Open Content, Closed Everything Else: Why K-12 OER Is Stalling

 OpenEd2026 kicked off earlier this week with a webinar entitled, Part 3. Working Together Across the OER Ecosystem. (I don't know what parts 1 and 2 were.) The subtitle was Finding Common Ground in K-12 and Higher Education. The suggested areas of common ground were Career Pathways/CTE, Dual Enrollment/Dual Credit, and AI Guidance / Implementation. The first two are really just Higher Ed issues that impact a small but important segment of K-12. I did a presentation five years ago at OpenEd21 with Lisa Lucas Hurst about using OER for Dual Enrollment/Dual Credit. It’s a great idea, but I’m not convinced Higher Ed is really interested in doing what needs to be done to make it work. Nobody had anything to say about AI Implementation during the webinar, so.... I have some ideas about how to do that, but it’s going to take a willingness to go offline, and there are few takers in the U.S. so far.

 I’ve been an advocate of OER for over fifteen years. OER arrived with the promise that high-quality, standards-aligned textbooks, lesson plans, and materials would be free, open, and legally adaptable for every classroom.  Some OER materials are now available, but wide use of OER is stalled because there is no unified, scalable, openly licensed way to assess student learning. And, assessing student learning is what K-12 is all about, but to give a test or even a formative assessment, a teacher moves open materials into Canvas, Schoology, a proprietary assessment system, or Google Classroom, and it stops being open.

As I argued in my post on Delmar Larsen's "National Priority for Open Education", open content hasn't displaced the infrastructure underneath it. The textbook is free; the system that enables the learning isn't. In K-12, Google controls most of that system, and Google is "kinda free," but definitely not open. Security makes things worse. When materials are public, a published answer key is a five-second web search away, so creators lock high-stakes assessments behind teacher-only firewalls. The testing layer of an open curriculum ends up closed.

Iowa is an example of how this goes wrong. The state spent $17 million in federal funds on full-year, LMS-delivered OER courses, but the openly licensed courses were never actually shared openly, and no professional development was funded to help teachers use them. Good content is not enough. As I wrote in Open Infrastructure has to Include People, training and support are part of the infrastructure, and the know-how has to stay with the institutions that depend on it rather than leave with a vendor contract.

Open, performance-based assessment (PBA) is the direction I favor. Students defend portfolios, conduct scientific inquiry, or tackle real-world modeling tasks scored with Creative Commons rubrics. But the logistics are brutal. An LMS grades a multiple-choice quiz in a millisecond; scoring thirty portfolios with a multi-trait rubric can take a teacher ten to fifteen hours. In an era of historic teacher burnout, that is a non-starter without real support. And for teachers in different schools to score the same project the same way, districts must invest in ongoing professional development and calibration, which most public systems lack the budget to do. PBA works only if capacity-building is funded alongside it. Using an open-source system like the one Andrew Lyman Butler explains here to do assessment is another good option, but that too will require investment in teacher professional development.

U.S. institutional architecture works against open assessment, too. States are pushed toward standardized data, and portfolios don't aggregate into statewide spreadsheets. So states stay tied to centralized multiple-choice exams. The providers of those exams are dominated by venture capital groups which have a strong financial incentive to keep assessment proprietary, centralized, and machine-gradable, and the lobbying power to keep teacher-led or open frameworks out of state adoption. Only a handful of U.S. public schools systematically use a truly open, Creative Commons-licensed mastery assessment system. Most schools still aren’t even using OER textbooks, and almost all use premium, proprietary, closed testing.

Any real fix has to pass the design test I proposed in this post: can it work in a classroom with no internet? In U.S. K-12, that isn’t a hypothetical. There are still too many classrooms and homes where connectivity is slow, metered, shared, or missing. Meanwhile, more states are limiting screen time, which makes always-on, cloud-dependent platforms a poor bet for the long term. Here is what passing the test requires.

A school or district should be able to run open assessment on hardware it controls. Students and teachers connect to it over the school’s own wireless network, with no internet needed. Small offline Moodle servers already do this in classrooms in northern Ghana, where the Sopala Project has paired them with local-language content. Nothing about the approach is specific to Ghana. A U.S. district can run the same kind of server. It also keeps student data inside the building instead of in a vendor’s cloud.

 Offline can’t mean a ‘less than’ system. Randomized practice, auto-graded quizzes, and teacher gradebooks should work without a connection, and results should sync when a connection is available, whether that’s overnight, at a library, or on a hotspot. The same logic extends to the home. Parents should be able to load content onto their own phones when they’re at school. That gives parents a way to see what their children are learning and to help with homework, even when the home has no broadband. A homework gap that depends on home internet is a design flaw. Elementary teachers, who work most directly with families where English is not the first language, would benefit first.

The content and the assessment infrastructure built directly on it should carry a license that keeps them free to adapt and redistribute but blocks commercial capture, such as CC BY-NC. Public schools can then deploy, translate, and modify the system without worrying that a company will repackage their work behind a subscription. Vendors can still sell support, training, and compatible tools to those who want them. They just can’t turn the core into a toll road.

Every district needs trained, paid facilitators, such as a teacher-leader or technology coordinator with time and a stipend, who can set up the local server, help colleagues build and adapt materials, and work with families who are new to the system. This role can be what media specialists used to do before IT took over materials. This is infrastructure, not a nice-to-have. State education agencies, districts, and federal programs already fund professional development, and some of that money should be directed here. Without paid facilitators, an open system only serves the schools that already have technical expertise, which compounds existing inequities.

Each trained teacher should add to what came before. When a fourth-grade teacher in one district translates a unit into Spanish, fixes a confusing question, or adds a better set of practice problems, that improvement should flow back to a shared, openly licensed repository that every other district can pull from, including districts that sync only occasionally. Facilitators are the natural link in this chain. They gather local improvements, check them, and share them, so the system gets better with every teacher who uses it instead of resetting each time a vendor changes its product.

Until  we have a secure, scalable way to measure mastery without surrendering to proprietary vendors, it will remain an incomplete experiment, a well-intentioned supplement rather than a systemic replacement. Free textbooks and lesson plans are great, but the part of K-12 that carries consequences are the quizzes, the unit tests, and the evidence that a student has learned something. “Secure” means student data stays under the control of the school and family. “Scalable” means the system works for a rural district with one part-time technology coordinator as well as for a large urban one, and keeps working on the school’s own hardware when the connection drops.

And don’t look to higher ed for help. Homework and testing platforms from Pearson, Cengage, McGraw Hill, Lumen, and many others are woven into course after course. Even where the textbook is free, open adoptions often depend on them for grading. Higher ed has learned to live with that dependency in order to ameliorate the brutally regressive practice of requiring students to pay for textbooks. Higher Ed starts with the best-connected learners, not with the K-12 classrooms where the hard problems are. K-12 has young children, shared devices, uneven home access, thin technical staffing, and strict privacy obligations. Solve for where it’s hardest and let Higher Ed adopt what works.

Open Infrastructure has to Include People:

 Delmar Larsen followed up on his argument for a national courseware ecosystem. I responded to Delmar's proposal in my last post. In his follow-up, Delmar says that FlatWorld, a commercial publisher, understands the publishing game because its platform connects textbooks to a homework system and to the instructor's actual workflow. His point is that the textbook is only one part of the ecosystem, that if open textbooks leave homework, assessment, analytics, adaptive learning, instructor tools, and LMS integration to commercial providers, they have solved only part of the problem.

I agree with that, and I want to add to it. (A note on terms: I use "institution" here for any body that adopts and supports these tools, whether that's a school, a district, a university, or a ministry of education.) 

As I said in my previous post, open content has not displaced the infrastructure beneath it, and Delmar's framing of the future as "an open infrastructure for teaching and learning" rather than a bigger repository is the right one. However, two important pieces are missing: infrastructure must include the people with expertise to make it work, and institutions that depend on it must be able to run it themselves, offline if necessary, or by choice.

FlatWorld is a for-profit textbook publisher, and it publishes comparison pages that set itself against open and commercial providers. Flatworld’s central claim is that a free textbook is not a working course: the $0 price covers the text, while homework, support, and accessibility checks sit outside it. Flatworld also claims faculty report spending more time preparing with open materials than with commercial ones. Its page on LibreTexts argues that the instructor sets up ADAPT homework, connects the learning management system, and checks each remixed title for accessibility, with support from community office hours and forums rather than a named contact.

FlatWorld's conclusion is that you should pay a vendor to do this work for you. I disagree. Our institutions are smart enough to manage the systems used for teaching and learning, especially when they collaborate. When training and support are bought from a proprietary vendor, the expertise stays locked inside the vendor's product and staff. If an institution's contract ends, or the company changes its pricing or its product, the know-how leaves with it.

Open projects can avoid that, but only if openness extends beyond the content. Delmar proposes a centralized ecosystem, and centralization has real advantages, but a centralized system must enable the institutions that depend on it to run it, repair it, and change it, including when they're offline?

LibreTexts' construction guide describes ADAPT as a centralized system, delivered through LibreTexts' own site and integrated with learning management systems such as Canvas and Blackboard. But, it can't be self-hosted, and it can't be run on an offline server such as a MoodleBox, a self-contained Moodle server that works in a classroom with no internet. Components like WebWork and IMathAS depend on server-side infrastructure.

Teachers can learn to build assignments and connect ADAPT to their LMS, but the ability to run, repair, and change the platform stays with LibreTexts. An institution can build some proficiencies, but not ownership. An offline classroom has no way to use it at all. A system can have openly licensed content and still be an infrastructure that institutions can only rent or borrow.

FlatWorld's OER page acknowledges that community support has real strengths because expertise is spread across many practitioners. Peers who have solved a problem can help the next person, and it institutions save money. OpenStax runs a hub, LibreTexts runs office hours and forums, and informal listservs and faculty networks fill in around them. FlatWorld presents this as a limitation. I see it differently. A community is as strong as the shared expertise it builds, and that expertise grows when people work on problems together rather than alone. A teacher who fixes a broken LMS connection, adapts a lesson for English learners, or checks a remix for accessibility has learned something the whole community can use. If that experience is documented, discussed, and folded back into the shared materials, it benefits every institution that adopts those materials.

The words we use for our higher ed institutions point the same way. "College" comes from the Latin collegium, an association of colleagues joined in a common purpose. "University" comes from universitas, which in the Middle Ages meant a corporation or guild: a body of teachers and students who organized themselves to learn together and governed their own work. Those early universities were not open in the way we use the word today. But they started from the premise that knowledge is made and passed on by a community working in common, not owned by a single provider and delivered to customers. Too much of today's professional development relies on proprietary, for-profit vendors to supply training and support. Funded collaboration and open licensing return us to our roots.

In K-12, where most of the institutions are public schools and districts, building capacity is even more important. Most K-12 is  publicly funded and publicly governed, a community working in common on the public's behalf. When a district spends its professional development budget on a proprietary vendor, public money buys training the district can't adapt, reuse, or share, and the know-how leaves when the contract does. In my last post I argued that K-12 is the real test of open infrastructure, because it is where teacher capacity matters most and where the largest and most varied population of learners is found. That is especially true for professional learning. A teacher in a small rural district, an under-resourced urban school, and a classroom in the Northern Region of Ghana all need training that fits their context and that the people using it can improve. Openly licensed professional development, built and supported collaboratively, lets the institution keep what it learns, adapt it for its own students, and pass it to the next school instead of paying again for each cohort of teachers. It also means that money spent on teachers' growth builds a shared public resource rather than renting access to a private one..

This also creates reliability without giving up openness. Support becomes predictable when it is shared work. Our institutions need funded facilitators backed by a growing body of openly licensed, documented practice that anyone can use. An open community can promise a response because it has built the capacity to deliver it, and everyone who contributes makes the next person's problem easier to solve.

If we can afford to pay money to for-profit vendors backed by venture capital, we can afford to pay for an infrastructure that is open. Pooled institutional membership is one route, public money is another. FlatWorld's page says California state funding supported ADAPT's development, which is why students there don't pay for it. Consortium dues favor institutions that can afford to join. Public funds follow their own geography, the trap my last post warned about. Paying for training on a platform an institution can't host doesn't build capacity; it builds dependence on someone else's infrastructure.

The funding should carry conditions. Money for training and support should go to platforms and materials that institutions can run, maintain, and modify independently, including offline, so an institution's investment doesn't disappear when a private equity company changes course. And it should reach the institutions with the fewest resources, first.

Delmar says he'll be writing more about this in the coming months. Alongside homework, assessment, and analytics I hope the conversation includes funded training and facilitation as part of the infrastructure, platforms that institutions can host themselves, and designs that work offline.

The measure of an open system is whether it gets stronger as more people use it. A vendor's support model scales by adding customers and staff on the vendor's terms. An open one can scale by turning users into contributors, so each trained teacher adds to what the next one can build on. That’s the difference between merely being a consumer of professional development and taking part in building it. No price cut, however welcome, can supply that.

Sources: Delmar Larsen, https://www.linkedin.com/feed/update/urn:li:activity:7512339710636154880/ ; Delmar Larsen, A National Priority for Open Education; FlatWorld, LibreTexts vs. FlatWorld and OER vs. FlatWorld, both updated September 16, 2026; LibreTexts, Construction Guide 3e, 1.2: ADAPT. FlatWorld is a commercial publisher and its analysis is its own; pricing and policy details should be confirmed with LibreTexts directly.

CC BY NC