Chapter 7. Epilogue — Fear of mimesis
Contents · Official record (SFU Summit)
Chapter 7. Epilogue
Lobsters have a special role in the imagination of male pontificators. Famed American author David Foster Wallace, for example, exhorted his readers to Consider the Lobster in an edited collection of essays in 2004. In Twelve Rules for Life, Jordan Peterson's New York Times-bestselling book, the lobster figures as a didactic proxy for the emergence of the naturalness of hierarchy. Peterson proposes lobster behaviour tells us that dominance hierarchies are a biological fact based on the fact that humans and lobsters share a common ancestor in evolutionary history (Channel 4 News 2018).32 Peterson's argument for human flourishing is an argument for hierarchy that reduces the human to biological functions and strips the legitimacy of any sociological explanation. In this view, the social is the animal kingdom, its hierarchy (and the means of navigating this hierarchy) a brutal structure for subject composition.
Peterson's analogy crystallizes a few of the main themes within this dissertation. Throughout, I have discussed the distinction between laws and axioms. The law functions as something that cannot be evaded or exceeded, an infrastructure that guides and shapes behaviours and subjects. Axioms, on the other hand, are that from which laws emerge–and, in turn, grant privileged access to the power of a law itself. The law is the realm of the superego, the paternalistic figure: it is what must be done, under threat of some negation of agency (discipline, punishment, control) that openly signifies domination. The axiom, on the other hand, indexes the coercive power of the law by tracing itself on the very phenomena it shapes according to this exact logic of prehension. As Deleuze argues, axioms are a "treatment:" they are involved in actively determining or giving form to that to which they are applied, and in this they carry the universal subject (Deleuze 1980, 13). What have historically been questions of law (metaphysical, Christian, logical) are exalted beyond question through the logic of the axiomatic. Thus while the law is known to hew to an authority that can be named (and hailed, and questioned), the axiom does a much better job hiding itself. Yet the axiom determines what counts as value, who is a "subject," and how objects circulate–with its own host of assumptions about what is and is not automatic, or taken for granted, and thus reducible to abstract levels of analysis (ibid. 14). Hence, another contradiction: the aestheticized rebellion outlined in this dissertation, performed through mechanistic conceptions of intelligence, perverts social changes intimately tied to this cultural positioning in the last hundred-odd years in the West. To put it simply: rebellion is the flavour of domination when the very idea of thinking axiomatically is exposed and the roots of the arborescent logic are named.
In Peterson's anecdote, the human is figured as subject to laws that always exist prior to that which makes us anything more than animals; the task of the intellect in this view is to enact law in accordance with this axiom. Peterson's particular brand of axioms–Jungian archetypes, rules for life, evolutionary chains–are mobilized to make his intellectualism a transfiguration of the archetypal onto the axiomatic. This tension runs throughout the episodes and figures in this dissertation, and this ambivalence reveals how “intelligence” operates less as truth than as affective and moral authority through the appeal to rationality. As Deleuze writes, rational and moral discourses form part of the state. It is not only institutions but all those social structures that nominally exist as abstracted forms. In this sense, the state itself is axiomatic to what it is to think and to be a subject. It is the logos of thought itself under these conditions, providing "goals and paths, conduits, channels, organs, an entire organon" (Deleuze and Guattari 1987, 374). "State philosophy" broadly construed is that mode of thought that characterizes the hegemonic way of thinking about representation, its subjects, and power. More precisely, as Deleuze describes it, it is that set of ideas that legitimate the concept of the State, qua institutionalization overall.
Hence, another theme that emerges in this dissertation, extending from the question of intelligence: self-styled philosophizing. From Silicon Valley "philosopher kings" to Girard's singular grand theorizing, the expansiveness of the intellectual projects at hand aspire for philosophy in the register of the axiomatic. The simplicity of explanation, the austere concepts, and the modularized, atomized units of analysis are truth claims grounded in abstraction. Their appearance as axiomatic makes them amenable to technical adaptations as a form of common sense: they are easily understood and, in turn, easily used. This common sense even hews to similar appeals to the natural, as "a compendium of well-tried knowledge, customary beliefs, wise sayings, popular nostrums and prejudices … [i]t seems to be outside time. Indeed it may be persuasive precisely because we think of it as a product of Nature rather than of history" (Hall and O’Shea 2013, 9).
"First principles" are another name for common sense, especially when we examine how they are discussed. In my focus on the epistemology present in Silicon Valley that all decisions can (and should) be made from a set of fundamental assumptions similar to an engineering problem, I show how this form of decision-making eschews the institutional and the macrological—and, in turn, the democratic—in its need to avoid association with any form of nomadic or relational way of reasoning. The hard logicism emergent from the first principle acts as both an apparatus for shaping thought and as a means to enclose the very possibility; in the introduction of the simple binary test of whether something accords to its axiom, the possibility for thinking (at all, otherwise) is eliminated. Remember that, as Peterson writes, the golden male warrior's "culture" that he defends is that which he cannot see. The scapegoat’s position of exalted defensiveness is self-referential. He sacrifices himself on the altar of a reason that masquerades as impartial. The true threat from "post-modernism" in this scenario is the questioning of reason itself. The Enlightenment thought it found its universal value that would couch its domination under the pretense that its order belonged to all.
This logos is intimately tied to patriarchal and colonial domination, as we tracked first principles to the thought of the progenitors of neoliberalism in the Austrian School of economics. By reducing all possibilities for knowledgemaking through a rationalist, deductive focus on activity, human action itself becomes an a priori category. Rather than relying on empirical data or historical contingency, praxeology posits a universal logic where all behaviour is purposeful, constructing economic laws from internally consistent reasoning. This strict logic of calculable agency reduces the affective, the social, and the unconscious to subsets of action and makes these actions amenable to economic and technical logics. Praxeology’s humble feint of accepting the impossibility of total knowledge paradoxically requires taking up the axiomatic and its colonization of sub-elements as the only legitimate form of knowledge; the axiom’s inability to account for otherwise mobilizes its adherents to defend it in a manner ideological. Victimhood is necessary emergent to the entire endeavour—any resistance to difference is a resistance to the legitimacy of reasoning praxeologically at all. Anything coded as feminine (the flowing, liquid demos) becomes existentially threatening.
Hence the popularity of René Girard amongst those who think Austrianly. Girard himself proffers an axiomatic theory of human desire. As Smith-Riui describes it, Peterson is a dumb person’s idea of a smart person; Girard, meanwhile, is a “practically-minded person’s idea of what a theorist is like” (Smith-Ruiu 2021). The similarities between the two illustrate part of the ease at which Girard can be taken up in these contexts. While Girard appears to have the cachet of a French intellectual, the very totalizing and axiomatic nature of his theory is reflected in his aims for what theory might (and should) be: mimetic desire is, for Girard, “a single explanation for a thousand different phenomena” (Girard and Haven 2020, 79). While one of the questions of this dissertation at the outset was how the right became the left (or, rather, an attempt to explain the intermingling of the two), Girard is an exemplar of this adoption. Contra Adorno or Foucault, cynically re-appropriated by upstart intellectuals with libertarian agendas, Girard’s own attachment to theory-with-a-capital-T was, as I detailed in the chapter, tenuous at best. Yet because of this he remains a metonymic answer to my question: the right never became the left. Instead, the social became an object of interrogation and study for the right in a manner similar to this conception of the “left,” the difference being that social relations are not taken as infinitely variable but amenable to abstraction and reification. Girard’s theory of mimetic desire and his approach to it are fundamentally conservative: it is as if the theory itself, once discovered, must be held sacred against any possible barrage of advancement or impractical collaboration. While it seems to innovate beyond the concepts of Peterson or other reactionary, conservative Christian notions of human nature, then, it is in reality emblematic of the situation where the axiom confuses its own imposition with truth.
A risk that emerged writing this dissertation is that by trying to write diagnostically or argumentatively, I could fall into the same processes of fetishization or valorization via abstraction that form this type of knowledgemaking and its kin politics. I would not be alone in such attempts to reckon with chaos; they have factored into the modern project from Nietzsche to Jung through the development of information theory. Nietzsche's Birth of Tragedy discusses the necessary dichotomy of Apollonian order and Dionysian chaos; Apollo represents the principle of individuation, harmony, progress, and those things central to conceptions of logos; Dionysus, on the other hand, represents all that exceeds order: emotion, ecstasy (the transcendence of the self), unity beyond division. Yet, as N. Katharine Hayles and others have shown, chaos has been a concern not only philosophical or psychological but scientific as well since the movement beyond Newtonian physics (Hayles 1990). Newtonian mechanics, as a branch of physics, ruled understandings of the physical world up until the turn of the 20th century. The physical processes underlying classical mechanics were those that presupposed a linearity between the state of matter and its location in space. Given the state of an object at a point in time, the laws of classical mechanics dictate such that it is possible to understand how such objects will move in the future and how they have moved in the past. This understanding was formalized, most notably, by Newton's three laws of motion: axioms that present necessary and sufficient conditions for any physical object to interact with a force upon it. Henri Poincare's proofs, that Newtonian equations could not be wrestled to solve nonlinear problems, expanded on the purview of classical mechanistic attempts to understand the universe and matter away from the attempt at logical, positivist, and enclosed understandings of phenomena and towards attempts to account for and understand the role that the unknowable might play in complex systems.
This attention to the chaotic is important given the broader technolibertarian project to eliminate it that I detail in this dissertation. What is sold through computationalism and its products is not the dream of technical utopia so much as it is the dream of a technocratic management that promises to eliminate uncertainty by subjecting the world to governance by its own rules. We see this in the conception and development of "ground truths" in AI circles. A "ground truth" is, in artificial intelligence design, the reference standard or correct answer against which model predictions are compared and evaluated (Bishop 2006; Hastie, Tibshirani, and Friedman 2009). The concept encompasses labelled examples of datapoints used to train supervised learning models, reference data for measuring model performance, and validation datasets for hyperparameter tuning. From the perspective of a machine learning algorithm's ability to develop explanations and claims about the validity of its dataset, ground truths act as the references against which predictive models are evaluated and validated, and they are manually–and deliberately–chosen. What appears, then, as the technological fantasy of the removed intellect of AI is in reality necessarily and deliberately constructed. As such, they are limited, arbitrary, and relational to both those that create both the algorithms and data that constitute them (Jaton and Bowker 2020). All these forms of verification act as the "ground" upon which Large Language Models can make their claims–and are, in turn, intimately tied up with the equivocation of intelligence as captured by a mastery over logical formalism and computational modes of thought (see, for example, (McCarthy 2022)).
Ground truths are the "first principles" of machine learning, and are intimately tied to our understandings of knowledge both natural and artificial. Current waves of artificial intelligence and large language models equate intelligence with the ability to pass the Turing test. As AI ethicist Momin M. Malik writes, "no quantification captures everything; like modeling more generally, a measurement is a reduction that solidifies some meanings while excluding others" (Malik 2020, 8). Malik references Yarden Katz's work on the relationship between behaviourism in psychology–which posits that psychological knowledge can only be gained through observation of human behaviour and the antecedent causal or logical deduction of more general principles–and the idea that intelligence itself can be determined by following a causal chain, its object or engine the axiom at the beginning. Yarden Katz describes this emergence as premised “on an input-output relationship that’s learned by tuning the model using error signals in data, without worrying about any internal states that govern that relationship” (Katz 2017, 6, qtd. in Malik 2020). This is reflected in the relationship that LLMs have to their training data and the knowledge that they are capable of producing. Emily Bender et al., in an influential paper, described large language models as "stochastic parrots" that repeat discourses embedded in their own training data, equivocating coherence with intelligence. LLMs present coherency in constructing meaning in their chaining of language; while the models they create and espouse from appear to the human eye as linguistically sound (in English), Bender et al. tell us that "an LLM is a system for haphazardly stitching together sequences of linguistic forms it has observed in its vast training data, according to probabilistic information about how they combine, but without any reference to meaning" (Bender et al. 2021, 617). Hence the tendency of LLMs to regurgitate and perpetuate hegemonic (specifically conservative, patriarchal, Westernized) narratives due to the relatively unfiltered data that makes up their training processes (ibid.). The stabilizing power attributed to ground truths involve a search for self-sustaining and self-governing social organisms, emergent from a felt knowledge and mastery over the logos prescribed by an axiom. This reduction, then, is an attempt to understand the social as a computational system. In these cases, intelligence is both the engine and the model of this escape, and it excludes through hierarchies along classic colonial lines. These escapes reflect the apotheosis of developments in "the quest to pin down intelligence" and this quest's reliance on objective measuring. In their attempts to measure intelligence according to the same dictates aimed for by artificial intelligence, they produce hierarchies of human worth that extend from the history of quantifying intelligence along eugenic lines (Katz 2022). As Beth Coleman writes:
Unsupervised learning in AI reproduces the society it mirrors, which is often in the form of a system of eugenics or homophily: these are practices of sorting and prioritizing inherited from pre-computational sciences and reflected in social and political standards … It is a practice of classification and the indexing of features that has its origin in the “science” of eugenics (Coleman 2021, 3)
Social escape and current conceptions of (artificial) intelligence both seek to, and are limited to, recreating the past (or some idyllic, nostalgic interpretation thereof). Intelligence as mechanism, like the politics of no politics, cannot account for the relationality of language and, in turn, politics.
Notes
While the common ancestor traces to before the development of skeletal structures, Peterson did not explicate this in either the cited interview or 12 Rules for Life. Serotonin also has a markedly different, almost opposite effect on invertebrates than vertebrates (Chiavegatto et al. 2001; Huber et al. 1997).