“Computers’re bringing about a situation that’s like the invention of harmony. Sub-routines are like chords. No one would think of keeping a chord to himself. You’d give it to anyone who wanted it.”
– John Cage, in the 1969 essay “Art and Technology”
“We see a future where intelligence is a utility, like electricity or water, and people buy it from us on a meter.”
– Sam Altman, in a comment during BlackRock’s 2026 U.S. Infrastructure Summit
Ian here—
This post began as a short housekeeping announcement but evolved into something more grandiose. (Though still attached to a housekeeping announcement.)
In the Spring of 2025 I taught a course titled “Closure and Infinity in Digital Media,” about the creation of theoretically infinite art. It opened in the 1960s with the process-based art of John Cage and the Fluxus movement, and worked its way to the contemporary dream of entertainment CEOs everywhere, selling AI slop to undiscerning viewers. The course was admittedly too broad in its focus (as evinced by its mouthful of a title). It took me to nearly the end of the quarter to actually grasp what its core concern had been all along: It was a course about the internet that every person in the room had grown up with, which was now dying.
You can draw a direct line from the sentiment expressed by Cage in the above quote—that computers will bring about an epoch of sharing comparable to the invention of harmony—to the utopian dreams of Wikipedia, open-access academic publishing, Reddit, TVTropes, even the humble user-submitted ASCII game walkthroughs still lurking in the bowels of IGN and Gamespot. That is to say, to the internet, as people idealistically conceptualized it in the 2000s, and which is still around today, though perhaps hanging by more of a thread than we’d like to admit.
Granted, there were skeptics of this vision of the internet. Way back in 2000, Tiziana Terranova dutifully tut-tutted the fiery optimism of the “Web 2.0” era, pointing out that capitalists were always waiting in the wings to monetize the free labor of credulous volunteers. The utopian venture of the world wide web relied on an unstable ideological union of hippy-dippy communitarianism and hardened anarcho-capitalism. It wasn’t built to last, because “free” didn’t mean the same thing to all parties.
But still, during its heyday, it was a grand thing to witness. Wikipedia’s 25 years of operation have stretched over my high school education, undergraduate education, graduate education, and now a decade of me being an educator myself. Throughout all of this, it has been not only an invaluable research tool, but also probably the best available model of socialism in action. (From each according to their research abilities, to each according to their research needs.) It’s no stretch to say that my path toward becoming an educator owes much to the general shape of the internet circa 2000–2015. It seemed like our shared intellectual commons was improving. Information was becoming freer every day. Professors were expanding beyond the classroom and academic publishing to become bloggers, posters, podcasters, and video essayists. Educators were rediscovering public communication, sharing their love of their subject matter more broadly than ever before. I was benefitting from it, and I wanted to contribute to it. It was a tantalizing moment to get caught up in.
And now LLMs are on track to end it decisively.
This past summer has seen some good reporting on the risk LLMs pose to human knowledge creation. Documents unsealed as part of the ongoing New York Times vs OpenAI case reveal that the people creating these technologies knew from the outset that they were a “product that destroys its supply chain”—i.e., destroys the ability of the people who authored its training materials to continue making a living. Amazon, once best known as a storefront for purchasing books, is now systematically hoovering up and destroying rare books for the purposes of training its AI. One could hardly hope for better headlines when it comes to pointing out the overall trajectory of this technology.
Yet even before these details became public over the past few months, it was evident that the product OpenAI and Anthropic offered was fundamentally at odds with the internet as it has existed for generations prior. Twenty years ago, if you had a question, you would search the internet. Hopefully, you’d find a helpful answer, but most likely you’d also find a community teeming with people who were passionate and ornery and who had opinions you perhaps had never encountered before. And maybe you would be intrigued, so you would follow a poster or subscribe to someone’s blog or start contributing to a wiki yourself. And the world would become richer for having your voice in it.
An LLM is a way to circumvent that process. If you have a question and ask a chatbot about it, it gives you an answer, strokes your ego, and disinvites further inquiry. The processes of knowledge-creation, and the community of people they involve, remain behind the curtain. Rather than an invitation to join into the cacophonous, sometimes discordant choir of voices that made up the internet in the lineage of Cage, you are encouraged to see knowledge as Sam Altman does: a utility, delivered to your screen like water to your tap after paying your monthly token bill to OpenAI, minus any understanding of the discourse involved in producing it.
Much has been made about the dangers LLMs pose to the economics of knowledge work: the journalists being laid off, the bloggers having their traffic strangled by Google’s AI Overview, law students finding the bottom rungs of their profession have been sheered clean off. Those economic concerns are real, but they’re not even the whole picture. The internet has always been built on free labor at least as much as paid labor: Not just book authors but also Wikipedia editors, not just travel writers but also Yelp reviewers, not just career Substackers but also anonymous Stack Overflow contributors. This labor offered no material renumeration, but it offered the pride of having one’s voice heard, prestige within a volunteer community, or the simple satisfaction of feeling you had made the commons a better place. My worry about LLMs goes beyond the economic problems that could theoretically be solved by a universal basic income. I fear that, by the process of knowledge-creation behind the curtain, by hiding the messy reality of research and debate behind the visage of a friendly chatbot, we’re sapping people’s basic curiosity, their desire to contribute to the conversation, to share their observations and join into debates.
No one’s paying me to write this blog—in fact, I pay Automattic an annual fee to keep it online ad-free. When I post a walkthrough for the game The Painscreek Killings, I get nothing from that but the satisfaction that, based on page views, over 200,000 people have benefitted from it over the years. But what motivation to I have to post a walkthrough once I know no one will ever see it, only see its details regurgitated by Claude or ChatGPT? Why would I post a syllabus, knowing that instead of being viewed by genuinely interested educators or students, that it’s instead going to be digested and regurgitated as the citations page for a term paper a student generated with AI? At a certain point, paranoia overtakes the natural generosity of the educator. All those materials we put online in the 2000s and 2010s, everything we posted in hope of fostering some grand new School of Athens that was open to everyone—was that all just a sucker’s game, getting us to translate knowledge into a form more easily accessible for AI training? In the age of Polymarket and Kalshi, are you a sucker for sharing any information with the public? Why not keep information to yourself—out of the hands of the prediction markets, out of the training data of the LLMs—and make a pretty penny on it? The utopian ideals of Web 2.0 were perhaps naïvely high-trust, but this new low-trust tech moment we’ve stumble into feels disastrously antisocial.
So that’s what it feels like being an educator on the internet in the current moment. What does it feel like being an educator in the classroom?
As the old saying goes, college instructors don’t assign essays because the world needs more essays on Plato’s allegory of the cave, or Foucault’s ideas about Bentham’s panopticon, or Bazin’s insistence that a photograph is ontological identical with the object it was taken of. There wasn’t some scarcity of essays that college students were very slowly redressing, that A.I. has now solved forever as it launches us into a new golden age of rhetorical abundance. The purpose of having a student write an essay is that it historically provided an opportunity for them to step into a designated knowledge domain, assess the facts, assess the multiple possible interpretations of the facts, and then persuasively argue for a particular intrepretation, while also imaginatively occupying another position (or positions) for the purposes of arguing against it and anticipating potential rebuttals. There is a long and storied tradition of ideas about how the essay, when properly employed within the broader liberal arts tradition, prepares students for the responsibilities of being an informed citizen in a democracy, able to distinguish between reliable versus unreliable sources of information and meaningfully deliberate the issues of the day.
(No, it doesn’t make much sense when you think about it. If we genuinely live in a democracy, where it’s vital to have an informed population of sharp critical thinkers, and a college-level liberal arts curriculum is how you get there, then why is college not free and easily accessible to all who are interested? Why does it instead resemble a high-priced credentialing service for the ruling class? Much to consider, certainly! But to repeat myself: in a prior era of tech optimism—and optimism about our democracy—it was psychologically easier to square this circle.)
Much as it would be a mistake to solely worry about the economic impacts A.I. is having on traffic-starved webpages, it would also be a mistake to solely lament the loss of the term paper as an assignment type. The problem cannot be reduced to just one headache for college faculty, or one particular career obstacle for journalists. The problem is the long-term erosion of incentives. The more students conceptualize intelligence as a utility, something delivered to you as a magical revelation at the push of a button, the less value they have for the skills of research, interpretation, skeptical inquiry, and persuasion. And despite what the A.I. maximalists say, I don’t think those are skills we can afford to let atrophy.
This erosion of incentives does not stop at term papers. In cinema and media studies, the tide has actually been turning against the traditional term paper for awhile now, since well before the current LLM crisis. Why have students write a traditional paper, the thinking went, when you could have them write a blog post peppered with dozens of still frames, mirroring the mode of close visual analysis David Bordwell was doing over on his blog? Why not have them do a video essay, pointing to Kogonada or Every Frame a Painting as models? Instead of preparing students for some archaic role of like a newspaper film critic, why not help them build the content-creation skills needed for a new online media ecosystem? Prepare them for a career making content on YouTube, or writing at The A.V. Club, Birth.Movies.Death, io9, or Polygon. My friend Hannah Frank assigned her students to create Buzzfeed quizzes about the movies they watched.
Go ahead: ask Claude how many of the above-named sites have been bought, sold, closed, stripped for parts, or otherwise experienced massive layoffs since their heydays. So now we’re back to square one. Why have students write blog posts or make video essays, when both blogging and long-form online video are rapidly becoming as niche and passé as newspaper columns in the 2010s? Sure, we could have students make short-form video content, instead … but at a certain point you run out of runway. Any assignment type broadly modeled off of “what people do for fun and profit on the internet” will have to reckon with the fact that an internet built by and for human beings is slipping further into the rearview mirror with each passing day.
All of this is to say that yes, a pedagogical crisis is hitting media studies at the same time as it is every other academic field, but there are also some subtle distinctions at play. Like everyone else, we’re scrambling to come up with new modes of AI-resistant assessment. But as we schedule oral exams for the first time in our teaching history, we’re not just dealing with the annoyance of not being able to use Canvas anymore. There are the additional pains of the field reaching middle age and losing its sense of cool. Just a few short years ago, we were the teachers with the fun and relevant assignments. We pointed our students to video editing suites and online content management systems, and could legitimately say “These are the tools you’ll be using if you pursue a media career. Our job is to get you comfortable with them, right now.” Now we joined the scrum of the other humanities faculty, passing out blue books alongside English and philosophy professors who have never once felt cool in their lives, all of us desperately slapping technology out of our students’ hands. (Technology that every employer that they encounter from here on out is going to insist that they use every waking moment.) The irrelevancy stings so much because it’s such a sudden, recent shift, like staring down your first grey hairs. We were techno-optimists, damn it. We believed in the future. Now even expressing that vision of the future makes us look old.
Which brings me, finally, to the housekeeping announcement.
I haven’t asked students post to this blog since the spring of 2024. For the past two years, the pedagogical risks of the blog post assignment type have outweighed the rewards. I’ve had students give in-class presentations on games that are simple regurgitations of analyses that either I’ve posted here or my former students have posted here. It’s something I wouldn’t mind at all if they mentioned the blog as inspiration. (Hooray for fostering a scholarly community that persists beyond the bounds of a single class!) But they don’t, likely because they aren’t aware of it at all. They asked an LLM about an indie game with a small profile, and the LLM spat out things said by previous students and by me, because the majority of words written on the internet about that game were right here, on this blog. My students and I had trained the tool that future cohorts would cheat with. I was watching orobouros in real time.
I can’t say for certain that I’ll never use blog posts as an assignment type again. But I am planning to re-orient this site away from student work and back toward my own output. For the past year I’ve been working on a writing project—one I’ve decided would make a poor fit for the video essay format—so you can expect some writing on here connected to that, in lieu of the video essays I’ve been periodically dropping over the past several years. I also intend to upload more syllabi, post more lesson plans, the sort of thing I used to do more often. I’ve gradually let some areas of this blog get terribly out-of-date. Everything I post will of course also be used to further train LLMs, but if I stop posting things to the internet altogether all I’d be doing is hastening the demise of the thing I’ve spent much of this blog post eulogizing.
That said, I’m also going to gradually take down the majority of old student posts from this blog.
This is a decision I did not come to lightly. In fact, I agonized over it for some time. In Winter 2026, I taught a course on the history of indie games, and the link rot I discovered when plumbing the indie games blogosphere of the late 2000s and early 2010s was thoroughly disheartening. At the peak of my own blogging here I would link to other people’s blogs all the time, and I’ve been gutted going back to old posts and seeing how few of those links still actually work. I’m trying to gradually replace the dead links with ones going to Archive.org (for as long as we can count on that, at least). It feels terrible, as I work on that long-term upkeep project, to further contribute to entropic decay of a slop-swarmed web by deleting posts.
But I’ve finally settled on my decision. My broad desire to continue propping up the internet that was has been outweighed in this specific instance by my belief in a right to be forgotten. Aside from instances where trans students posted things under their deadnames—which I’ll be prioritizing for deletion in ongoing coming process—I just don’t think that people need or want to have writing they did for a class one time available to the world in perpetuity. It’s one thing if it’s good writing: writing that I remain really proud of them for, and think that they should continue to be proud of as well. I will keep some hand-picked posts up if I think they still reflect well on the students who made them. But posts that were just garden-variety schoolwork, or posts with busted formatting because students hotlinked images from another source no longer online … those I’ll be clearing out.
So that’s my housekeeping announcement. And look, it only took 3000 words of doomerist pontificating to get there! Expect changes to the blog coming soon.
