This is one of talks from the CHCI panel, "The Humanities We Can Build Right Now" in Banff, Alberta. My explanation of the context and my introduction to the panel are here.
**
Thanks for having me, folks. I had been planning to devote my short presentation to an outline of the crisis facing STEM fields, and how understanding that crisis can make a difference for how we think about the assaults on the humanities. I’d be happy to talk about that in our conversation after this panel. But after listening to Professor Alondra Nelson’s fantastic opening keynote on Monday, I found myself wanting to pick up on her insights, and run with them a little bit.
As you recall, Professor Nelson walked us through a distinction between stochastic and epistemological frames for knowledge, where the stochastic names the random, unpredictable play of the material world, which nonetheless always holds out the promise of the predictability of phenomena, while epistemological language holds out a different promise, the knowledge of how and why things work the way they do. And she pointed us to the idea of agnotology, which if I followed right, describes both the production of obfuscatory anti-knowledge, and study of that obfuscation.
You probably also recall that Professor Nelson deftly unpacked the cynical production of ignorance among tech capitalists as they pivot between a masters-of-the-universe epistemology deployed in pitches to potential investors, and a disingenuous, shoulder-shrugging powerlessness when they tell the general public that the sheer stochastic scale of AI makes it a black box.
What I want to suggest in my presentation is that agnotology is not only about the production of ignorance, but about deskilling labor. I will also suggest that this deskilling is an old capitalist strategy – indeed it is as old as capitalist machinery itself. This is probably a familiar argument to many of you. I will add five things to it.
First: deskilling is never absolute, and the relative forms it takes reveal an important capitalist dynamic, which is that inter-capitalist competition takes the form of pitting different kinds of labor against each other.
Second: the epistemological conditions created by this shell game make many things not seem like products of capitalism at all.
Third: we can schematize this epistemological problem as the reduction of capitalism to a mere category of social life, instead of a living, breathing reproductive project that demands not categorization but conceptualization.
Fourth: we have the philosophical tools to develop this concept, and they’re not that hard to work with.
Fifth and finally: our training as humanists can help us understand capital as a moving concept because our experience with aesthetic objects invites us to see in multiple registers at once, but also to recognize how those registers form living, ever-shifting wholes.
I begin with a belief in Marx’s law of the tendency of the rate of capitalist profit to fall, because this is where deskilling begins. Capitalists do not like the rate of profit to fall. They will therefore do anything they can to each other, through the medium of us, to counter that tendency.
In this regard, stochastic and probabilistic strategies for making us work harder and consume more are especially useful to capital because they promise to make up in sheer volume what droops and sags in terms of the rate of return on investment.
As the information scholar Justin Joque points out, long before the statisticians formalized the stochastic techniques that now drive large language models, capitalists had stumbled upon those techniques and developed them without a specific name for what they were doing, like the character in Molière’s play The Bourgeois Gentleman who exclaims, “Good heavens! For more than forty years I have been speaking prose without knowing it!” Joque shows how, in giving capitalists the gift of a formalized and replicable method for their two centuries of ad hoc discoveries, statisticians, data scientists and their fellow travelers have once again reignited the dream of what the business press currently calls “zero marginal cost”: the dream of a practice in which the effort to realize profit approaches zero because, at scale, it only take a few hits to make the minimal costs worth it. The catch, of course, is that only a very small number of firms can operate at the scale required to achieve anything even approaching that dream.
There is a brutal math behind this: one that keeps capitalists in a low-grade panic, and one whose costs they pass on to each other by way of passing them on to us. The reason capitalists place hope in mechanizing their labor processes is because they see it as a way to compensate for a competition-driven fall in prices per unit with the ability to sell more units.
Marx describes this as an inverse ratio between the rate and the mass of profit: if one goes up, the other must go down. What we now call stochastics is an attempt to avoid the irreducibly zero-sum character of capital, which despite its compulsion to expand forever can only expand through rounds of destruction. Think of it as a game of musical chairs or hot potato, in which every capitalist is trying to fob off the costs of business on all the others, with the main weapon being one version or another of cheapening labor. Think of it as an attempt to be infinite and zero-sum at the same time.
This strategy – and this problem – is as old as capitalism, though capitalists’ attempts to leverage the tech sector into the yellow brick road out of the 2008 crisis has made it freshly visible in the capitalist core. The historian Henry Snow refers to it as the drive to make us all “stochastically waged.”
Because big tech has been tasked with getting deindustrialized capital out of its late-20th century messes, it is especially good at describing the dream of the fully stochastic wage, at least in its own sector. The brilliant communist geographer and independent scholar Phil A. Neel points us to a good example from Lukas Biewald, who founded CrowdFlower, a company that paid people in the 2010s to label data:
Before the Internet, it would be really difficult to find someone, sit them down for ten minutes and get them to work for you, and then fire them after those ten minutes. But with technology, you can actually find them, pay them the tiny amount of money, and then get rid of them when you don’t need them anymore.
And here’s how stochastic labor looks from a proletarian perspective, in this case represented by Neel, himself who has worked training AI to supplement his income:
If a worker were to walk by at [any given] moment and politely ask me what I was doing, my response would be frank: “I don’t fucking know.” The first lesson in AI work is dealing with an unknowing, persistent dissonance: complete this task for no apparent reason according to these hyper-specific (but continually updated and often contradictory) instructions; you are an employee but not an employee ...your wage can change at any moment, surging and dipping by as much as $20 an hour; you rely on the work for a major chunk of income, but it is not your “job.”
Neel’s description of his tasking is darkly hilarious: he spends a day asking the model questions; another day answering its questions. He is told to wear a GoPro camera and record audio in crowded places. He is warned never to utter an identifying name, to cover his tattoos, to look the other way if someone approaches. He writes that the training document he was given “gives arcane guidelines that read like dark age superstitions.”
The thrust of Neel’s argument is not that we are leaving capital behind and entering an era of “techno-feudalism,” however, but that his work conditions are the latest iteration of an old capitalist class deskilling tactic. The historian of science Matteo Pasquinelli puts it this way: “the current paradigm of AI – deep learning ... emerged not from theories of cognition ... but from contested experiments to automate the labour of perception, or pattern recognition” (page 21). The tactic, it turns out, it to turn jobs into tasks, and to stitch them together or separate them as needed. This is as true of the working day as it is of the tasks that populate it. And it is also true of the global recipe for profitability, which must necessarily include as wide a variety of types of labor as possible.
I said I would translate this into philosophical terms, so here goes capital moves according to the logic of essence and appearance. When Marx discovered this, he was leaning on Hegel, who was following Aristotle, knew this logic very well. Just as there is not a creature walking the earth called “the animal,” Aristotle points out, but only animals, capital cannot appear in any one way. It is abstract in its essence, but this is because it is endlessly appearing in material form. It has to.
This is why the theorist Beverley Best describes the activity of capitalist accumulation as a dynamic of “social averages and tipping points,” and why she has coined the term “perceptual physics” to describe our experience of this ever-shifting movement. The sheer variety of tactics capitalist classes have developed to accomplish their one task can make so much of our experience feel like it’s not exploitation or capitalism at all; it can feel like a side hustle, where your boss says he’s not your boss, that you’re not an employee, and that the money you make after all the fees and penalties is not a wage. Or it can feel so overwhelming, like the genocide in Gaza, that grasping it through the concept of capital seems like an unhelpful dogmatism, at least until you look at the artists’ renderings of the Dubai-like “Gaza 2035” commissioned by the Netanyahu government, dotted with electric vehicles and innovation labs, or until you read about the important role that anticommunism played in the British and American support for Zionism leading up to the Balfour Declaration, and onward to the founding of the Israeli state, because Zionism was seen as a bulwark against Jewish socialism.
This combination of infinite variety and relentless sameness creates a schizophrenic epistemological situation in the liberal academy. On the one hand, liberals will treat Marxists as though they only think categorically: oh blah blah blah, I get it, phenomenon X goes in the capitalism bucket, so does everything, do you have anything new to say? On the other hand, liberal discourse will keep the real meat of Marxist analysis, which is not the category but the concept of capital, out of view because it feels too overwhelming to confront. By construing Marxists as categorical thinkers, liberals can pride themselves on advocating pluralism, nuance, and the sensuous variety of the particular, and dismiss Marxism as a monological discourse that flattens all that beautiful humanities particularity.
There are complicated historical reasons we are in this strange situation, where a category of people devoted to truth-seeking (humanities scholars) do not even seem to realize that they have cut themselves off from strong conceptual knowledge of the objective conditions that shape our lives and link our lives to countless others.
But one way to name this complex of reasons is to identify it as the long shadow of anticommunism. Chris Hedges and Samuel Moyn have detailed, respectively, liberal purges of leftists from intellectual leadership in the academy, the arts, and religious spaces, and from the humanistic disciplines. John McCumber has meticulously documented the role of anticommunist politics in shaping the historical arc of a former anchor discipline, philosophy, which between the beginning and the middle of the 20th century mutated from the practice most suited for questions about justice and the good life to a technical formalism focused on the possibility of determining the conditions for the validity of sentences. And if you came of age in the (anti-) humanist academy of the 90s you know the puzzlement masking a contempt in the postmodernist descriptions of Marxism as “totalizing.”
But the contempt the puzzlement was masking was itself masking a fear, hardly conscious at all. The story of American anticommunism isn’t only of intellectual arrogance or post-war capitalist triumphalism or post-Stalinist bitterness or post-1968 disillusionment. It’s the story of fear, fear that McCarthyism, however defeated, lurks dormant in our institutions. The University of California still requires newly hired faculty to sign an anticommunist loyalty oath, which I signed in 1996 because I wanted a job.
It’s also the fear that beyond American McCarthyism lies far worse, namely the anticommunist violence the Dulles brothers perfected abroad, long after McCarthy, a bloody campaign that Vincent Bevins compasses in The Jakarta Method, in which the murder of nearly a million Indonesians provided the formula for mass murder in Chile, for torture techniques and assassinations in the Congo, in El Salvador, in Guyana, and on and on. Every post-colonial attempt to launch an alternative to subordinating integration into global capital was rerouted, delegitimized, and terrorized out of existence. But because of our collective miseducation, liberals look at the global rise of the right and think, what is wrong with these people?
So I want to close by returning the term of art to which our brilliant keynote speaker introduced us, agnotology. I want to suggest that we are constantly at risk of practicing it on ourselves. Reducing capitalism to just a word that describes our terrible world is a kind of unnecessary self-hobbling in which capital becomes a Kantian thing-in-itself, at once all too familiar and utterly unknowable. But as Best points out, Marx himself felt that the complexity of capital’s shapeshifting was graspable by everyday people; as she writes, “Marx bet the farm on the potential for such popular cognitive work, in aid of which his own analysis was intended” (page 114).
This is a humanistic project – or it should be. As we know, the tech lords are fond of theorizing about the character, the fate, and the destiny of the human. We have abundant tools with which to reply to their bloviating. The human stares back at us, evolving, contradictory, somehow singular and universal at once, in both the art we make and the tools we use. This is why, once we get hold of the moving, contradictory character of capital, singular and general and universal by turns, we can arrive at a kindred kind of conceptual clarity around the aesthetic as well – or vice versa.
Late in his life, Georg Lukács argued that we can attain this clarity when we reflect on the thunderclap of aesthetic pleasure and on the dynamics of capital alike: it turns out that the grisly thing and the beautiful thing share a logic, a “unity of substance” born of the endless interplay of singularity and generality and universality. We can use either one to get better at understanding the other. This conceptual thinking is the source for a more objective and honest collective pursuit of the meanings of the human than anything that Silicon Valley will ever create. We cannot let them seize that.
0 comments:
Join the Conversation
Note: Firefox is occasionally incompatible with our comments section. We apologize for the inconvenience.