of components are included in the set of interconnected structural elements that provide the framework supporting education?

I can’t imagine a way to conduct a program of education without all four of the following components: competencies or learning outcomes, educational resources that support the achievement of those outcomes, assessments by which learners can demonstrate their achievement of those outcomes, and credentials that certify their mastery of those outcomes to third parties. There may be more components to the core education infrastructure than these four, but I would argue that these four clearly qualify as interconnected structural elements that provide the framework underlying every program of formal education.

Not everyone has the time, resources, talent, or inclination to completely recreate competency maps, textbooks, assessments, and credentialing models for every course they teach. As in the discussion of permissionless, democratized innovation above, it simply makes things faster, easier, cheaper, and better for everyone when there is high quality, openly available infrastructure already deployed that we can remix and experiment upon.

Historically, we have only applied the principle of openness to one of the four components of the education infrastructure I listed above: educational resources, and I have been arguing that “content is infrastructure” (Wiley, 2005) for a decade now. More recently, Mozilla has created and shared an open credentialing infrastructure through their open badges work (Mozilla, 2014). But little has been done to promote the cause of openness in the areas of competencies and assessments.

Open Competencies

I think one of the primary reasons competency-based education (CBE) programs have been so slow to develop in the US – even after the Department of Education made its federal financial aid policies friendlier to CBE programs – is the terrific amount of work necessary to develop a solid set of competencies. Again, not everyone has the time or expertise to do this work. Because it’s so hard, many institutions with CBE programs treat their competencies like a secret family recipe, hoarding them away and keeping them fully copyrighted (apparently without experiencing any cognitive dissonance while they promote the use of OER among their students). This behavior has seriously stymied growth and innovation in CBE in my view.

If an institution would openly license a complete set of competencies, that would give other institutions a foundation on which to build new programs, models, and other experiments. The open competencies could be revised and remixed according to the needs of local programs, and they can be added to, or subtracted from, to meet those needs as well. This act of sharing would also give the institution of origin an opportunity to benefit from remixes, revisions, and new competencies added to their original set by others. Furthermore, openly licensing more sophisticated sets of competencies provides a public, transparent, and concrete foundation around which to marshal empirical evidence and build supported arguments about the scoping and sequencing of what students should learn.

Open competencies are the core of the open education infrastructure because they provide the context that imbues resources, assessments, and credentials with meaning – from the perspective of the instructional designer, teacher, or program planner. (They are imbued with meaning for students through these and additional means.) You don’t know if a given resource is the “right” resource to use, or if an assessment is giving students an opportunity to demonstrate the “right” kind of mastery, without the competency as a referent. (For example, an extremely high quality, high fidelity, interactive chemistry lab simulation is the “wrong” content if students are supposed to be learning world history.) Likewise, a credential is essentially meaningless if a third party like an employer cannot refer to the skill or set of skills its possession supposedly certifies.

Open Assessments

For years, creators of open educational resources have declined to share their assessments in order to “keep them secure” so that students won’t cheat on exams, quizzes, and homework. This security mindset has prevented sharing of assessments.

In CBE programs, students often demonstrate their mastery of competencies through “performance assessments.” Unlike some traditional multiple-choice assessments, performance assessments require students to demonstrate mastery by performing a skill or producing something. Consequently, performance assessments are very difficult to cheat on. For example, even if you find out a week ahead of time that the end of unit exam will require you to make 8 out of 10 free throws, there’s really no way to cheat on the assessment. Either you will master the skill and be able to demonstrate that mastery or you won’t.

Because performance assessments are so difficult to cheat on, keeping them secure can be less of a concern, making it possible for performance assessments to be openly licensed and publicly shared. Once they are openly licensed, these assessments can be retained, revised, remixed, reused, and redistributed.

Another way of alleviating concerns around the security of assessment items is to create openly licensed assessment banks that contain hundreds or thousands of assessments – so many assessments that cheating becomes more difficult and time consuming than simply learning.

The Open Education Infrastructure

An open education infrastructure, which can support extremely rapid, low cost experimentation and innovation, must be comprised of at least these four parts:

• Open Credentials
• Open Assessments
• Open Educational Resources
• Open Competencies

This interconnected set of components provides a foundation that will greatly decrease the time, cost, and complexity of the search for more effective models of education. (It will provide related benefits for informal learning, as well). From the bottom up, open competencies provide the overall blueprint and foundation, open educational resources provide a pathway to mastering the competencies, open assessments provide the opportunity to demonstrate mastery of the competencies, and open credentials which point to both the competency statements and results of performance assessments certify to third parties that learners have in fact mastered the competency in question.

When open licenses are applied up and down the entire stack – creating truly open credentials, open assessments, open educational resources, and open competencies, resulting in an open education infrastructure – each part of the stack can be altered, adapted, improved, customized, and otherwise made to fit local needs without the need to ask for permission or pay licensing fees. Local actors with local expertise are empowered to build on top of the infrastructure to solve local problems. Freely.

Creating an open education infrastructure unleashes the talent and passion of people who want to solve education problems but don’t have time to reinvent the wheel and rediscover fire in the process.

“Openness facilitates the unexpected.” We can’t possibly imagine all the incredible ways people and institutions will use the open education infrastructure to make incremental improvements or deploy novel innovations from out of left field. That’s exactly why we need to build it, and that’s why we need to commit to a strong conceptualization of open, grounded firmly in the 5R framework and open licenses.

References

Creative Commons. (2014). Metrics. https://wiki.creativecommons.org/Metrics

Hewlett. (2014). Open Educational Resources. http://www.hewlett.org/programs/education/open-educational-resources

Raymond, E. (2000). The Cathedral and the Bazaar. http://www.catb.org/esr/writings/cathedral-bazaar/cathedral-bazaar/ar01s08.html

Thierer, A. (2014). Permissionless Innovation.

http://mercatus.org/permissionless/permissionlessinnovation.html

Von Hippel, E. (2005). Democratizing Innovation. http://web.mit.edu/evhippel/www/democ1.htm

Wikipedia. (2014). Infrastructure. https://en.wikipedia.org/wiki/Infrastructure

Wiley, D. (2005). Content is Infrastructure. http://opencontent.org/blog/archives/215

Wiley, D. (2013). Where I’ve Been; Where I’m Going. http://opencontent.org/blog/archives/2723

Remembering Brent Lambert

I just learned that my colleague and friend Brent Lambert has passed on. As I’m reflecting on our relationship this morning, I want to share a few thoughts and feelings and stories.

I met Brent when he entered the PhD program at USU. He came to us with a Masters degree (in CS), but no Bachelors. That always cracked me up about him – he was quirky like that. He loved making software and was extremely firm in his commitment to openness (and the other principles that guided his life). He had a passion for building things that would make the world a better place. Here’s an old blog post where I include his thinking on learning objects in a brief review of related thinking by folks like Wayne Hodgins, Stephen Downes, and Andy Gibbons.

My relationship with Brent was the beginning of many good things in my life. We worked together on a number of crazy projects that were each ideas WAY ahead of their time. There wasn’t an insanity I could propose that Brent couldn’t code. In 2003 we launched our first collaboration, an open source system for post-publication peer reviewed journals we called Pitch.

That same year we began work on two bigger projects. Open Learning Support was open source, online discussion software designed to allow learning communities to self-form, self-organize, and self-manage. This software was integrated with MIT OCW and Connexions for a time, providing the places for people to ask and answer questions about what they were learning.

And then there was eduCommons. This was definitely the most impactful work we did together, and also started in 2003. eduCommons is an OpenCourseWare Management System – server software that colleges and universities use to run their OCW publishing initiatives. This work we timed just about right. eduCommons came into its own about the same time the OCW Consortium did, and since MIT never did open source the platform they use to manage their OCW initiative, it was the best game in town for a number of years if you wanted to run and manage your own OCW. (Brent would have persuaded you that, even if MIT had opened their platform, eduCommons was still better.) By our count in 2007 or 2008, a full 1/3 of all OCWs in the world were running on eduCommons. Many still do. It’s quite a legacy for Brent, because while many of us worked on eduCommons, it was really his baby.

Brent was a fellow traveler. I remember very early on – maybe in 2004 – we were tapped by the Hewlett Foundation to head down to Rice to do a technical review of an early version of the Connexions platform. As always, Brent stuffed everything he needed for the trip into a single grey backpack. We flew down the evening before the review, headed to the hotel room, and began messing around with the platform. After some initial frustrations and minor injury to our geek pride, we vowed we would not sleep until we figured out how to author a module and then fork it. It ended up taking about 90 minutes.

We went from being a pair of troublemakers, to being a small team, like with Pitch (which we worked on with Corrine Ellsworth), to what we called the OSLO Group (Open, Sustainable Learning Opportunities Group), to COSL (the Center for Open and Sustainable Learning). Brent was still leading the eduCommons team inside COSL when I left USU in 2008.

I broke a lot of new ground with Brent. I wrote my very first Python with him. I joined my first SourceForge project with him. I was always asking him one question or another about why my code wouldn’t work, and he would always patiently correct and teach me.

Brent and his family moved down to Utah Valley a few years back. We never really managed to get together with all the things going on in the lives of two families with lots of kids. I always had it in the back of my mind that we should get together “soon,” as I expect he did. Now it will be a while until I’ll see him again, which makes me sad. But it’s great to think that Brent is on the other side, working just as energetically there to make it a better place.

I shed a few happy tears remembering these good old days with Brent. I’m sure I’ll shed more next time I’m stuck writing some python, go to ask him about it, and remember that he’s temporarily out of reach. But I shed some bitter ones, too, thinking about Michelle and the kids who will miss him so much more than I will. My heart and my prayers go out to them, together with my reassurance that, despite it being a tragically, annoyingly long time, they will absolutely be together with him again one day.

Here’s to you, Brent. Thanks for everything. See you on the other side.

A Response to “OER Beyond Voluntarism”

Well, this has turned into a rather enjoyable conversation. To recap what has unfolded so far:

• It began with Jose Ferreira inviting me to appear on a panel at the Knewton Symposium,
• on the panel, I made the claim that in the near future 80 percent of general education courses would replace their commercial textbooks with OER,
• after the conference, Jose responded to my claim by telling publishers why I was wrong,
• I responded by explaining that the emergence of companies like Red Hat for OER would indeed make it happen, using the Learning Outcomes per Dollar metric as their principal tool of persuasion, and
• Michael Feldstein argued that it depends.

Yesterday, Brian Jacobs of panOpen published an essay contributing to the conversation. While I agree that some in the field have yet to pick up on a few of the points he makes, I’m a little perplexed that he would choose to position these points as a response to writing by Michael, Jose, and me. By making these points in a response, he implies that we have yet to understand them. Take this bit for example:

Their comments, though, didn’t tackle what I’ve come to see as the core issue for the OER movement, a foundational assumption that has crimped its progress. The assumption holds that because open-source educational content is like open-source software…its application and uses should follow in a similar way. The short history of the two movements makes clear that this is not the case.

I’ve been accused of many things in my life, but never of missing the difference between open content and open source. As the person who coined the term “open content” sixteen years ago specifically for the purpose of differentiating it from open source, I’ve never had to defend against this particular allegation. Not sure what to say.

Or this:

The OER movement’s almost singular focus on cost can obscure the larger objective — actually getting more students through to graduation while ensuring that they’ve learned (and enjoyed learning) something along the way.

when I spent almost half of the post he is responding to laying out the Learning Outcomes per Dollar metric for empirically measuring the impact of OER use on students’ academic performance. And then demonstrating with actual data from an OER adopter the incredibly powerful ways that OER adoption impacts learning.

Perhaps the article isn’t a response to Jose, Michael, and me at all. Maybe Brian is just using the conversation as an opportunity to underline a few unrelated points he feels need making, and that’s fine. And these little tidbits aren’t what I actually wanted to write about, anyway. Sorry. What I really want to do is unpack and comment on the core argument of the essay. First I’ll disagree, and then I’ll agree.

Disagreeing

As important as [the OpenStax] project is, it doesn’t yet realize the promise of OER as disaggregated high-quality content created and modified from anywhere.

Overworked and underpaid instructors are looking to content and course technology to make their lives easier, not to take on the additional responsibility of managing their own content without financial recognition for that labor.

From these and other portions of the article, I believe Brian’s argument is based on two premises:

• In order for students to get the full benefit of OER, their faculty need to be aggregating, revising, and remixing OER – really tailoring and customizing it to meet their specific needs
• This is a lot of additional work for faculty, and they won’t do it unless they are provided with additional incentives

Arguing from these assumptions, he arrives at the following conclusion:

This can be done by charging students nominally for the OER courses they take or as a modest institutional materials fee. When there are no longer meaningful costs associated with the underlying content, it becomes possible to compensate faculty for the extra work while radically reducing costs to students… a system for distributed content development also needs to be accompanied by a system of distributed financial incentives.

So, just stating each step of the argument explicitly to make sure I’m getting it right (hopefully he’ll correct me in the comments if I’m getting it wrong):

• if we charge students a little when faculty adopt OER,
• we can use a portion of that revenue to incentivize faculty to do the work of curating disaggregated OER and engaging in the revising and remixing process,
• (because if we don’t incentivize faculty by paying them, then most will never engage in these activities), and
• if faculty aren’t aggregating, revising, and remixing disaggregated OER, students won’t get the full benefit of OER.

I largely agree with Brian’s premises, but disagree somewhat with where he takes the argument based on them. (As I’ll argue below, this disagreement is both healthy and a Good Thing.) Here’s where I think the primary differences in our thinking lie.

The “Full” Benefit of OER

First, while I agree in theory that students don’t get the full potential benefit of OER if their faculty don’t engage in the aggregate, revise, and remix process, it’s unclear to me how much benefit students miss out on when faculty simply adopt OER “as is” (though we’re studying this question now). For example, the overwhelming majority of faculty in the college algebra example from my previous post – where passing rates increased from 48% to 60% after faculty switched to OER – did zero aggregating, revising, or remixing. Maybe the change in pass rates would have been even higher if they had, but are we really going to poo-poo an increase of 12 real percentage points in the pass rate? If students are getting much of the potential benefit even when faculty don’t aggregate, revise, and remix, is it worth incurring the additional costs necessary to achieve 100% of the full benefit? This brings us directly back to the Learning Outcomes per Dollar discussion in my previous post. What’s the delta in learning we would place in the numerator? What’s the delta in cost we would place in the denominator?

The INTRO Model

On Monday we’re submitting an article (for a special issue of EPAA) that introduces a new funding model we call the INTRO model – INcreased Tuition Revenue through OER. In this article we use actual enrollment, drop rate, tuition, policy, and other data from a large OER adopting institution to show that:

• when faculty adopt OER, drop rates decrease significantly
• when drop rates decrease, the institution refunds significantly less tuition
• when they refund less tuition, the institution has more funding to spend on things like supporting OER adoption among its faculty

In this particular example we demonstrate that, if the current OER pilot was expanded to all sections of the 20-some courses currently piloting OER, the institution could expect to retain over \$100,000 a year in tuition that they’re currently refunding. Some of this new funding could be used to pay for services supporting faculty adoption of OER without charging students.

I’m sure there are other models for funding OER adoption support services out there if we’re creative and open-minded enough to find them.

Parallel Experiments

And I am in total and complete agreement with this statement from Brian’s piece:

What’s needed are lots of entities — for-profit and nonprofit — to experiment with funding models.

YES! We need more experimentation happening, and we need it happening in parallel instead of serially. We can’t all stand around watching the Flat World Knowledge experiment, and only start trying something different when it becomes clear that their approach isn’t quite the right one. As Linus said, in what is possibly my favorite quote:

And don’t EVER make the mistake [of thinking] that you can design something better than what you get from ruthless massively parallel trial-and-error with a feedback cycle. That’s giving your intelligence _much_ too much credit.

Even though I disagree with some of Brian’s conclusions (which is why I’m experimenting with a different business model), I absolutely want him out there experimenting with his particular business model. If I’m sufficiently humble, I’ll learn a thing or two from him before it’s all said and done. (If he’s sufficiently humble, Brian may learn something from me, too.) From this learning a new generation of models will emerge and be tested. They will be followed by another, further refined set of models. That’s how the field moves forward in its understanding of how to support OER adoption at scale, and it’s how at least 80% of general education courses will end up adopting OER in place of commercial textbooks.