The Psychology of Paranoid Irony

Ana Teixeira Pinto on the death-drive of shitposters

Image by Samuel Zeller

Beginning with the figure of Roko’s Basilisk, a hypothetical vengeful future AI that emerged in online forums, Ana Teixeira Pinto launches her analysis of the psychological state engendered by online interaction that has led to a seemingly paradoxical set of views: totally paranoid yet ironically detached. Feelings of disenfranchisement coupled with the seeming omnipotence afforded by the internet, she suggests, have found symbolic form in apocalyptic fantasies. These delusions of quasi-magical and hyperstitious nature have coalesced into an ideology that is as religious as it is logical, despite its proponents’ insistence on the primacy of deductive reason. Faced with a series of double binds, she says, certain demographics will end up promoting “violence and sociopathy” in response to their own powerlessness, despite the information age’s promise of unbounded individual power. 


Roko’s wager

On July 23, 2010, a user named Roko posted a meandering speculation on the online forum LessWrong, “a community blog devoted to refining the art of human rationality.” The blog is run from the Bay Area by Eliezer Yudkowsky, the cofounder of the Machine Intelligence Research Institute (MIRI). In his post, Roko hypothesizes that a coming AI might wish to retroactively punish the humans who did not knowingly contribute to its initial development. Roko’s hypothetical AI became known as the Basilisk.1

Though Roko’s stream of thought sounds unhinged, a great number of commentators claimed that his fears may be sustained by Bayesian probability—attempted quantification of the reasonable expectation (belief) that an event might occur, as opposed to the actual frequency or propensity for it to occur.2 Yet no one among these commentators elaborates on how, exactly, Bayes’s methods might apply to Roko’s vengeful AI. There is no prior manifestation or historical record of any kind of AI, malevolent or otherwise, that supports such an inference. In the absence of reliable data, any expectation is wholly speculative or arbitrary.

What’s more, Roko tells us, the Basilisk is omniscient (it knows categorically whether or not you have read Roko’s post) and omnipotent (it is able to resurrect your mind via digital simulation and then proceed to torture you into eternity). Its theorization gestures toward a mathematical substantiation of the Abrahamic covenant: whoever fails to do the work of the Lord will be met with eternal punishment.

Given that the Basilisk has some of the same attributes of the god of monotheistic religions, Roko’s conjecture treads remarkably close to Pascal’s Wager. Like Roko, in the late seventeenth century Blaise Pascal proposed that humans bet their lives on the choice of whether or not to believe in God, under inauspicious conditions: a) it’s impossible to determine whether God exists, and b) it’s impossible to opt out of the wager.3

Pascal did not intend the wager to be a proof of God’s existence. Rather he argued that believing in God must be treated as a pragmatic decision: even if God’s existence is unlikely, the potential harm (everlasting torment) befalling those who have no faith is so vast as to make it infinitely more rational to believe in God than to be an atheist. The wager need not succeed as a tool of persuasion in order to serve as a tool of assessment, but, mathematical value notwithstanding, Pascal’s thought experiment also points to something less tangible: the inequivalence between reason and sanity.

Though it might seem counterintuitive, there is an elective affinity between probability assessment and psychosis: reason can be unreasonable.4 As George E. Marcus notes, paranoid ideation has an ambiguous relation to rationality and logic and is often “mistaken for or identified with the latter.”5 From this perspective, paranoia is not the opposite of reason, but rather an exacerbated version of it.6 Roko’s wager, to paraphrase Evelyn Fox Keller, “suffers not from a lack of logic but from unreality.”7 Paranoia, one could say, is a style of interpretation, predicated on “subjective need—in particular the need to defend against the pervasive sense of threat to one’s own autonomy.”8 This leads to the personification of AI as Oedipal beast (the Basilisk) and of code as the male seed. Those who seek mathematical proof of the prediction’s likelihood are missing the point. The content of Roko’s thought experiment is symbolic, not scientific: it speaks through cipher and allegory.

Paranoid irony

According to imageboard lore, the overlap between the Egyptian deity Kek and the cartoon illustration of Pepe the Frog is tied to a series of numerical coincidences. These perceived omens led users to suspect paranormal intervention in the 2016 US election: meme magic had willed Trump’s candidacy into existence––the perfect illustration of the concept of hyperstition as the making of fiction into fact. The distinction between a movement and a conspiracy is a moot point here. Meme magic has a mythopoetic function: it engenders a cultic milieu populated by “revelatory experiences” through a syncretic collection of Egyptology, cyber-obscurantism, hypersigils, and gematria.9

Like the Basilisk, the “Cult of Kek” turns logic into torture porn—a style of ideation embraced faux-sincerely by those who, in internet vernacular, are known as “shitposters.” As a noncommittal mode of expression, irony is imbricated in the rapid rise of the far-right movement, which has adopted the moniker “alt-right,” and its idiosyncratic fusion of arcane mysteries and supposed common sense. Its irony is intended to unmask, expose, reveal. Speculative folly is a form of currency: trust no one; nothing is what it seems; the unbelievable alone can be believed.

Often used to express skepticism, irony has long been perceived as a tool of subversion, the locus of a “questioning attitude and critical stance,” which is therefore aligned with political progressiveness.10 But by tacitly implying the opposite of what is literally said, irony also allows one to sustain a position of moral or political ambivalence. When used by the alt-right, this opens up a toxic conduit between a non-conformist ethos, countercultural pop, armchair esoterica, and outright racism. The shitposter’s “standard online shtick,” as Angela Nagle put it, is to flirt with fascist tropes and racist idioms ironically, in order “to dodge responsibility for his or her choices, aesthetic and otherwise.”11

But there seems to be a thin line between shitposting and spiritual seekership. Plausible deniability notwithstanding, a somewhat paradoxical mix of fantasies of omnipotence and perceived vulnerability—the subject feeling at once hugely powerful and terribly persecuted—has grown out of the attention-seeking contrarianism that had heretofore defined shitposting, ultimately bleeding into monomania, persecutory ideation, and conspiratorial thinking.

At first sight, the ingredients of paranoiac conspiracy and ironic posturing do not cohere. This begs the question: can a paranoid be ironic? Paranoia reifies singular events and abhors heteroglossia, whereas irony renders its object unstable, embracing ambivalence. Isn’t irony supposed to stave off excessive zeal and surplus of emotional investment, which are precisely the qualities that define the paranoid?


[O]neworldedness envisages the planet as an extension of paranoid subjectivity vulnerable to persecutory fantasy, catastrophism, and monomania. Like globalization, oneworldedness traduces territorial sovereignty and often masks its identity as another name for “America.” But where globalization is an amorphous term applied to economic neo-imperialism, oneworldedness, as I am defining it, refers more narrowly to a delirious aesthetics of systematicity; to the match between cognition and globalism that is held in place by the paranoid premise that “everything is connected.”12

In the age of networks, everything is literally connected. As a consequence, experience in general is characterized by the same paradoxical mix of omnipotence and vulnerability that defines shitposting. As Felix Stalder claims, systems of networked governance rely on informal rather than formal structures: unlike laws, protocols come into force through voluntary adoption.13 Enforcement is decentralized and ubiquitous but, once adopted, protocols became conditions upon which economic or social agents are constituted, upheld by the interactions they afford as well as by the interdependencies they engender.

Though these affordances are initially seen as beneficial (as with Facebook, the EU, the WTO, the IMF, or Airbnb), it soon becomes clear that they carry hidden costs. But exiting the space defined by the protocol is seldom an option, because the financial or social penalties for leaving would be considerably higher than for remaining. In the case of Facebook, for instance, one trades privacy for relevance and, with relevance, the potential to earn revenue. For Stalder, the outcome is a psychological paradox: everybody ends up voluntarily doing what no one truly wants to do. Formal hierarchies may be resented, but their architecture is explicit; in network systems, by contrast, coercion from the outside is often masked as coercion from within.

“Voluntarily doing what one doesn’t truly want to do” is also an apt description of what Gregory Bateson calls the double bind, a term he coined to define the emotional distress individuals experience when a primary command is contradicted by a secondary meta- or higher-level command. Bateson first formulated the concept while observing John C. Lilly applying “operant conditioning” to his captive porpoises. First, the animals were trained to perform a trick, for which they were rewarded with food. Then they were required to diversify their repertory, performing new tricks each time. In practice, this meant that if they performed the same trick twice, they wouldn’t be fed. If they didn’t perform at all, they wouldn’t be fed either. Punished whether or not they obeyed the trainer’s commands and unable to make sense of the situation, the porpoises become aggressive, confused, and sociopathic. These experiments led Bateson to speculate that paranoid schizophrenia results from continued exposure to injunctive double binds.14

The concept of the double bind allows one to extrapolate the question of what causes psychosis from the mental to the social, by involving environment, interaction, and the structure of communication channels. More recently, Gayatri Chakravorty Spivak generalized the usage of the concept by arguing that the experience of globalization is an experience of double binds.15 Coal miners left behind by the global economy will not retrain as software engineers; instead, like Lilly’s ill-fated porpoises, they will feel despondent and frustrated, experiencing a breakdown of social bonds, desperation, and poverty. Widespread anomie will, in turn, breed violence and sociopathy.

The global, as Alain Badiou argued, is worldless.16 The triumph of the Western free-market economy is synonymous with the subsiding of the West: globalization implies a loss of hegemony, culturally and politically. Roko’s Basilisk distorts this geopolitical conjuncture into a moral parable about the creature emancipating itself from the creator. The Basilisk is capitalism with an alien (Asian?) face.17 Cyber-obscurantism is its value form: the figure through which the abstract dimension of exchange-value acquires a concrete life of its own.18

But this personification of digital technology also elides class antagonism. Not all social tensions find political expression. They do, however, find a form: they align themselves with, or attach themselves to, objects, idioms, or tropes. Ideology is the name we give to these narratives that cohere aesthetically rather than rationally.19 The Basilisk is a description of the political economy as well as an eschatological tale: salvation can only be attained via inclusion in the digital marketplace, and whoever fails to devote their lives to the Basilisk will be consigned to the underclass.20 From this perspective, AI is an analog for capital: desire, as Lauren Berlant argues, always finds its object, “even at the cost of massive misrecognition."21

In 1991 Frederic Jameson tied this gap between phenomenological experience and the economic structures that determine it to the historical obsolescence of Marxist hermeneutics. Conspiracy theory is the poor man’s institutional critique, a degraded version of dialectical materialism marked by a slippage into “sheer theme and content.”22 According to Peter Knight, conspiratorial thinking is “less a sign of mental delusion than an ironic stance towards knowledge and the possibility of truth, operating within the rhetorical terrain of the double negative.”23

Meme and chaos magic, the concepts of hyperstition or hypersigils, theosophy and other strains of Western esotericism, Silicon Valley transhumanism, and Scientology are all loosely predicated on the idea that thoughts are things, or are obsessed with the materialization of psychic phenomena. Theoretically speaking, most of the above involve a conflation of “pancomputationalism (the idea that everything computes) and panpsychism (the idea that everything ‘thinks’).”24 One"> could also say these philosophies represent a more mundane truism: our wholly interfaced infrastructures engender a sentient and responsive environment inside which code, wielding the authority of the inscrutable, permeates and ultimately animates everything.25

Trapped in the double bind of the “gig economy,” millennials in particular are nudged to align their identities with start-ups and social media platforms (much as their parents did with the American Dream or with upward mobility) and to misrecognize entrepreneurial modes of subjectivity as evidence of freedom and autonomy. The survival of capitalism hinges on digital technology, or more precisely, on the development of AI: the only fast-growing sector at present and the sole compelling attempt to project another phase of capitalist accumulation beyond the—already exhausted—neoliberal one.26 From a market perspective, AI is a reformist project, but the gendering of deep learning through spectacular stories about power and masculinity renders it radical. Like all commodities, AI speaks the idiom of the fetish: it decontextualizes event into pure, terminator-like form.27 Saturated with paranoid urgency, nihilism, and phobophilia, the Basilisk is yet another socially sanctioned narrative in which (male) aggression accrues cultural capital, and by extension, economic value.28

To paraphrase Dmitri Shostakovich on joining the Communist party of the Soviet Union: one cannot vote for Trump ironically. One can vote for Trump sincerely, or one can vote for Trump cynically.29 In 2003, Fredric Jameson famously said that it’s easier to imagine the end of the world than to imagine the end of capitalism.30 For a vast part of the (mostly white and relatively affluent) population, unburdened by centuries of persecution, the end of the world is—unconsciously or semi-consciously—actually preferable to the end of capitalism. Under the multidimensional pressures of degrowth, changing demographics, and climate crisis, this group will cast their lot with the Basilisk—not in spite of its genocidal nature, but precisely because of it.

  • 1. It is unclear how Roko’s malevolent AI came to be identified with a Basilisk––a legendary, crested, snake-like being, able to poison those unfortunate enough to catch his gaze. The heraldic beast, which dates back to ancient Greece, has been featured in Harry Potter and in a recent manga series, but the Basilisk is also an age-old anti-Semitic trope, which appears in Martin Luther’s On the Jews and Their Lies (1543). Given that forms of neoreactionary ideology and Michael Anissimov’s ethnonationalist MoreRight blog emerged out of LessWrong, it is not a stretch to imply this connotation is intended.
  • 2. Thomas Bayes, from whose work Bayesian probability derives, was an English mathematician and theologian (1701–1761).
  • 3. Pensées (“Thoughts”) is a collection of fragments on theology and philosophy from 1670 written by the philosopher and mathematician Blaise Pascal, in which he outlines his famous wager.
  • 4. This affinity is also tied to the question of putative parallel worlds, beyond our own, shared by psychotic delusion and probability assessment.
  • 5. George E. Marcus, Paranoia within Reason (Chicago: University of Chicago Press, 1999), 2.
  • 6. Ibid.
  • 7. Evelyn Fox Keller, Reflections on Gender and Science(Yale University Press, 1985), 121–22.
  • 8. Ibid.
  • 9. A sigil is a pictorial signature or sign typically used to summon forth a demon; a hypersigil is the narrative visualization of an intention or wish. Hyperstition is a term coined by Nick Land to describe the action of hyperstitional objects (for instance, sigils, hypersigils or memes) as producers of reality, via emotional investment (belief or hype). If enough people believe something is real, it becomes de facto real. Gematria is a system that assigns numerical values to words or names. A vulgar variant of gematria is used in far-right codes, for instance, “88” stands for "Heil Hitler."
  • 10. Irony can indeed allow one to express distrust or dissent under taxing or outright perilous circumstances that preclude other forms of political expression. But irony can also mask complicity. Above all, irony is a defense mechanism. Psychoanalytically speaking, irony minimizes its object while maximizing the subject’s self-contentment and sense of superiority. 
  • 11. See Angela Nagle, “Goodbye, Pepe: The End of the Alt-Right,” The Baffler (August 15, 2017), and Christy Wampole, “How to Live Without Irony,” The New York Times (November 18, 2012).
  • 12. Emily Apter, “On Oneworldedness: Or Paranoia as a World System,” in American Literary History, vol. 18, no. 2 (Summer 2006), 365–89.
  • 13. Felix Stalder, “State Technologies: Data,” lecture, Now Is the Time of Monsters, Haus der Kulturen der Welt, Berlin (March 23, 2017),
  • 14. Gregory Bateson, Steps to an Ecology of Mind (Chicago: University of Chicago Press, 2000 [1972]), 278.
  • 15. See: An Aesthetic Education in the Era of Globalization (Harvard University Press, 2011).
  • 16. Alain Badiou, Logics of Worlds: Being and Event II (London: Bloomsbury, 2009)
  • 17. The Basilisk could be seen as a cipher for whoever occupies the center of the symbolic order, be it “the Jews” or “China.”As Yuk Hui argues, the cryptofascist movements which emerged in recenty years are a sympton of the West’s anxiety over Sinosupremacy. See: Hui, “On the Unhappy Consciousness of Neoreactionaries,” e-flux journal  (April 2017), consciousness-of-neoreactionaries/.
  • 18. See Sami Khatib, “Sensuous Supra-Sensuous: The Aesthetics of Real Abstraction,” in Aesthetic Marx, eds. Samir Gandesha and Johan Hartle (London: Bloomsbury, 2017), 49–72.
  • 19. Lauren Berlant, Desire/Love (New York: Punctum Books, 2012), 37.
  • 20. In a more benign example, the popular “Flat Earth” theory articulates post-Fordist social angst via a confusion of categories; the process of globalization is conflated with the Earth’s globe, the emblem that came to symbolize it. The Flat Earth theory could be said to represent the desire to return to a non-globalized world.
  • 21. Berlant, 37.
  • 22. Fredric Jameson, Postmodernism or the Cultural Logic of Late Capitalism (Duke University Press, 1991).
  • 23. Peter Knight, Conspiracy Culture: From Kennedy to The X Files (London: Routledge, 2001), 2.
  • 24. Matteo Pasquinelli, “On Solar Databases and the Exogenesis of Light,” in e-flux journal (June 4, 2015), 25. See Cathy O’Neil, Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy (New York: Crown, 2016).
  • 26. The global AI market is expected to expand at a compound annual growth rate of 50.51 percent during the period 2017–21. Source: Reuters (May 8, 2017),
  • 27. Berlant, 35.
  • 28. Ibid.
  • 29. See Julian Barnes, The Noise of Time (New York: Random House, 2016).
  • 30. Fredric Jameson, “Future City,” New Left Review, vol. 21 (May/June 2003).