The Case for Closed


David Wiley is the Chief Academic Officer of Lumen Learning and Education Fellow at Creative Commons. He has written extensively about the benefits of making instructional materials as inexpensive and as flexible as possible, and is a leading advocate for open educational resources (OER).

Dear David:

I heard a story on NPR not too long ago about how John Deere has locked down its tractor software, preventing farmers from making tweaks and repairs. I find John Deere green one of the most beautiful colors on the America landscape and am sure the company has its reasons. Still, my overwhelming reaction to the story was, “Screw those guys.”

I enjoyed your recent blog post, then, about the virtues of open resources in education. As important as copyrights and paywalls are for encouraging innovation, I share your concern about how they restrict access and stifle downstream creativity. If our goal as a community is for instructional materials to have maximal impact, it seems reasonable that we’d want them to be as accessible and as flexible as possible.

However, let me push back on that a bit.

Let Freedom Ring?

As a curriculum developer, I’m skeptical of the notion that content should be free. While many view content as just a “commodity” — a characterization that’s especially pervasive among technologists in Silicon Valley — my view is that platforms are only as valuable as the content that’s on them…which explains why Netflix’s stock price tanked when it couldn’t re-up its Starzz contract, why Steve Jobs cared so much about getting the Beatles on iTunes, and why LAUSD’s $1.3 billion iPad purchase was such a debacle: The tablets worked but the curriculum didn’t. Every time someone floats a freemium approach in which publishers make their resources free and charge for professional development, I can’t help but wonder whether they would have advised Beethoven to give away the Ninth in order to sell more piano lessons.

While I appreciate the upside of unfettered access, I think free educational resources are only virtuous if they’re also good; otherwise they can be terribly costly. The best educational content has tremendous value and typically requires a level of investment that defies its giving away. That said, I’m sensitive to the social loss that results when educators opt for low-quality free resources over high-quality paid ones, and I’m increasingly convinced of the value in making exemplar content free: not because I think free is inherently better but simply because I recognize the problem it solves in allowing educators to evaluate resources on quality alone.

When it comes to making exemplar content fully open, though, I’m not so sure. In fact, I fear that opening educational materials may in some cases make them worse. To illustrate why, let me share with you the thinking that went into XBOX Xponential, the Mathalicious lesson in which students use exponential growth to explore how video games have changed over time.

XBOX Xplained

In the mid-1960s, Intel co-founder Gordon Moore predicted that processor speeds would double every two years. In 1976, Atari released the first video game console with an internal microprocessor: the 1.2 MHz Atari 2600. So have subsequent consoles evolved the way Moore predicted they would? To answer this, we can (1) write an equation to predict how fast we’d expect speeds to be, then (2) research how fast they actually were. If the speeds don’t match up with Moore’s Law, we can (3) create a regression to compare actual speed growth to predicted growth and discuss what this implies for the future of video games. Mathematically, it’s a fairly straightforward process.

Pedagogically, though, it’s not. Creating an opportunity for students to successfully engage in this process requires expertise and careful planning.

As you can tell, XBOX Xponential is a narrative that builds over a series of questions, where each question is designed to serve a specific mathematical purpose. Once we feel confident that a lesson is ready to be published, should we provide a way for users to edit it? As a former teacher myself, my immediate instinct is to say yes. As a curriculum developer, though, I’d offer that the issue is less clearcut than it may seem.

Two Outcomes

There are two likely outcomes when someone edits a lesson: the lesson gets better or the lesson gets worse. In the case of XBOX, the most obvious way for it get to better would be for a teacher to remove questions. For instance, Q1 and Q2 act as guides towards the Moore’s Law equation, and editing them out would make the process more challenging for both students and teachers. If a teacher is comfortable providing the missing scaffold, this could result in a more meaningful learning and teaching experience.

Mathalicious lessons are already pretty lean, though, and it wouldn’t take too many deletions until the lesson became, “How have video games changed? Go.” This is effectively project-based learning, a level of inquiry-based instruction that may be unfamiliar to and difficult for many teachers. In our experience, most teachers appreciate some level of support from the resources they use, and I haven’t seen any instance in which a teacher pared down a lesson.

I have, on the other hand, seen plenty of instances where teachers built them up, and added so many questions that they inadvertently undermined the opportunity for students to engage in productive struggle.

Actual Hypothetical
  1. Based on Moore’s Law, how fast would you expect the processors to be in each of the consoles below?
  2. Write an expression to estimate how fast consoles should be in 2077, a century after the original Atari.
  1. How many years after the Atari was the Intellivision released?
  2. How many times must you multiply 1.2 MHz by to find the speed of the Intellivision?
  3. Calculate the expected speed of the Intellivision and write it in the table. Then repeat for the remaining consoles.
  4. When we multiply 1.2 by 2 over and over again, what type of growth is that?
  5. In your table, create two new rows: one for the number of years that have passed since 1977 (aka “video game years”), and another for the number of times the 1.2 MHz should have doubled.
  6. Write an expression for the relationship between “video games years” and the number of doublings.
  7. Write an expression to estimate how fast consoles should be in 2077, a century after the original Atari.

We often hear that Mathalicious lessons are “hard for students.” When students struggle, it can be tempting to jump in and provide a roadmap to the answer, or to write lessons in a way that preempts struggle altogether. Yet as helpful as this may feel in the short-term, it’s counter-productive in the long. As Carol Dweck, Jo Boaler, and others have pointed out, students need an opportunity to wrestle with mathematics. While making lessons editable would make them more customizable, it might also make them less valuable to students.

It might make them less valuable to teachers, too. Teaching an exemplary lesson involves a range of skills, from recognizing misconceptions to facilitating discourse. A well designed lesson supports teachers in realizing best practices and is one of the most powerful tools for professional development. The more scaffolded the lesson becomes, the less it serves this purpose.

Areas of Expertise

Of course, I understand how condescending it sounds to suggest that a teacher might suffocate a lesson. The thing is, it’s exactly what I did when I was in the classroom! Not only that, it’s what I did when I first started Mathalicious. In my first year, many of the lessons I wrote were over 50 questions long — anything I could include, I did — and it was only when I saw Dan Meyer’s TED talk that I began to appreciate the philosophy of “be less helpful.” Even still, it took me another few years to hit my stride. In the end, I needed three years of teaching, one year as a math coach, two years of grad school, and three years of collaborating with my colleagues at Mathalicious before I felt like I was good at writing lessons.

If we assume that teachers are normally distributed when it comes to instructional quality, and if we agree that high-quality lessons are an important tool for informing instruction, then the question of editability may come down to this: When it comes to exemplary resources, do we want the resource to move towards the teacher, or do we want the teacher to move towards the resource? Put another way, do we want the previous version of me to edit a lesson that the current version of me helped write?

The discussion around open educational resources tends to focus on maximizing the upside. I think it’s important to consider minimizing the downside, as well. I’d love for teachers to be able to tweak lessons for the better. However, making lessons editable makes it possible for lessons to get worse, and the expected net effect of this is strictly a function of where the resource is relative to the curve.

This isn’t to say that resources shouldn’t be editable. It’s simply to say that making them so comes with a risk. With many resources — perhaps even with most resources — the risk is fairly minimal; a word problem that uses baseball to teach fractions can probably be rewritten to be about basketball without sacrificing much, as can the thousands of one-off lessons on aggregator sites like OER Commons and Teachers Pay Teachers. With resources like Mathalicious, the risk of editability is greater; each lesson relies on a tight narrative and a pedagogical coherence which may not be obvious at first glance (indeed, which may not even be evident until after the lesson is taught). The risk associated with editability is greatest when it comes to core resources like Connected Math and Discovering Algebra; not only do these depend on coherence at the lesson level, but they also depend on coherence at the unit and ultimately curriculum level. Once you start pulling on this thread, you risk unraveling what may have been a very intentional and thoughtfully crafted educational experience, one refined over many years by a team with significant expertise.

And it’s this word — expertise — which I think makes the question about openness such an interesting one…and also such a difficult one. Teachers are experts on what’s happening in their classrooms, and it’s important that they be able to modify instructional resources to meet their students’ (and their own) needs. At the same time, effective curriculum developers are experts on how concepts develop and how to balance scaffolding with inquiry, and it’s critical for the profession that this expertise be preserved.

Harmory of the Spheres

I had lunch recently with Jere Confrey, a math education professor and one of the most thoughtful people I know. She and I were discussing ways to improve XBOX Xponential. One of the most challenging moments comes in Q5, when students compare how processor speeds were predicted to grow (Moore’s Law) with how they actually grew (regression). Because the equations are written in different terms — every two years versus every year — it’s an apples-to-oranges comparison. Jere suggested that we could resolve this by priming students to think of Moore’s growth on an annual basis earlier in the lesson. Q2 currently asks students to write an expression for the speed after 100 years (1.2 x 2100/2). If the question also asked about the speed after one year (1.2 x 21/2), then students might recognize the annual growth rate of root-2 more quickly.

If we implemented this change, students would have multiple ways to write the Moore’s Law equation in Q3. That would be good. On the other hand, a discussion of fractional exponents so early in the lesson might obscure the arguably more important insight into the relationship between years passed and number of doublings. That would be bad. While asking about the annual growth might help students understand Moore’s Law more deeply (good), since the prediction is specifically described in terms of “every two years,” students might wonder where the question was coming from (bad).

This conversation with Jere was admittedly in-the-weeds, and I include it simply to highlight two points. First, while every lesson can be changed, sometimes even the most thoughtful changes are lateral ones where the result is just a different version of very good. In cases like this, I’m not sure how much the educational community gains from making resources editable.

Second, as esoteric as the conversation was, it’s exactly the kind that thoughtful curriculum developers have. Not everyone is interested in the issues that Jere and I were debating…and not everybody needs to be. And this is the point which I think too often gets lost in conversations about open resources: Writing a lesson and teaching a lesson are very different roles.

In many ways, the relationship between a teacher and a curriculum developer is similar to the relationship between a conductor and a composer. Beethoven may have written the Ninth, but Gustavo Dudamel (Los Angeles), Alan Gilbert (New York), and Marine Alsop (Baltimore) will all approach it differently. Similarly, Anne Wicks (Chicago), Ed Campos (Visalia), and Brooke Powers (Lexington) will all teach a math lesson differently; they all have their own style of questioning, their own approach to classroom management, and their own ideas about which conversational threads to pursue. While OER advocates rightly note that teachers are likely to feel more ownership over lessons they customize, they often overlook the fact that even when a lesson isn’t personalized, the experience of teaching it is still intensely personal.


Thanks to the ubiquity of new technologies, we live in an era in which traditional power structures are being eroded. Blogs and Twitter are democratizing journalism. iBooks Author and Lulu are democratizing publishing. In education, services such as GeoGebra and Curriki are empowering alternatives to the hegemony of Pearson, McGraw-Hill, and Houghton Mifflin Harcourt. This is good. The democratization of creation allows for a more vibrant community, and I can appreciate why you advocate for open educational resources that facilitate the “five Rs:” retain, reuse, redistribute, remix, and revise.

Assuming we can figure out economic models that incentivize and sustain expert-level development, I agree that maximizing access is an unequivocal good; retention and reuse make resources more dependable, and redistribution makes teaching more collaborative. Yet while the first three Rs address how content can be used, remixing and revision address what the content actually is. Here, I’m not convinced that complete openness is as incontrovertibly advantageous as many seem to think.

The Department of Education recently hired its first-ever open education advisor, and is even considering a new requirement that any materials created with public funds be free and fully open. While part of me applauds this, another part wonders whether it’s rooted in a deep understanding of the relationship between teachers and curriculum, or whether it’s simply a reaction to the larger social phenomenon of Etsy and Pinterest: a new zeitgeist in which the line is blurred between user and creator and where expertise is but an anachronism to overcome. As much as I agree with the spirit of openness, I fear the unintended consequences of treating it as a goal in and of itself: Editability may result in better outcomes with most resources, but is likely to result in worse outcomes with the best resources. Of course, if someone wants to cut up an exemplar lesson and insert dozens of additional questions, nobody can prevent that. But the issue of openness isn’t about whether we want to make that possible. It’s about whether we want to make it easy.

For me, that’s a difficult question, and one I’ve been wrestling with for years. It’s also one that’s especially germane right now. We’ve recently begun to work on the next generation of Mathalicious, and are exploring ways to make the experience even better for teachers and students. While our priority is to continue supporting good instruction, it’s possible that we could develop the new lessons in a way that makes them easier to change than they are now.

If we can, though, do you think we should?

March 24, 2016