Revised September 2026. The original version of this piece made a number of claims without sources, borrowed a passage from Martin Weller without crediting him, and described a lawsuit that had already been dismissed. This version fixes those problems and adds what has happened since, including the Canvas breach of spring 2026.
You’ve felt it, even if you couldn’t name it. The study tool students relied on locks its best features behind a subscription. The course platform that promised free education now shows you the first module and asks for a card. The learning management system keeps adding features that make everything more complicated while support gets slower and the bugs pile up. Cory Doctorow gave this a name, ensh**tification, and once you see the pattern you find it everywhere in educational technology.
The pattern
Doctorow’s argument describes a predictable life cycle for digital platforms. First a platform is good to its users, often at a loss, to draw them in. Once the users are locked in, it turns to serving its business customers at the users’ expense. Once those customers are locked in too, it squeezes both to move the surplus to shareholders. Then it dies, or limps on in a degraded state that everyone tolerates because leaving is worse.
What makes this dangerous in education is the asymmetry of power. When a social network degrades, you can delete your account. When a course platform paywalls its lectures, a student partway through can’t switch mid-semester. When a district signs a multi-year LMS contract, the teachers inside it have no exit, whatever the platform does next. Students and educators are captive audiences, and captive audiences are exactly who the third stage of the pattern is built to extract from.
The MOOC promise
The clearest case in education is the one that began with the most idealism. Coursera launched in 2012 promising to make the best education in the world freely available to anyone. edX, created by MIT and Harvard the same year as a nonprofit, was meant to be the alternative to the commercial platforms. Neither promise survived.
Class Central, which has tracked MOOC monetization for years, documented Coursera’s progression: full free access in the early years, then paid certificates and specializations while auditing stayed free, then, in mid-2025, a Preview Mode that replaced auditing and let non-paying learners see only the first module of a course. Coursera’s own announcement described this as a more meaningful way to explore content. The timing is what gives the game away. Class Central noted that the change arrived under a new chief executive, an Amazon veteran, at a moment when the company was raising its revenue guidance toward $746 million, generating free cash flow, and holding roughly $775 million in reserves. Rather than a platform fighting for survival, this was a platform whose stock price was still far below its IPO, looking for a lever.
edX went further. In June 2021, MIT and Harvard sold it to 2U, a for-profit online program manager, for $800 million. The universities framed the sale as the only way to keep the mission alive against better-funded competitors. Three years later, 2U filed for bankruptcy, its shares having lost most of their value within a year of the deal, so whatever the sale was, it wasn’t a plan for sustainability.
When everyday tools turn
Student-facing tools have followed the same script at a smaller scale. On August 1, 2022, timed to the start of the school year, Quizlet moved its Learn and Test modes behind its Plus subscription after years of free access, limiting free users to a handful of rounds per set and removing a popular study game entirely. The response came mostly from students themselves, in school newspapers across the country, and one of them noticed the sharpest detail: Quizlet’s own page on the science of studying recommended exactly the modes it had just locked away. The company’s own pedagogy became a premium feature.
The LMS and the question of data
Learning management systems deserve their own section, since they combine lock-in with something the study apps don’t have: a complete record of what students do. The switching costs are enormous. Proprietary formats, deep integrations, and multi-year contracts hold institutions in place even as the platform degrades, and the lock-in works on people as much as on organizations. I wrote about being known for a platform I don’t control in Known as the Canvas Guy.
In March 2025, parents in California and Maryland filed a proposed class action against Instructure, the company behind Canvas. The complaint alleged that the company collected far more than an education record, including messages, discussion posts, grades, search activity, and uploaded essays and creative work, and that it shared student information with a large number of other companies. In August 2025 a federal judge dismissed the suit, finding the plaintiffs hadn’t pleaded specific enough facts about what the company actually did with the data. That’s a ruling about the complaint rather than a finding that the practices didn’t exist, and it’s worth being precise about the difference.
What came next made the question harder to dismiss. In late April 2026, Instructure disclosed a breach of Canvas claimed by an extortion group, affecting institutions worldwide in the middle of final exams. The company reached an agreement with the attackers and later confirmed a second intrusion through a separate vulnerability. By mid-May, more than two dozen federal lawsuits had been filed. The breach didn’t settle the argument about what Instructure collects, but it did make plain how much sits in one place, and how little anyone inside a locked-in institution can do when that place fails.
The broader picture isn’t better. In 2022, Internet Safety Labs tested more than 1,300 apps recommended or required by a random sample of 663 American schools and found that 96 percent shared student data with third parties, 78 percent of the time with advertising or analytics companies, usually without the knowledge of the school. The free tools aren’t exempt, either. Before a July 2020 policy change, Zoom’s education product pages allowed third-party advertising cookies, a practice the Parent Coalition for Student Privacy and Common Sense Media both flagged during the spring when every classroom in the country was suddenly running on it.
AI as accelerator
Martin Weller, writing on The Ed Techie in late 2024, argued that AI turns the pattern from a platform problem into a life problem. His case is that ordinary services get downgraded to an AI-by-default version, and that the things people currently take for granted, including talking to a human, get sold back as the premium tier. What makes AI different, in his account, is the combination of good-enough performance and low cost: it can be substituted almost anywhere without losing the customer base, and once everyone has done it there’s nowhere left to go. He calls it an engine for the process, and the phrase has stuck with me.
In education, the companies building these systems have positioned themselves as partners rather than vendors. In July 2025 the American Federation of Teachers launched a National Academy for AI Instruction with $23 million from Microsoft, OpenAI, and Anthropic. The criticism was immediate and predictable: one reviewer called the funding a way for the companies to make inroads into the education market, and the objection from some educators was that a union exists to protect its members’ autonomy, not to broker their adoption of a product. Whether the academy turns out to be a guardrail or a sales channel is still an open question, though the structure of the arrangement isn’t.
Does the sustainability defense hold up?
When platforms defend these moves, the argument is always financial necessity, and the record doesn’t support it. edX was sold in the name of sustainability and the buyer went bankrupt. Coursera paywalled its courses while holding hundreds of millions in cash. Quizlet locked its core features at the peak of its popularity. In each case the free tier did the work of building a locked-in user base, and the paywall arrived once leaving had become expensive. That looks less like a company running out of money than like the third stage of the pattern, on schedule.
Khan Academy is the exception that clarifies the rule. It remains a nonprofit, funded by foundations and donors, with its core content free and without advertising. Its persistence suggests the degradation isn’t a technical or business necessity so much as what happens when venture capital and public markets are the ones asking for returns.
Living with the pattern
Recognizing the pattern doesn’t undo the asymmetry, but it changes how you make decisions inside it. For anyone choosing a platform, the questions that matter are the ones that predict stage three: who owns the company, what pressure its investors are under, whether your data and content can leave in an open format, and how long the contract locks you in. A free tool with no visible business model is a tool that hasn’t reached its paywall yet. For anyone already stuck with a platform, the work is to keep your materials in formats you control outside it, to avoid building anything on features that could vanish behind a premium tier, and to document what depends on what, so that when the price doubles or the vendor is acquired you have a map instead of a panic. And for those of us who train other people, the transferable lesson is to teach the pattern itself. Platform-specific skills lose their value on the platform’s schedule. The ability to recognize when a tool has started to turn doesn’t.
None of that is a solution. It’s a way of staying clear-eyed inside a system that counts on you not paying attention. The question isn’t whether the pattern will continue. It’s whether we recognize it, name it, and keep building the alternatives that put learning first, even knowing most of them will be smaller than what they replace.
If you’ve watched a tool you depend on start to turn, I’d like to hear about it. You can reach me at licht.education@gmail.com, and there are more tools, articles, and resources at bradylicht.com.
