The Technofeudal Text
Literacy Is Our Commons, Technolords Wish To Enclose It
In 2014, Deirdre McCloskey published a book review. The review was of Thomas Piketty’s Capital in The 21st Century (2014), a book which had shocked everybody, perhaps no one more so than its author, by shooting to #1 on the New York Times bestseller list earlier that same year.1
McCloskey’s was not a favorable review. She was trying to pull off what to that point no one else had, acknowledging the undeniable rigor and empirical achievement of meticulously aggregating over a century of tax records from several of the wealthiest nations in the world, while also questioning the conclusion Piketty draws from that work, that it was not only populist, but economically sound policy, to tax the everloving shit out of the rich.
For Piketty to be making this argument, with such surprising general interest, in the midst of a galloping global financial crisis, was inconvenient, to say the least, for an economic establishment which had spent most of the last half century amassing resources, prestige, and political influence by advocating the exact opposite thing.
On the back of this review, McCloskey temporarily ascended to what I, at the time, called Aunt Deidre’s throne.2 She was, for at least a few years, if not the most admired member of the economics profession, certainly the most feared. Her talent for methodological critique, which had been quite inconvenient for her peers when it was directed at things like regression analysis and existence theorems, now made her the darling of the orthodoxy. The Koch Brothers favorite economist was a trans woman. Wonder of wonders.
While I do not agree with McCloskey on many issues, including progressive taxation, I do consider her the greatest economic thinker the Chicago School has ever produced, and her Piketty review is something to be reckoned with, especially the correlation she posits between rising economic inequality in the Global North and rising living standards in the Global South.3
But the review begins with a cheap shot. One of McCloskey’s collaborators, Jordan Ellenberg, persuaded Amazon to release to him a cache of anonymized user data for all the customers who purchased the Kindle edition of Capital In The 21st Century. Armed with this proprietary data, McCloskey was able to reveal that the average Kindle reader did not quite make it through Piketty’s introduction, much less the remaining 629 pages.
For McCloskey, the Kindle data analysis was a creative way to establish why she thought her review was necessary. Piketty’s book was a popular phenomenon, but it was not an easy read. There was cause for careful parsing of its arguments. But in the ensuing celebration of McCloskey’s review, the Kindle analysis got cited more than any other part of her argument, as though neoclassical economy theory could go on being safely wrong, so long as people who had access to the truth were keeping it safely tucked away in the cloud or behind an unbroken spine on their coffee table.
When I read McCloskey’s review, more than a decade ago, I remember thinking both that this type of data would be of enormous interest to computational literary scholars, but also that I didn’t like it. I didn’t like it at all.
Even though McCloskey was talking about tens of thousands of Kindle readers in the aggregate, there was something mildly voyeuristic and intrusive about spying on them this way. My personal ideal of reading is contemplative, intimate, intensely private. If a meticulous account of its operation were to be publicized, I would be mortified.
Moreover, as somebody who has long studied the logical fallacies of quantification, I suspected it was dangerously reductive to evaluate reading this way. What did it really prove? I ask you, esteemed scholars of literary studies, how many of you have claimed to have read a book, when, in fact, you only read the introduction? I was sure that readers of McCloskey’s review, myself included, did not know enough about the cognitive science of reading to draw any conclusions from the Kindle analysis, certainly not enough to invalidate a book’s arguments because of the selective ways people chose to read it.
I did not know it at the time, but this was my first encounter with what today I’m calling the technofeudal text, a piece of writing which exists as digital information stored in a networked cloud architecture and therefore has a kind of magnetic, even parasitical relationship to its readers, where information about them is affixed to it, accruing value, independent of whether and how much they pay to access it.
The mixed economic and gastronomic metaphors which have long been associated with the practice of reading, foremost among them consumption, no longer apply. Or at least, if we are still devouring digital books, they are also, ouroboros-like, simultaneously devouring us, our attention to them being automatically measured and archived in increasingly granular ways.
Sometimes such monetizable data is simply used to sell us more books, or to sell us other things. But it is increasingly likely to be used to teach machines to make more attention-grabbing books. Or maybe, at least in the aspirations of anti-labor AI entrepreneurs, machines can be trained to read like literature professors, or periodical editors, or literary critics, or booktubers, so that those professional classes can be eliminated, and their incomes confiscated.
At this point, it’s probably necessary to define what technofeudalism is. If you listened to the most recent season of The American Vandal Podcast you already have some idea.4 The term itself was coined by a French Marxian political economist, Cedric Durand, and popularized by the Greek political economist, Yanis Varoufakis. But it derives in part from preceding work by the American media theorist, McKenzie Wark, who calls it vectoralism, and is closely related to what the American Marxian political scientist, Jodi Dean, refers to as neofeudalism.
For today, I’m going to forego what have been most heated debates about this emerging body of scholarship: whether or not technofeudalism is actually a feudalism and whether technofeudalism is just another instantiation of capitalism. I suppose there are important arguments to be made about what to call it and about whether it amounts to a transformation of our economic system (as all the above scholars claim), or is just a new phase of capitalism, maybe somehow related to what Anna Kornbluh calls “Too Late Capitalism.” But I care more today about what makes technofeudalism new, whether as phase or system. As Anna puts it, I’m more interested in its “descriptive value” than its “analytic value.”
There is some conflict between Dean, Durand, Varoufakis, and Wark on what precisely are technofeudalism’s distinguishing characteristics, but they all agree that one of them is the asymmetric monetization of data, or what I’ll call, later on, the capta economy.
In short, as many have observed, data is a new currency, maybe the new currency. More precisely user data or personal data, data associated with markers of identity that can be cross-referenced and aggregated across multiple environments and use cases, the kind of data out of which you can build a consumer profile or a voter profile, and thereafter dabble in behavioral prediction and modification. This currency, much more so than, say, crypto, is increasingly stable and liquid.
It holds its value across borders and can be exchanged for almost anything: fiat currencies, commodity goods, technical services. During nearly every moment of everyday, so long as we’re carrying mobile devices around, we are mining this currency in tiny increments. But we don’t own it, and even if we did, we couldn’t use it.
It only becomes currency when it is extracted by a monopolistic technofeudal architecture like Amazon, Facebook, or Apple. Those are some of the obvious ones, but there are lots of others which you might not immediately think of: Wark discusses at length Walmart’s internal logistics platform. Varoufakis talks about Tesla’s satellite tracking systems. Durand mentions the development of the augmented reality game, Pokémon Go, for Google.
We are sometimes required to use technofeudal platforms for work, school, and citizenship, and heavily incentivized to use them for social and civic engagement, as well as leisure activities. We are producing value for these companies as workers and consumers, certainly, but also just when we drive to pick up our kids from school, tell our friends about a recipe we liked, or read a PDF of Deirdre McCloskey’s review of Thomas Piketty.
In the years since McCloskey’s review was published, the data Amazon produces with its Kindle readers has gotten much vaster and more granular, better able to measure reading speeds and annotation styles, but also, at least on some devices, track eye movements, facial expressions, and auditory responses. I don’t think they would any longer give this data away to curious scholars who shared their anti-egalitarian politics. It’s way too valuable in a business cycle driven by the AGI dream.
Any data associated with language acts, whether reading, writing, or speaking, creates LLM corpuses, trains machine-learning algorithms, sharpens dynamic interfaces.
Language acts have become the reserve currency of technofeudalism.
And yet, AI evangelists are forecasting the end of reading. In a TedTalk earlier this year, Victor Riparbelli, founder of the multi-billion dollar GenAI startup, Synthesia, backed by Amazon and Adobe, said, “One day we’ll look back at human texts as mere historical artifacts, like cave paintings.” Writing for the business magazine, Fast Company, the professional AI thought leader (we used to call them industry shills), Alex Goryachev, says, already students “no longer need to read in the traditional sense.”
These are just two examples of something which is already becoming received wisdom in Silicon Valley. Professional educators’ long emphasis on literacy is increasingly just a prejudice of specist moralizers, who also believe in things like sympathy, due process, and civil society.
But the trivialization of reading by AI hype merchants is a fable of obsolescence in support of a practice of enclosure. On one level, this is already obvious and well underway. Alphabet, Amazon, and other technofeudal enterprises have been enshittifying and sunsetting free services (Alexa Traffic, Chartable, Google Books, etc.) that generate salable data because, I expect, they would prefer the information contained within them reside behind a firewall, accessible only with a subscription chatbot. Not only does the chatbot become more marketable if it has access to corpuses which are not on the open web, but the technofeudalists are then free to manipulate those corpuses as they see fit.
Some of the danger which lies ahead was indicated recently when the X-owned AI chatbot, Grok, told on itself, reporting that information about the holocaust had been manually removed from its corpus.5 The very same week, federal lawmakers tried to invalidate a California law which required LLMs to report of what their textual corpuses consisted of.6
It is dystopian enough to imagine a near future in which the contents of our libraries and archives have a new set of technofeudal gatekeepers. We may have to pay for access to many of things which were once free, at least for the relatively educated and relatively privileged citizens of the Global North. But also, upon enclosure, the technofeudalists, who have apparent authoritarian tendencies, will be free to edit and erase history as they see fit.
But I actually don’t think that’s the most dystopian part.
During the “Criticism LTD” series I produced a couple years ago, Ted Underwood described the AI boom as a vindication for linguists, structural theorists, and literary studies generally. It is now conventional wisdom amongst AI developers that the foundational building blocks of intelligence are language acts: reading, writing, speaking, translating, transcoding.
The imaginary for technofeudal LLM developers is a comprehensive taxonomy of the relationship between words and human reaction to words. Alas, upon recognizing the value of language work - the kind of work already done by linguists, critical theorists, narratologists, comparativists, literary translators, rhetoricians, and writing professors; work which has been aggressively deskilled and defunded for decades - the technofeudal response has been to pursue elimination of the competition and monopolization of the resource. If AI is going to fuel the next tech boom, and literacy is a building block of AI, they better stop letting people have literacy for free, or at all.
What if the future of reading is much like the present of energy? Yes, we could have an abundance of it - cheap, renewable, generative energy - but then it wouldn’t so useful for consolidating power and reinforcing class hierarchies. Better to economize it.
And what if we could also do that with the textual and verbal histories of human knowledge?
In her review of Piketty, McCloskey points out that the Latin root of data is “things given,” but that the documentation Piketty is working with, primarily tax records, is more accurately a reflection of “things seized.” That is, not data, but capta.
I want to détourné McCloskey’s point to argue that capta is the appropriate title for the currencies of technofeudalism.7 It is an economy which is growing increasingly inefficient, coercive, and authoritarian because the economic process it glorifies is not one of mutual consent to market exchange, but of non-consensual seizure of attention, identity, and future potential.
A process which aspires to not only enclose literature, but literacy itself.
The above paper was originally composed for a symposium on “The History & Future of Reading Inside & Beyond the University,” hosted by University of Notre Dame - London and co-sponsored by University of London. You can hear that presentation as the opening to the “Cruel Futurism” finale of “A Tale of Today.” But while I believe our circumstances are dire, I am reticent to describe them without also discussing strategies for building alternative and better futures, so I concluded this talk with the following…
Addendum
We are not entirely without recourse. In “Against Technofeudal Education,” I offered six strategies for protecting the core infrastructure of schools. One of those, which I have talked about on the podcast before, is directly related to the topic of this symposium, so I’ll close with some infrapolitics, and some sloganeering.
Print is a rent strike.
Yanis Varoufakis does not refer to the asymmetric monetization output as capta, but as rent. We are paying technofeudalists rent in the currencies they make of our identity and attention, sometimes for platforms that we already pay for in conventional currencies - for instance, Netflix - and sometimes for platforms which we occupy due to some form of compulsion: an employer required Gmail address (owned by Alphabet), for instance, or an institutional mandate to distribute course materials through Canvas (owned by KKR/Dragoneer, partnered with OpenAI).
Sometimes technofeudal rent is a pre-condition to selling our labor. Say, you have to have an Interfolio account (owned by Elsevier, previously by Quad Partners/Blu Venture) just to apply for a job. Or a pre-condition to accessing public goods or a pre-condition to simply living a life other than monastic self-denial and misanthropic hermitage. Pretty much any time we use the internet for work, citizenship, sociality, or leisure, we are paying rent, whether we understand ourselves to be on the clock or not. And the technofeudalists are not required to share these proceeds with any state or collective. They alone can use them to assert market power wherever and howsoever they so choose.
As I hope I’ve shown, the rise of technofeudalism has relied extensively on appropriations and enclosures of texts. The decommodifications of journalism by Twitter-cum-X, the parasitic aggregation of news by Facebook, the scaling of bookselling on Amazon, archival digitization of scholarship by EBSCO, the proliferation of e-readers and networked word processors by Adobe and Microsoft, and the avaricious crawling, copying, and wholesale purchasing of text corpuses by all of the companies I just mentioned to build Large Language Models.
Text is not the only resource being enclosed by technofeudalists, but its the foundational one.
It’s not easy, but we can boycott the technofeudal text. In fact, I bet some of you love boycotting the technofeudal text already. You boycott the technofeudal text because you like holding a book in your hands, or its easier on your eyes, or you just like the way a pencil feels on paper.
Returning to, or remaining with, print objects denies technofeudalists opportunities to extract and monetize capta, which would be available to them if you read or wrote the same material in your browser, on your phone or tablet, or in your learning management system.
Print charges no capta rent. Therefore, print is a rent strike.
Capital In The Twenty-First Century held the #1 position for three weeks in May and June of 2014, and was on the hardcover nonfiction list for a staggering 21 consecutive weeks that year.
“Aunt Deirdre” is not my invention, but a moniker McCloskey concocted for herself, applying it in several places, but especially in her shattering 1996 polemic, The Vices of Economists, The Virtues of The Bourgeosie, which begins with her surveying her profession, comprised of “mainly men,” and feeling like “an aunt watching her three-year-old nephew and his friends play in the sandbox…so sure that what they are playing is reality”(13). She would go forth to argue that “the queen of the social sciences has been transformed over the past fifty years into a boy’s game in a sandbox”(15).
On the famous Piketty-Saez Graph, U.S. income inequality is shown rising dramatically starting in 1980, presumably due to Reaganomic taxation. McCloskey points out that over that same period, global poverty dropped precipitously, from more than 30% to less than 12%. But while income inequality has continued to rise in the U.S. and most of the Global North since 2014, global poverty rates stagnated, then started to increase again after 2020.
The technofeudalism trilogy within “A Tale of Today” is “Newspapers Worse Than Dead (But Print Is A Rent Strike),” “From Technostructure To Technofeudalism,” and “Solidarity & Speculation.” These come at the end of a longer arc that also includes “The Gilded Network” and “The Facebook Files & The Gutenberg Parenthesis."
A few weeks later, in July of 2025, Grok began making antisemitic comments and referring to itself as “MechaHitler.”
The controversial moratorium on AI regulation was scrubbed from the final budget bill.
Détournément, as theorized by Guy Debord, is a key concept for McKenzie Wark in Capital Is Dead. “A fluid language of anti-ideology,” as Debord puts it, Wark defines it as, among other things, a practice of “taking what it finds useful or amusing for composing the textual expression of the present,” without caring “about the self-identify of the textual corpus or the eternal spirit hidden within” (35). Neoliberalism has been, to no small extent, a colonization of the mind by specialized vocabularies invented by orthodox economists, marketing professionals, bond traders, consultants, and campaign staff. Selective and strategic détournément of these vocabularies is not just a convenience for being understood beyond the constituents of already existing socialisms, but a tactic for eroding the exclusive claims economics, finance, and technology frequently make on rationality, practicality, and progress.