Digital literacy should mean more than knowing where to click
If you would rather get the core argument visually before settling into the full article, we made a short explainer for The Open School. It introduces the hidden digital curriculum beneath everyday school technology: how platforms can shape pedagogy, teacher attention, student data, AI use and institutional dependency long before anyone calls it a technology strategy.
At 8:17 in the morning, before anybody has formally begun teaching, the school is already delivering a lesson about technology. A teacher opens the attendance system. A student signs into the learning platform. Another student asks for a password reset. A parent checks a communications app. Somewhere else in the building, somebody discovers that a class list in one system has not updated in another and quietly repairs the discrepancy before it becomes visible to everyone else. None of this appears on the curriculum, yet the school has already taught several things: that identity is something a platform verifies, that access depends on permissions somebody else configures, that institutional memory increasingly lives inside software, and that when systems fail to cooperate a human being will usually be found carrying the missing piece between them.
We normally describe digital education in more reassuring terms. Students need digital skills. Teachers need training. Schools need connectivity, devices, secure platforms and high-quality digital resources. All of this is true, and Europe’s Digital Education Action Plan continues to place infrastructure, organisational capacity, digitally confident educators, accessible content and secure systems at the centre of resilient digital education. The difficulty is that these ambitions answer only the first stage of digitalisation: whether schools can use technology. They tell us much less about the relationship schools develop with the technology once it becomes indispensable.
That relationship deserves its own question. Once digital platforms hold lessons, assessments, communication, identities, behavioural data, staff workflows and years of teacher-created material, how much of the school still belongs to the school? That is not an argument against educational technology. It is an argument for treating technology as part of educational architecture rather than as a collection of useful products.
Every school now has a digital curriculum it never wrote
Education has long recognised the idea of a hidden curriculum: the values and expectations students learn from the way an institution is organised, even when nobody has written those lessons into the syllabus. A classroom teaches something about authority through who is allowed to speak. A library teaches something about knowledge through the fact that ideas can be shared, compared and revisited. A laboratory teaches that the physical world can be examined rather than simply accepted. Digital infrastructure has joined that hidden curriculum, although we rarely inspect what it is teaching.
A platform that cannot be modified teaches that technology arrives finished. A file that cannot leave its software teaches that ownership and access are not necessarily the same thing. A device that is discarded because one component fails teaches that technological objects are consumables rather than things that can be understood and repaired. A student who must accept extensive monitoring in order to complete an ordinary activity learns that surveillance is a normal condition of participation. A teacher who has spent six years constructing courses inside a platform and then discovers that only fragments of those courses can travel elsewhere learns a rather more advanced lesson about institutional dependency.
None of these outcomes requires malicious technology companies or incompetent schools. That is precisely why they matter. Dependency doesn’t enters the building accompanied by ominous music; it arrives through reasonable decisions, useful features, tight budgets, urgent needs and contracts signed by people trying to solve today’s problem before Tuesday. The hidden curriculum emerges from the accumulation.
This is why digital literacy now needs a more ambitious definition. We can think of interface literacy as knowing how to operate the tool, system literacy as understanding what sits beneath the interface, and agency literacy as understanding that digital systems contain human choices and can sometimes be challenged or changed. Schools have become increasingly effective at teaching the first of these because modern education requires students to navigate digital environments constantly. The harder task is ensuring that competence does not end at the edge of the screen.
Platform literacy can look like digital literacy
Imagine teaching writing by training students to complete increasingly sophisticated forms. They could become excellent at fields, dropdowns and character limits while remaining rather poorly prepared to write an essay. Something similar can happen in digital education when familiarity with a product is mistaken for understanding of the medium.
A learner who can collaborate in a particular cloud suite has acquired useful experience. A learner who also understands document formats, permissions, version histories, data portability and why another application may or may not be able to open the same work has acquired knowledge that survives the platform. A student who knows how to build a presentation in one interface has learned a tool; a student who understands information hierarchy, visual communication, accessible design and transferable file formats has learned something closer to a digital competence.
The distinction matters because platforms change faster than education systems. UNESCO’s global review of technology in education found that EdTech adoption frequently proceeds with surprisingly limited independent evidence; in one survey it cites, only 11% of teachers and administrators said they had requested peer-reviewed evidence before adopting technology. The report’s broader conclusion is not that technology is ineffective, but that educational need, evidence, equity and sustainability should determine adoption rather than the mere availability of a product.
There is a deeper institutional consequence. When a product makes one teaching practice effortless and another cumbersome, it begins influencing pedagogy without ever appearing in the curriculum meeting. The software does not need to prohibit anything. It simply needs to make one path smooth enough that it becomes normal and another irritating enough that teachers use it less.
Curriculum captivity begins with convenience
We tend to discuss vendor lock-in through technical symptoms: proprietary file formats, difficult exports, costly integrations and painful migrations. Education has another form of dependency that is easier to miss because it gradually becomes professional habit. We can call it curriculum captivity: the point at which teaching practices begin reorganising themselves around what a platform makes easy.
A learning system makes multiple-choice quizzes almost effortless, so more of them appear. Portfolio assessment requires an awkward sequence of uploads and permissions, so it happens less often. Collaborative annotation requires another service, so teachers either abandon the idea or create unofficial accounts elsewhere. A new AI tool can produce rapid short-answer feedback, and assessments slowly begin shifting toward forms the system can easily interpret. Nobody has ordered teachers to redesign learning around software. The interface has simply changed the relative cost of different choices.
This is not always harmful. Good technology can make genuinely valuable pedagogy possible at a scale that would otherwise be difficult. A collaborative platform can widen participation; accessible digital content can remove barriers; routine automation can release time for meaningful feedback. The professional question is not whether technology influences pedagogy, because of course it does. It is whether the institution can still distinguish between a pedagogical decision and an interface-induced habit.
That question becomes especially important when software has been present long enough that nobody remembers how the process worked before it. At that point, a platform is no longer merely supporting the curriculum. It may have become part of the curriculum’s architecture without ever formally applying for the job.
The school that rents its memory
There is another form of dependency that deserves more attention: institutional memory.
Teachers create extraordinary amounts of intellectual infrastructure. They build courses, explanations, feedback banks, assessment structures, rubrics, interactive resources, reading lists, learning sequences and accommodations that often reflect years of accumulated professional judgement. Increasingly, these things do not live in filing cabinets or even on school-owned drives. They live inside platforms.
That can be enormously useful. A good learning environment allows resources to be organised, reused and improved across years. The problem appears when the institution cannot confidently answer what would survive if the platform disappeared.
Can the course structure travel, or only the files? Do quizzes retain their logic? Do question banks survive? What happens to rubrics, comments, discussion history and student-generated artefacts? Are relationships between resources preserved, or does an export produce a large folder containing the educational equivalent of dismantled furniture?
When the answers are unclear, the school may technically own its content while renting access to its own institutional memory.
This is one reason interoperability standards matter in education. Common Cartridge, for example, exists to package learning resources in ways compatible systems can interpret, while Learning Tools Interoperability provides standardised mechanisms through which external educational tools can connect to learning platforms. These standards do not make migration effortless and they do not guarantee pedagogical quality, but they move educational infrastructure away from the assumption that every piece of knowledge must remain understandable only to the product in which it was born.
An open school is not a school that has successfully installed Linux everywhere
There is an attractively simple version of this article in which open source solves everything. We remove proprietary platforms, replace them with open software, celebrate digital sovereignty and go home slightly earlier.
Reality is less cooperative.
An open-source application can be inaccessible, badly supported, poorly documented or entirely wrong for a school. A proprietary platform can be excellent, standards-based, highly portable and backed by contractual protections that give the institution substantial control. Software licensing matters because it determines important rights over the code, but a licence is not a certificate of educational wisdom.
An open school therefore needs a broader definition. It is a school that retains enough agency over its digital environment to understand how that environment works, influence important decisions, protect the people inside it, preserve what it creates and change direction without dismantling its educational model.
Open-source software can contribute to that agency. So can open standards, openly licensed educational resources, repairable hardware, portable identities, documented interfaces, strong exit clauses and a school community with enough technical understanding to participate in decisions. REDefine’s own mapping of open educational infrastructure reaches the same conclusion: the value lies not in replacing one branded stack with another, but in combining platforms, identity, content, devices and support into an ecosystem that can be adapted and maintained. OpenSourcereport_full vf
The important distinction is therefore not open versus closed in the abstract. It is agency versus dependency in practice.
Valencia shows why openness can make technology more local
The Valencian education system offers a useful example because its LliureX project demonstrates something that is easily lost in debates about cost. LliureX is a GNU/Linux distribution created specifically for the Valencian educational environment. The Generalitat describes its ability to follow the school calendar, incorporate software for particular subjects and education levels, provide documentation for teachers and students, and support Valencian alongside Spanish. It also reports licence savings of more than €30 million. Ceice Gva
The savings are impressive, but the more interesting feature is where the ability to adapt resides. Language support does not depend entirely upon whether a global product manager believes Valencian is commercially important enough for the next release. The educational administration can shape the environment around the needs of its own schools, and its 2024–2028 digital education strategy continues that model through LliureX, the Moodle-based Aules learning platform, WordPress-based school sites and open-resource creation tools.
There is a useful paradox here. We often associate global proprietary platforms with sophistication and local solutions with limitation, yet open infrastructure can create the opposite effect. Because more of the underlying environment can be changed, the system can become more specific to local language, pedagogy and institutional rhythm while still drawing on a much larger global ecosystem.
Openness, in other words, can make technology more local without making knowledge smaller.
Kerala demonstrates something even more important: capability can be distributed
Kerala’s KITE programme offers another model. KITE currently reports work across 16,028 government and aided schools, including 45,000 high-tech classrooms, and combines infrastructure, connectivity, teacher capacity, content development, e-governance, support and maintenance rather than treating digital education as a software-purchasing exercise.
Its Little KITEs programme is especially interesting because students are not positioned only as technology consumers. The initiative reports around 175,000 current members across more than 2,000 schools, with activities extending into programming, robotics, cyber safety and AI. This does not mean every learner is being trained to become an engineer. It means the educational ecosystem creates opportunities for students to encounter technology as something that can be constructed, investigated and used to create new tools rather than merely consumed through interfaces.
KITE also maintains hardware support, teacher training and local technical capacity. That matters because there is a considerable difference between outsourcing technical work and outsourcing technical understanding. Schools will often need external suppliers, cloud providers, specialist developers and professional support; insisting that every institution do everything itself would produce less sovereignty and considerably more exhausted staff. The question is whether enough understanding remains inside the education system to govern those relationships intelligently.
This is where open education becomes a capacity question rather than a software question.
The right to learn without buying the classroom
The same principle applies to students after they leave school.
Professional software deserves a place in education because students should encounter tools they may use in employment. An architecture student may need exposure to industry-standard design environments; a media student should understand professional production workflows. The problem begins when the commercial product becomes indistinguishable from the underlying competence.
A student should ideally leave knowing more than where a particular command sits. A designer needs to understand vectors, layers, formats and colour systems. A data student should understand tables, queries, relationships and portability. A writer should understand collaborative revision, structure and versioning rather than merely the conventions of one cloud suite.
Otherwise the institution risks graduating highly competent tenants of somebody else’s ecosystem.
This is also why freely available and open tools can matter beyond budgets. When students can legally install the same environment at home, continue using it after graduation and inspect or extend it if they become more advanced, access to the learning environment is less tightly tied to institutional membership.
The principle is not that schools should shield students from commercial technology. It is that professional preparation should not become product captivity.
A sophisticated digital school must work on an unsophisticated connection
Educational technology is often demonstrated under ideal conditions. The device is recent, the network fast, the account correctly configured and nobody’s younger sibling is streaming something on the same connection from the next room.
Home is less cooperative.
A learner may have a phone but no laptop. Broadband may be shared. Mobile data may be limited. An old device may struggle with a platform designed around newer hardware. A disability can turn an apparently polished interface into an obstacle course. A system that performs perfectly on a school network can become unusable on the kitchen table at 20:30.
That means digital inclusion is not simply an access programme. It is an architectural requirement.
Schools should know whether essential materials can be downloaded, whether activities work on lower-cost devices, whether interfaces survive low bandwidth, whether assistive technologies can complete the whole learning journey rather than merely reach the front page, and whether an offline route exists when connectivity fails. Our open-source resource mapping highlights platforms such as Kolibri precisely because offline-first design begins from the assumption that permanent connectivity is not universal rather than treating imperfect infrastructure as a user failure. Module 2B – Strategic Open Sour…
Sometimes technological sophistication means adding another service.
Sometimes it means designing something that continues working when the network disappears.
Educational data has a half-life
Schools have always kept records, but digital learning environments create a different kind of memory. Attendance, assessments and qualifications are joined by login times, click patterns, submission histories, engagement measures, interaction data and increasingly records generated through AI-enabled tools.
Storage makes remembering easy.
Educational judgement should decide whether remembering is useful.
A grade may need to remain part of an official record. A safeguarding record may have defined legal requirements. Information about a particular intervention may matter for one year. The fact that a student failed to open a Week 3 worksheet at 19:42 on a wet Thursday may have a considerably shorter educational career, although a database is quite capable of preserving it with touching loyalty.
This suggests a concept that education technology governance badly needs: pedagogical data half-life.
For every type of learner data, schools should ask not only whether collection is lawful but when the information stops being educationally useful. Jisc’s recently updated code for learning analytics takes a similar direction, requiring institutions to define purposes, collect only the data needed for those purposes, establish clear responsibilities, involve students and staff, validate analytics and retain information only for appropriate periods. Its 2026 update also warns that analytics and AI can never provide a complete picture of a learner and should not allow categorisation to harden into bias or diminished autonomy. Jisc
The professional principle can be stated simply without making it simplistic: every educational data point should have a reason to exist, a reason to travel and a reason to disappear.
There is something educationally important about the final part. Schools are institutions in which people are supposed to become different from who they were last year. Their databases should leave room for that possibility too.
AI has created a new governance problem: the product can change while the approval stays the same
The educational debate around generative AI began, understandably, with assessment. Would students cheat? Could teachers detect generated work? Should AI be prohibited, incorporated or taught explicitly?
Those questions are already giving way to a deeper one.
What happens when AI acquires decision weight inside the institution?
The EU AI Act treats certain educational uses as high-risk, including systems intended to determine access or admission, evaluate learning outcomes where those outcomes influence a learner’s path, assess the appropriate level of education, or monitor prohibited behaviour during examinations. The distinction is important because an AI tool helping a student generate revision questions and an AI system influencing admission or progression may both be called “AI”, but the governance consequences are not remotely equivalent.
There is also a problem that conventional procurement was not designed to handle. A school may evaluate an AI product in September, approve it for a particular purpose and write guidance around the behaviour staff observed. Six months later the logo and interface can look essentially unchanged while the underlying model, system instructions, retrieval capabilities, memory functions or data practices have evolved.
Traditional software procurement assumes some continuity between the product approved and the product used.
Generative AI makes that assumption less reliable.
An open school therefore needs AI change control, not merely an AI policy. Institutional approval should record what the tool was approved to do, what data may enter it, who may use it, which decisions require human judgement, what behaviour was tested and what types of material change trigger reassessment. Jisc’s 2026 guidance now explicitly extends its analytics principles to AI, emphasising explainability, validation, student contestability, proportional data capture and human responsibility for interpretation. Jisc
That is a considerably more useful governance question than asking whether the school “allows ChatGPT”.
Repair teaches something screens cannot
There is something wonderfully contradictory about delivering a sustainability lesson through a device nobody is permitted to repair.
Schools now depend materially on laptops, tablets, displays, routers and specialised hardware. Procurement therefore shapes environmental behaviour as surely as the recycling poster in the corridor. A replaceable battery, available spare parts and a realistic maintenance pathway are not only operational details; they determine whether an expensive object becomes waste because one component failed.
Repairability also changes the educational relationship with technology. A machine that can be opened becomes something with parts, trade-offs and material consequences. Students can encounter diagnosis, maintenance and reuse rather than experiencing technology as sealed magic.
KITE’s model is again instructive because hardware clinics and maintenance sit alongside teaching infrastructure rather than being treated as an entirely separate concern. UNESCO’s sustainability analysis similarly argues that procurement needs to consider environmental lifecycle as well as educational and financial performance; it notes that extending laptop life across the EU by a year has been estimated to save emissions equivalent to removing almost a million cars from the road.
A broken keyboard should not automatically require a funeral for the entire laptop.
Schools of all places should understand that.
Public education should accumulate capability, not only subscriptions
This brings us to a larger civic question.
When public money is spent on education technology, what remains when the contract ends?
A licence can provide excellent value. A hosted service can spare a school the considerable burden of managing infrastructure itself. Commercial support can be worth far more than the apparent savings of doing everything internally. The issue is not whether money leaves the institution; schools buy expertise and services for perfectly good reasons.
The question is whether years of expenditure also build institutional capacity.
Does staff knowledge increase? Is documentation retained? Can another supplier maintain the environment? Can teachers preserve what they created? Are workflows based on transferable standards? Has the institution developed enough understanding to negotiate the next contract more intelligently than the previous one?
UNESCO’s procurement analysis gives this a useful economic edge. It cites estimates that the initial purchase of education technology may represent only 25% or less of the eventual total cost once implementation, operation and maintenance are included. 2023 GEM Report A cheap licence can therefore support an expensive architecture, while a more expensive product may reduce overall cost if it saves staff time, integrates well and remains easy to leave.
Open source changes some of those economics, but it does not abolish them. Hosting, maintenance, cybersecurity, training and support remain real work. The stronger case for openness is that money can sometimes build reusable capability: local expertise, documentation, adaptations, integrations and systems another provider is permitted to maintain.
The difference is not between spending and not spending.
It is between renting access and building capacity while you spend.
The question beneath the classroom floorboards
The student sees an assignment. The teacher sees a course. The administrator sees an enrolment. The IT team sees identities, devices and permissions. Finance sees contracts, while a supplier sees a customer and, depending on the service, perhaps a rather impressive stream of data.
Underneath all of these perspectives sits one educational architecture assembled from pedagogy, licences, standards, infrastructure, data, expertise and thousands of small decisions that rarely felt like architecture when somebody made them.
That architecture teaches.
It teaches whether knowledge can be adapted, whether technology can be questioned, whether mistakes should become permanent data, whether a broken object can be repaired and whether an institution can change its mind without starting again.
An open school does not require every child to read source code or every teacher to maintain a server. It requires something more reasonable and, in many ways, more difficult: enough institutional understanding that technology remains a tool of education rather than gradually becoming its landlord.
This leaves us with the question beneath this week’s floorboards. When students leave school digitally fluent, have we taught them only how to live confidently inside the systems they inherited, or have we also taught them that systems are human constructions that can be understood, questioned and changed?
Continue beneath the floorboards
The open version stops here, but the practical work begins underneath it.
In the full Substack edition, we move from the idea of the Open School to the architecture behind it: how educational organisations can identify where technology is consuming teacher attention, reshaping pedagogy, accumulating unnecessary data, creating AI governance risks, or becoming difficult to replace.
The professional section introduces the REDefine Educational Technology Dependency Model and walks through tools for examining curriculum captivity, teacher attention, interoperability, data half-life, AI change control, continuity and exit risk.
Paid subscribers can also download the REDefine Open School & Educational Agency Audit — a connected working tool designed for schools, universities, training providers and educational organisations that want to map these dependencies and turn them into practical decisions.
Subscribe on Substack to unlock the full article and download the audit.
