Version 1.0, September 2026. A living document. See the changelog at the bottom of this page.
Educational technology is not neutral. Every platform, every pricing model, every data policy reflects a set of decisions about who benefits and who adapts. This document is about who those decisions belong to, and who they currently belong to instead.
I am writing from inside one of the places where those decisions get made and lived with. I work at Black Hills Special Services Cooperative, an educational service cooperative serving districts across the Great Plains region. My division provides training and coaching to rural districts, mostly in South Dakota but increasingly through contracts that extend further. I manage our Canvas learning management system for adult learners, develop and facilitate courses, and lead trainings on educational technology, often on generative AI. I also do grant work, including a rural post-secondary economic development project building self-paced courses for high school students considering teaching as a profession.
What I see from this position is a particular pattern. Rural districts purchase educational technology not because the technology meets a need they have identified, but because their neighboring district has purchased it, or because a sales pitch landed during a leadership transition, or because they did not want to look behind. The decision is rarely framed as a structural choice with downstream consequences. It is framed as a routine procurement. Then five years later the district is locked into a platform whose defaults shape how teachers think about their work, whose data flows out to a vendor whose interests do not match the district’s, and whose costs have crept beyond what the original decision accounted for.
What I see
The pattern I keep noticing is that small districts are paying for platforms they did not choose by their own assessment, do not fully use, cannot really change, and cannot easily leave.
They did not choose by their own assessment because the channels that shape ed-tech purchasing in rural districts run through influencer presentations, vendor-sponsored conferences, and informal comparison with peer districts more than through structured needs assessment. A district considering Canvas does not usually start with what kinds of mentoring or coaching structures would fit local needs. The conversation begins with what other districts use, what workflows are already documented somewhere in a vendor knowledge base, and what configurations are available. A teacher who happens to hold a Microsoft certification ends up advocating for the Microsoft option. A superintendent who sat next to a sales rep at a conference dinner ends up softening on a tool they were skeptical of in the morning. None of this is corrupt. It is how purchasing happens in small districts that don’t have the time or the staff to do it differently.
They do not fully use what they pay for because the platforms are built for markets the rural districts are not really in. Vendors design for the contexts they actually represent in their development process, which is predominantly suburban and urban districts with larger student populations, dedicated IT staff, and the institutional capacity to participate in user research and beta testing. A small district in western South Dakota encountering a mismatch between a platform’s assumptions and local need has fewer mechanisms available to register that mismatch in any way that affects future design decisions. So they use the platform’s surface features and leave the rest unused, paying for it anyway.
A specific version of this dynamic shows up around language. I have asked major ed-tech vendors whether their platform supports Lakota, the indigenous language still spoken by people across the region we serve. Most of the time the people I am asking do not recognize the word. Their platforms support dozens of languages, sometimes hundreds, but the languages they support are the languages of large markets. Lakota is not a large market. It is also a language whose romanization carries political weight, so it is not something a vendor could plausibly add by spinning up a quick translation pipeline. A tool that supports learning Lakota in a way that respects the communities the language belongs to is not a tool a platform like Duolingo will build. It is a tool a local educator with the right relationships and the right knowledge could build.
They cannot really change what they bought because the design authority over a vendor platform sits with the vendor, not with the customer. Educators and administrators can choose settings within permitted ranges, but they cannot alter the underlying logic that determines what choices exist. Buying Canvas is not just buying a tool. It is buying an argument about how courses are organized, how assignments are designed, what counts as engagement, and which information about students gets made visible to whom. Those decisions were made by a product team optimizing for a scalable market, not by anyone with a stake in the specific learning happening in a specific classroom in a specific town.
They cannot easily leave because the lock-in is structural rather than contractual. The annual seat cost is only part of the price. There is also the time teachers spent learning the platform, the time spent reformatting materials to fit it, the institutional muscle memory built around its conventions, and the cost of pulling years of accumulated content off it if a different platform were chosen. The platform becomes the infrastructure. The infrastructure becomes invisible. Eventually no one questions it, not because it is good, but because questioning it would be too expensive.
The pattern I am describing has a name in the critical education literature. Michael Apple calls it technical control, the structural arrangement in which design authority becomes separated from the practitioners who carry out educational work, sustained by what Raymond Williams called selective tradition, the cultural process by which particular arrangements become naturalized as the legitimate way of doing things. I am not building from this framework so much as recognizing that the patterns I keep noticing in my work have been observed before and named by people who studied them carefully.
There is a second pattern underneath the first one, and it is harder to talk about because it involves expertise. When a district buys a major ed-tech platform, the deal includes implicit access to expertise: documentation, professional development materials, certification programs, conference content, knowledge bases. That expertise feels neutral, but it is not. It has been shaped by the vendor’s profit motive. The most heavily promoted “best practices” are the practices that drive deeper adoption of the platform’s features. The certification programs build a workforce of practitioners whose professional identity is tied to the platform’s continued use. Even the conference talks that look like independent scholarship are often delivered by people whose careers benefit from the platform’s success. None of this is conspiratorial. It is what happens when a single commercial entity becomes the dominant intermediary in a professional field.
The third pattern is about how teachers themselves are positioned by these arrangements. A teacher in a small rural district is increasingly asked to be an implementer of decisions made elsewhere. The curriculum was purchased. The platform was purchased. The assessment was purchased. What remains for the teacher is execution. The pedagogical reasoning, the design judgment, the response to particular learners and particular contexts, all the things that constitute teaching as a profession rather than as a job, get squeezed out of the work over time. This is true on the platform side, and it is also true on the staffing side. In some of the districts I work with, the teaching workforce includes a substantial number of teachers brought in through Teach for America or through international recruitment programs that place teachers from outside the country into rural classrooms, sometimes via remote arrangements. Whatever the merits of any individual program, the cumulative effect is to treat the teaching role as interchangeable enough that the professional standing of any one teacher carries less weight. Platforms deprofessionalize teaching from one direction. Staffing patterns deprofessionalize it from another. They are different mechanisms but they point in the same direction.
There is a dimension of all this that I have not yet named directly, and I want to be honest that it exceeds what this document can fully engage. The region I work in includes nine sovereign tribal nations and significant numbers of students attending schools on or adjacent to reservations. Pine Ridge, Rosebud, Standing Rock, Cheyenne River, and others are part of the same geography this work touches. The patterns I have been describing about vendor dependency, deprofessionalization, and the foreclosing of local design authority apply in tribal contexts as well. They also intersect with histories and power dynamics that the patterns alone cannot account for. The history of educational technology in Indigenous communities is bound up with the history of education itself in those communities, which is bound up with treaty relationships, boarding school legacies, language loss and revitalization, and questions of sovereignty that this document cannot resolve.
The Lakota example I mentioned earlier is one small entry point into this terrain. A platform that does not recognize the word Lakota has decided, by default, that this place and these people are not part of its market. Even a platform that wanted to support Lakota would still have to navigate the politics of romanization, the relationships between different community-based standards for written Lakota, and the question of whether the platform is the right institution to be doing that work at all. These are political and relational questions, not technical ones. A locally built tool, designed in conversation with the people whose language it serves, can do something a vendor platform cannot do regardless of how much development funding the vendor has.
I want to name this dimension of the work even though I cannot fully engage it in this document. A dissertation that takes up the questions I am raising will have to engage it more carefully, and I am still working out what that engagement requires of me. For now, the most honest thing I can say is that the alternative I am pointing toward has different stakes in tribal contexts than in other rural contexts, and that the difference is not incidental.
What I believe
I’ll keep this section as plain as I can.
Educational technology is not neutral. Every tool encodes an argument about how learning happens, who is responsible for what, what counts as evidence of progress, and which forms of attention are valuable. A district adopting a tool is also adopting those arguments, whether or not anyone in the room names them.
Design authority over the digital environments of education should sit with the people who teach and learn in them. Not with vendors whose interests are not aligned with the local context. Not with technology coordinators whose mandate is to keep things running. Not with administrators whose primary relationship to the tool is the purchase order. With the teachers and learners themselves, with the educators who serve them, and with the communities those educators work in.
Rural communities and small districts are well-positioned for this kind of work, not poorly positioned. They have constraints that larger districts do not have. They also have strengths that the dominant ed-tech discourse does not recognize. People know each other. Teachers see students at the gas station. Principals deal with classroom behavior personally, which keeps them accountable to the daily reality of teaching. When a tool gets built locally and a problem arises with it, the person who built it is reachable. Resilience is built into the relationship structure of a small place. The current discourse treats rural districts as places that lack what larger districts have. They lack some things, but they also carry things the discourse misses.
Students are not learners. Or rather, students are learners only in the trivial sense that everyone is learning all the time. What “student” names that “learner” does not is the institutional relationship: a person in a school, with a teacher, under a curriculum, subject to rules and expectations they did not write. The institutional specificity is the point. To call a student a learner is to abstract away the institution, and to abstract away the institution is to make the power dynamics of education invisible. I want them visible.
Students are participants in the institutional arrangement, not objects of it. They have wants and needs and developing autonomy. They are also vulnerable to forms of harm they cannot yet recognize, especially harms that arrive on a delay. A student posting personal information on a discussion board does not always understand that what they posted persists in ways other forms of disclosure do not. The adults around them may not always understand it either. Part of what schools owe students is protection from arrangements whose consequences are not yet visible to anyone in the room.
What I am trying to protect students from has two layers. The first is more visible in the immediate term. Students deserve to be taught by people who live in the same community, understand the local context, and can adjust their teaching to the actual students in front of them rather than to a model of students built somewhere else. Every vendor platform that mediates the teaching relationship is an entity that does not know the student. The more of those entities sit between the teacher and the student, the more the relationship becomes about the platforms rather than about the people.
The second layer matters more in the long run, and it is the one I think about most. A student who only ever experiences technology as something that arrives prepackaged from a vendor, that they cannot modify, whose terms they cannot question, learns a particular set of habits about what technology is and what their relationship to it can be. Those habits do not stop at graduation. They become the default frame for how that person interacts with technology as an adult worker, citizen, voter, parent. I want students to leave school with the capacity to imagine that technology can be made, modified, refused, and that those choices belong to them and to the communities they live in. That capacity will not develop on its own. It has to be cultivated, and the institutional environment of school is one of the easier places to cultivate it.
Teaching is relational. It happens between people. When a piece of technology is inserted into the relationship without the teacher having chosen it, the technology becomes a third party in the conversation. Sometimes that third party is useful. Often it is intrusive. The teacher’s ability to decide which is which is a professional judgment the current arrangement frequently does not allow them to exercise.
Design is a political act. It is also a craft. It is sometimes a service and sometimes a gift. What it is not is a neutral technical activity. Every tool reflects the values of the people who made it, the constraints they accepted, and the relationships they prioritized. Pretending otherwise is what allows the current arrangement to persist.
Change is continuous. Nothing about the current arrangement of educational technology is permanent, even though the institutional habits surrounding it can feel permanent. Octavia Butler’s line, that all you touch you change and all you change changes you, is a working principle rather than a literary flourish. The act of building a tool changes the builder. The act of using a tool changes the user. The current arrangement was built. It can be rebuilt.
How I work
The beliefs above are not abstractions. They show up in how I do the work, and the practices I try to follow are also the practices I try to share with others.
When I build something for a district, the first concern is whether the district can actually own what I built. Ownership in this context means more than possession. It means the district has the code, the documentation, the configuration access, and the institutional knowledge to maintain and modify the tool without me. Most of the small tools I build are hosted in Google Apps Script bound to a Google Sheet, which is itself a tradeoff. Google is a vendor, and the critiques I make of Canvas apply at one level to Google as well. The practical reality is that a district can copy a Google Sheet, set up the script in their own account, and run it for free. They don’t need to learn GitHub or pay for hosting or rely on me to keep something running. The tradeoff is shareability and accessibility in exchange for some platform dependency. I have not fully resolved the contradiction. I try to name it openly when it comes up.
I keep the code transparent and put configuration in a visible place, usually a settings tab in the sheet, so that non-coders can adjust the tool without touching the underlying logic. I document what I built and why, including the reasoning behind specific decisions, so that whoever inherits the tool inherits the thinking with it.
I try to build for the right lifespan. Not every tool needs to last forever. A spoon analogy is useful here. A plastic spoon makes sense if you are serving a lot of people once. A titanium spoon makes sense if you are building something to last. Building everything to last is wasteful. Building nothing to last is also wasteful. A tool’s lifespan should match its purpose, and most ed-tech today is built to a single default lifespan, perpetual subscription, that does not match most of what schools actually need.
I build for educators, not for districts as administrative entities. The distinction matters. A district can buy a tool that no teacher wants and impose its use through hierarchy. An educator-led adoption looks different. It starts with someone in the building who has a problem, builds a tool that fits the problem, shares it with colleagues, refines it through use. The institution may eventually formalize the arrangement, but the locus of authority stays with the people who actually use the thing.
I refuse projects that automate teachers out of their work, or that promise to make teachers more efficient without first asking why the inefficiency exists. Efficiency is usually a euphemism for compressing work that should be expanded. I also refuse projects whose framing relies on terms like “student success” or “future ready” or “21st century skills,” because those terms tend to obscure what is actually being asked rather than clarify it.
I try to keep the design principles small and operational. Calm design means cutting features and visual noise that do not serve the user. Life-centered design means accounting for the environmental and social costs of the tool, including the power and water consumed by AI inference and the data centers behind it. Biomimicry, for me, is mostly about looking for craftier and lower-energy approaches when the obvious approach is heavy. If a chatbot can do keyword matching against a small index before sending a query to a model, that is biomimicry in practice. Sustainable systems thinking is mostly about asking what happens when the tool fails, when the builder leaves, when the funding runs out.
Documentation is part of the ethos, not a side task. A tool without documentation is owned by whoever built it. A documented tool is ownable by whoever can read the documentation. This is one of the easiest things to skimp on and one of the most important things to maintain. I am not always good at it. I am trying to be better.
I work with teams, not heroes. A tool built by one teacher who shares it with no one is more fragile than the same tool built by three teachers who share it with each other. In rural contexts, turnover is real, and the resilience of a tool depends on whether more than one person knows how it works. This is part of why documentation matters. It is also part of why I push back when a district wants to designate a champion rather than build a small team.
What I’m pointing toward
I want to be careful here, because the alternative I am pointing toward is easy to romanticize and harder to actually build.
The future state I see is one in which small groups of educators in a region build, share, and maintain their own tools for their own needs. Not all tools, and not in opposition to vendors entirely. Vendors will still have a place. The default position would be inverted. Instead of starting from what platform should we buy, districts would start from what do we actually need, and can we build it. Some answers would be vendor products. Many would not.
In that future, teachers are not just consumers of educational technology. They are also among its makers. That shift changes more than what gets built. It changes how teachers relate to all the technology in their work, including the vendor tools they continue to use. A teacher who has built a small tool understands what tools are made of, what they cost to maintain, and what assumptions are embedded in them. That understanding makes the teacher a more capable user of every tool, not just the ones they built.
In that future, districts own their own data in the practical sense, not just the legal sense. Their information lives on servers they can access, in formats they can read, in tools that do not lock them out when the contract ends or the rates change. Local LLM tools like LM Studio, hosted on a district’s own hardware, become realistic options for the inference work that is currently outsourced to commercial APIs. The technical capacity already exists. What is missing is the assumption that this is something a district could do.
In that future, the gap between what teachers know about their students and what software companies know about those students closes. The teacher remains the person who knows the student best, because the student is in their class every day, because the teacher lives in the same town and shops at the same store and sees the family at the football game. The software stays in its appropriate role, as a tool for specific tasks, rather than expanding into the role of an intermediary that knows everything.
Cooperative work in the Great Plains
The alternative I am describing is not new to the Great Plains. It is older than the technology it would apply to. When private utilities decided that electrifying rural America was not profitable enough to bother with, communities in this region built electric cooperatives instead. The Rural Electrification Administration of the 1930s provided the financing and the legal framework, but the work of actually wiring the farms and running the lines was done by farmers, organized as members of cooperatives they owned together. The result is that today, much of rural America gets its electricity from cooperatives that have outlasted multiple generations of investor-owned utilities.
The same pattern shows up across other forms of infrastructure that private capital declined to build at sufficient scale or quality. Telephone cooperatives took on the work in places where the Bell System found it unprofitable to extend service. Grain elevators were built and owned by the farmers who used them rather than by distant traders. Credit unions emerged where the depositors were also the members. Agricultural marketing cooperatives pooled the output of many small producers into volumes large enough to negotiate with major buyers. Mutual aid in fire protection, water management, and irrigation followed similar lines.
The pattern is consistent enough to be its own tradition. When market forces will not serve the place adequately, the place organizes to serve itself, and the institutional form it tends toward is the cooperative. Not because cooperatives are ideologically preferred, but because they work in this kind of geography and this kind of economy. Sparse population, distance from centers of capital, climate that punishes individual failure, agricultural economies with shared timing constraints, all of these conditions produce institutions that pool risk and effort. The Great Plains taught itself how to build shared infrastructure that the rest of the country was buying from somebody else, and that institutional knowledge is still in the ground.
There is a complication I should name. The cooperative tradition I am describing is also entangled with histories that this region has not reckoned with adequately. The same period that produced the agricultural cooperatives produced the policies that dispossessed Native nations of their land. The electrification work of the 1930s did not extend to reservations on the same terms. The cooperative tradition is real, but it is also a tradition that was built largely by and for settler communities, and any extension of it into the present has to account for that. I am pointing to the cooperative form as a usable inheritance, not as an uncomplicated one.
Educational technology in the Great Plains has not yet caught up to this tradition. Districts continue to buy their digital infrastructure from vendors at scales that mostly serve other markets. The cooperative model has not yet been extended to the digital tools of teaching and learning. There are partial precedents in educational service cooperatives, which I will come back to in a moment, but those have largely operated as intermediaries between districts and vendors rather than as builders of shared infrastructure in their own right. There is no fundamental reason for this to remain the case. The same logic that built the electric grid in this region could build a different kind of ed-tech infrastructure if the institutions and the people in them decided to do the work.
This is where the educational service cooperative model becomes structurally interesting. ESAs in South Dakota, BOCES in New York, AEAs in Iowa, ESDs in Washington, and similar agencies across other states sit between districts and the larger marketplace by design. They were built as multi-district shared infrastructure, governed by boards of member districts, operating at scales no single small district could justify. The same structural form that explained why an electric cooperative made sense in 1936 also explains why an educational service cooperative made sense when it was created and continues to make sense now. Both exist to provide infrastructure that the market does not provide at appropriate scale or terms.
What educational service cooperatives have not generally done, including the one I work for, is take up the work of actually building tools and infrastructure for member districts at the digital level. The work has mostly been in training, coaching, and helping districts navigate vendor relationships. There is no structural reason it could not also include the work of building locally. The institutional form is already there. The question is what the institutions inside the form choose to do with it.
My own role
I want to be honest about my own role in this picture. If the alternative I am pointing toward succeeds at scale, my current role becomes less necessary. I do not want to build a profession of people who coach educators through tool-building forever. The point is to make the work transferable, not to make it a permanent service relationship. Right now I am at the front of this work because I have the time, the privilege, and the institutional position to be there. Ideally, in ten years, the work I am doing now is being done by the educators themselves, and my role has changed into something else.
I do not know yet what that something else will be. Maybe it is writing. Maybe it is helping educational service cooperatives in other states do this kind of work. Maybe it is something I cannot see yet. The honest answer is that the future I am pointing toward is one in which my current role is provisional, and I am comfortable with that.
A closing note
This document is not a manifesto, exactly. Manifestos commit you to a future state. What I have written is closer to a working theory of where the field is, where it could go, and where I am standing while it figures out which direction it is going to take. I will keep revising this as the work develops.
Changelog
Version 1.0, September 14, 2026. First published June 28, 2026 as part of the vibe coding in EdTech series under the same title. Moved to this page with no changes to the text, so that it can be revised openly and cited stably as the work develops. The reasoning behind the move is in An Ethos With a Changelog.
