Though taking place years apart and in a rather different context, the senator’s theatrics brought me back to another encounter I had, in 2016,when I was at the Google Cultural Institute in Paris for a summit on machine learning—not least because of another Singaporean connection. At that time, Google had just released DeepDream, a program that uses neural networks to produce surreal, psychedelic images by amplifying patterns within existing images. Artists across the world flocked to the application, captivated by what felt like an entirely new form of vision produced by AI, and it was on this basis that Google had convened the summit to explore potential projects at the intersection of art and machine learning.
A demonstration of the application Portrait Matcher at the Google Cultural Institute, Paris, 2016. Courtesy of the author.
There, I watched as the French artist and interaction designer Cyril Diagne demonstrated his application Portrait Matcher. Developed during his residency at the Google Arts & Culture Lab, the application allows users to upload any photograph with a human face, which is then analyzed through machine learning to be matched with a face appearing within the forty thousand digital reproductions of artworks in Google’s collection.
That moment, I noticed a familiar face—one that I immediately recognized as belonging to an iconic painting permanently displayed at the National Gallery Singapore: Chua Mia Tee’s Epic Poem of Malaya (1955). However, any sense of serendipity I felt quickly dissipated upon seeing that, besides their recognizably Asian features, there was little resemblance between the two faces matched by the application. Given the dismal underrepresentation of Asians in the dataset upon which the application was trained, perhaps it’s unsurprising that when given the face of an Asian man, the application would return a painting that I already knew all too well.
Yet, revisiting this moment today—just as generative models are turning out thousands of images by the minute and putting pressure on prevailing regulatory frameworks—I’m less interested in addressing questions of representation in proportional terms than in understanding the order of representation through which AI perceives, processes, and generates these images. For one, what happens to the logic of representation when a copy of a painting made at the height of the anticolonial movement in Singapore moves from a national collection into the artificial neural network of a multinational corporation where it becomes just one set of data samples within a vast computational archive? How can the transformation of this image—from art historical emblem to spatially redistributable samples—speak to the shifting role of the state as the long-standing referent of meaning when everything computable is on the verge of dissolving into the Cloud? What is the political order of an era saturated by images that appear to have no history?
Perhaps it’s fitting that the properly historical answer to these questions can be found by turning to Chua’s painting itself. As we shall see, it was exactly in the years when the painting was produced, amid the traumatic birth pangs of the nation-to-come, that the terms of our contemporary discourse pitting the state against global networks of technology and capital began to take form—most tellingly, behind the armed checkpoints and barbed-wire fences erected to keep a once formidable network at bay.
In Epic Poem of Malaya, a group of Chinese students are listening intently to an orator recite from a book on Malayan history against a dramatic rustic landscape. Completed in the year of the Rendel Constitution’s implementation, which granted many among Singapore’s mostly immigrant population the right to vote for the first time, the work reflects the self-study movement that had blossomed within the Chinese community at the time as part of the burgeoning of a shared Malayan identity.
That this vision of a Malayan Singapore failed to come to fruition is clear enough today, with Singapore existing as a sovereign city-state and Malaya as a political entity having been entirely dissolved into present-day Malaysia, a federation encompassing the peninsula and the Bornean states of Sabah and Sarawak. Although I won’t go into the tumultuous series of events that led to Singapore joining the federation in 1963 before making an unceremonious exit less than three years later, the actual process through which Singapore gained its freedom from colonial rule was evidently more chaotic than what Chua depicted, with movements and groups of all political orientations fighting for influence to determine the precise contours of the nation to come and the terms of the union. Quite unlike a scene of students quietly listening to a recital, a more representative image of the era would have been of the masses taking their protest to the streets, having been spurred by networks of progressive and radical activists into taking direct action against colonial rule. Of all these networks, none rattled the British more than the network of Communist Party affiliates that had infiltrated trade unions, schools, and other organizations across the region.
Once the unlikely wartime ally of the British, the Malayan Communist Party comprised an extensive and clandestine network of cadres across the peninsula that had only grown in strength after the British turned against the party and forced its operations underground.
Chua Mia Tee, Epic Poem of Malaya, 1955, oil on canvas, 107 × 125,5 cm. National Gallery Singapore collection.
Officially known as the Malayan Emergency (1948–60), the protracted war which ensued was launched by the British to safeguard their economic interests in a region where the demands for decolonization could no longer be suppressed. Seeing that there was no bigger threat to these interests than a communist Malaya, the colonial government took pains to ensure that the Communist Party would play no part in the postcolonial government that would follow. As they eventually found out, however, to destroy a network so pervasive and tenacious, violence alone was not enough; a space needed to be constructed where the people’s exit from the network could be secured.
In the peninsula, large parts of the population were forcibly resettled into so-called New Villages (kampung barus)—military-run internment camps where villagers were kept under close watch to prevent them from providing food and shelter to guerrilla insurgents. Reflecting the crude conflation of ethnicity and political affiliation that has been restaged time and again into our present-day geopolitical rivalries, the communities targeted for resettlement were overwhelmingly Chinese. But the Briggs Plan—officially named as such after the director of operations, Harold Briggs—was more than a security operation; it also formed a crucial part of a broader ideological campaign to win the villagers’ “hearts and minds.”[01] Through the resettlement, the targeted communities—many of whom came from squatter settlements on the fringes of plantations—not only gained access to sanitation, education, and healthcare but also, for the first time in their lives, had the option to apply for land titles.[02] Such incentives were conceived in the hope that, by experiencing the material comforts that came with living in the planned community, the newly pastoralized villagers would be convinced of the merits of social democracy over communism, allowing them to be assimilated into the body politic of the incipient nation-state.
However, it didn’t take long for the British’s strategy to unravel, as the internees grew to resent their containment and surveillance over time, with some even throwing their lot with the communists. Having failed to win their hearts and minds, the colonial authorities soon realized that what the villagers actually thought or believed was irrelevant as long as they acted in a manner that produced useful information.[03] It was upon this reckoning that British counterinsurgency operations took a communicative turn. Instead of simply relying on propaganda leaflets and radio broadcasts, the authorities doled out increasingly generous monetary rewards in exchange for information while routinely rounding up the villagers to fill in questionnaires and drop them into sealed boxes to offer information anonymously.[04]
Aerial view of the New Village in Malay Peninsula (kampung barus), 1950s. Image: IWM (K 13796).
As recounted by former New Village internee Tan Teng Phee, such practices were implemented in tandem with the progressive introduction of elections at a local, state, and finally federal level, and were explicitly designed “to develop a democratic spirit” among the villagers.[05] Yet, given the use of draconian measures such as collective punishment upon villages where demands for information were met with reticence, democracy as practiced here was less about the right to speak or to express one’s political opinion by the vote than about the performance of a set of standardized procedures that reduced civic participation to feedback—a process through which a population and its administrators adjusted their respective behaviors to reach a mutual accommodation of each other. What started as an intelligence operation by way of manual data collection thus evolved into a technical system imbued with its own “intelligence,” capable of constant adaptation by processing and acting on the results of its own outputs.
This renewed strategy proved relatively successful, so much so that administrators across the British Empire and beyond began adopting it as a template for their own counterinsurgency operations: first by the British in Kenya in 1953, then by the US and South Vietnamese authorities during the Vietnam War in 1962. While the Vietnamese experiment in building “strategic hamlets” modeled after the Malayan New Villages ended in disaster— thanks to corruption and a lack of civilian cooperation—the “villagization” policy introduced by British administrators in Kenya’s Great Rift Valley came closer to its model in fulfilling its objectives.[06] There, the construction of fortified villages was a direct response to the outbreak of the anticolonial Mau Mau uprising and aimed to deradicalize a native population whose political inclinations threatened the emerging world of global markets where the British sought to retain its stronghold.
In a striking instance of what can only be called the military-industrial-theory complex, it was also in the Kenyan “new villages” where the media theorist Marshall McLuhan found his inspiration for his famous concept of the “global village”—a concept first explored in his 1962 book The Gutenberg Galaxy, which is widely seen to have anticipated the development of the internet. Indeed, rereading McLuhan today with this context in mind, it is astonishing how often McLuhan invoked Africa as part of his comparison of literate and so-called primitive nonliterate cultures. Central to McLuhan’s thesis is that the advent of electronic new media in the twentieth century allowed Western society to return to what he called “the Africa within,” in reference to a more sensuous mode of experiencing the world that had been displaced by the invention of the printed word.[07] A world connected through electronic new media would therefore feel much smaller and more intimate, like the African village of McLuhan’s imagination.
As the architectural historian Ginger Nolan points out, despite the abundant scholarship on McLuhan, scant attention has been devoted to the fact that some of the literature cited by McLuhan on Africa was in fact developed through colonial research carried out on the continent in the final years of British rule.[08] Especially revealing are the writings of ethno-psychiatrist John Colin Carothers, cited repeatedly in The Gutenberg Galaxy, who was specifically invited by the colonial government to conduct research in the Kenyan new villages. Drawing upon Carothers’s theory of the African mind as one steeped in the amniotic depths of the auditory—as opposed to the cold detachment of the visual upon which Western print culture was founded—McLuhan laments the “detribalizing power of the phonetic alphabet” while hailing electronic new media for enabling us to live “pluralistically in many worlds and cultures simultaneously [. . .] in a single constricted space resonant with tribal drums.”[09]
It is not a coincidence that the media that most fascinated McLuhan— namely, the “tribal drum of radio”—was extensively used across both Malaya and Kenya for information dissemination.[10] But the Kenyan operation also differed from its Malayan model in one important regard: While in Malaya the preference for audiovisual media was made in light of the villagers’ low literacy rates, in Kenya it was less a lack of reading ability than a perceived predisposition for the sensuous immediacy of the aural that informed the British strategy. McLuhan himself was notably fascinated by reports of African reception of film, observing that, unlike in the West, the African is never a “passive consumer” but will always accompany a screening with live commentary, effectively turning every act of media consumption into an oral and participatory experience.[11]
This capacity for providing immediate feedback, which would become an essential built-in feature of later technologies, was decisive in allowing colonial administrators to gauge receptiveness toward a circulated message. As noted in the work of Leonard W. Doob, another colonial researcher studied by McLuhan, such strategies had long been taken up by missionaries who “in the attainment of their crusading goal” sought to identify existing cultural predispositions among Africans that could be harnessed to allow the gospel to resonate better. Yet—and here’s the crucial bit—Doob also emphasized that “no amount of feedback can modify some sections of Christian belief.”[12] What this implies is that while both Africans and Europeans were wired into the same feedback loop, allowing them to recalibrate the signals sent to each other, it is the Christian alphabetic man who wielded the semiotic power in the final instance. The same can be said for the forms of techno-mediation that villagization entailed: From Malaya to Kenya, the new village was precisely that “single constricted space” where the power to make resonant was deployed to transform an agitated public sphere into an unimpeded site of information flows in the image of the market economy.
To be sure, none of this suggests that the forms of domination and control embedded in our contemporary networked condition—the dark underside of McLuhan’s sanguine vision—are in any way equivalent to the brutal regime imposed through the global chain of new villages, from which McLuhan, knowingly or not, turned away. Rather, what demands our scrutiny are the designed structures of containment, often underwritten by the state’s monopoly on violence, that fall from view in the abstraction of the “village” from a historically situated infrastructure of colonial dispossession into a metaphor for a kind of electrified global intimacy—structures which explicitly served to enfold a population into the “open space” of the “free market.” The topographical flatness of the village through which information looped was secured vertically through imperial command and centralized control—the Briggs Plan was, after all, a plan. One could thus also observe that British counterinsurgency experiments in those years merely anticipated the much more audacious US Global Plan that unfolded as the Cold War began in earnest, of which the Marshall Plan was just a modest beginning. We need only remember that it was the central planning of the Pentagon, which channeled billions of dollars into its war machine in Southeast Asia, that enabled market economies to proliferate around US bases, supplying them with everything from refined oil to cheap sex.[13]
Indeed, contemporary debates framing the Plan and the Network as diametrically opposed obscure not only the extent to which networks almost always involved some form of central coordination but also how planning was then the first order of the day wherever one looked. Even outside of the communist world, prevailing social-democratic norms, combined with a planning regime left over from the war, meant that governments from Europe to its existing or former colonies in Asia and Africa were deep in the throes not of globalization but of “planification.”[14] Nowhere was this more evident than in the provision of housing, where it was Singapore, more than any other country, that succeeded in transforming a city of squatters into a nation of home-owning citizens.
In fact, it was not long after the New Villages were founded on the peninsula that efforts to rehouse the city’s population into multistory modernist apartment blocks came into full swing in the separately administered city-state—first by the British-run Singapore Improvement Trust and then by its more prodigious postcolonial successor, the Housing and Development Board. Compared to the horizontal displacement of squatters into villages in Malaya, the vertical ascent of the displaced population ultimately served less to model a form of communal living that could dissuade the population from aligning with the communists (whose networks in Singapore had already been decimated by a series of crackdowns in 1950) than to forge a disciplined and incentivized labor pool for the emerging industrial wage economy. By the early 1960s, with the Malayan Emergency declared over and Singapore on the cusp of entering Malaysia, it appeared by all accounts that the (national) Plan had defeated the (communist) Network.
Yet if there’s one country that’s in the best position to shift the discourse in favor of networks—specifically from the nation as an object of planning with or against networks to the nation as itself a network held together by a Plan—it would be the newly independent Singapore just a few years after, when the collapse of its dreams of a common Malayan market meant that its only option was to supply the global market. In a landmark speech delivered to a room of journalists in 1972, foreign minister S. Rajaratnam proclaimed that Singapore was now a “global city” that no longer worried about the loss of the Malayan hinterland, for the world had become its hinterland. As a global city, Singapore, according to Rajaratnam, was becoming an important part of a chain of cities connected through “the tentacles of technology,” allowing it to trade with any place in the world and not just its immediate neighbors—this was decades before sociologist Saskia Sassen popularized the now-banal moniker.[15]
For sure, the kinds of networks into which Singapore was embedding itself were categorically distinct from the communist networks so feared by the British colonial government and its postcolonial successor. Nonetheless, policy planners in Singapore were deeply cognizant of the tensions between the developmentalist trajectory it was pursuing, which emphasized central planning and hierarchical control, and a liberal economic agenda promising the free movement of goods and capital in the context of an immigrant city-state still shaping its sense of identity. As Rajaratnam saw it, it was accordingly the task of the government to “create the necessary anti-bodies” to give the national body immunity against the many risks posed by its heightened exposure to foreign markets and capital.[16]
Rajaratnam was not known to have read McLuhan, but the affinities between their respective concepts are hard to miss. It’s worth mentioning that even before The Gutenberg Galaxy, McLuhan had in fact already conceived of the global city in relation to the global village. In a 1960 letter to Jaqueline Tyrwhitt, a close colleague at the University of Toronto, McLuhan spoke of the necessity of building a global city as a center “for the village margins”—a thought likely influenced by Tyrwhitt’s work as a United Nations technical adviser in India where the British urban planner constructed “village centers” as sites of education to enfold villagers into the emerging networks of “global development.” Drawing upon the psychological principle that “a center without a margin” leads to mental paralysis, McLuhan believed that given the interchangeability of center-periphery positions enabled by electronic communications, such a city would be required “to coordinate and concert the distracted sense programs of our global village.” He added that its function would be similar to a massive airport managing air-traffic flows—a description that could very well apply to Singapore’s famed Changi Airport, often taken as a synecdoche for the city-state itself.[17]
With this, we arrive at a preliminary sketch of the political order of the global city/village—a point upon which McLuhan, given his reticence over his own political beliefs, never quite expounded. Combining McLuhan’s and Rajaratnam’s insights, we could thus imagine the nation as a body, with the state functioning as its brain, from which networks of labor, trade, and capital extend toward other nodes within a global technostructure where center-periphery relations are defined as much by the traditional concentration of wealth and firepower as by the throughput and quality of information flowing through a given site. Yet, as we shall soon observe, this vision of a world composed of interconnected networked states, as avantgarde as it was for its time, would quickly be challenged by an even more radical political program—one that sought to rewire the entire planet in the image of a brain without a body.
In June 2014, a group of computer scientists at the University of Montreal introduced a new class of machine-learning models known as generative adversarial networks (GANs), with the ability to generate photorealistic images. While often seen as the start of the ongoing generative AI boom, the breakthrough achieved by GANs was in fact decades in the making, having originated in early cybernetic projects that sought to create artificial neural networks based on the structure of the human brain. To produce accurate results such networks require immense computing power, vast pools of data, and an enormous standing reserve of informal workers to label the training data, all of which were not available before the arrival of graphics-processing units, the internet, and platforms like Amazon Mechanical Turk.[18]
So impressive were the results produced by the first generation of GANs that portraits of so-called imaginary individuals quickly made their rounds on the internet, in the process raising ethical concerns over the potential for such “deepfakes” to spread misinformation and manipulate public opinion. As such images proliferated, it also became apparent that, when it comes to reproducing faces, many of these models appear inclined toward and more proficient at recognizing and reproducing Caucasian features— an issue I witnessed myself in 2016 during my encounter with the image of Chua Mia Tee’s painting in Paris. While with each model it is often unclear where the bias is located, much attention has been directed to potential biases in the datasets used to train these models, in turn spurring litanies of critique on the colonial genealogy of generative AI models.
For all the validity of such concerns, the suggestion of a direct line of transmission between the archive and the model leaves unexamined the logic of the neural network and the challenge it presents to traditional representational regimes. Today, more stable and efficient models, such as transformers and diffusion models, have replaced GANs in the creation of some of today’s most widely used prompt-based image generators. Yet, the fundamental logic of the feedback loop remains operative: unlike in rule-based systems, the behavior of each neuron in the neural network results from a randomized process of trial and error without any kind of intervention from the programmer, as if the neurons were characters who wake up in a time loop with no context whatsoever and are left to figure out for themselves the best way to navigate the space in which they find themselves so as to escape the loop for good.
Here, it’s important to emphasize the essential role played by randomness throughout this process. As the basic building block of a generative model, the noise sample is a randomly generated unit of data that replaces authorship in the conventional sense with a network effect, swapping semiosis itself for a probabilistic process of iteratively refining a sample through numerous feedback loops. Insofar as the “creativity” of such models results from their ability to produce recognizably original images, randomness is the indispensable precondition that allows a model to abstract—rather than simply replicate—the data upon which it is trained by progressively recalibrating noise samples to hit a target probability distribution. This also means that abstraction here works quite differently from how it does in language, to the extent that it does not logically entail any kind of figuration at all. From the viewpoint of the model’s machinic vision, the difference between the random noise samples with which it began its training and the final output image is purely mathematical.
Given this, to the extent that such models reinscribe the inequities of the colonial archive, it is not because they have usurped its semiotic power but because they have pulverized it entirely into randomized clouds of noise, fundamentally divorcing the inscribed from the act of inscription while reproducing its likeness as a quantitatively abstracted trace. Put simply, such models are set up for repeating the mistakes of history because they literally cannot remember them in any qualitative sense—as mistakes. This also means that instead of clamoring for stronger laws to protect copyright owners or for the diversification of datasets—which will merely reascribe or redistribute semiotic power—the more urgent task involves coming to terms with the logic of the emerging representational order and how it’s already upending existing ethical and legal frameworks.
This coming anti-semiotic regime has already been in formation for some time, its origins going back to the inception of what has become the metanetwork of our age: the internet. In 1969, the ARPANET, the precursor of today’s internet, went online as the world’s first large-scale packet- switching computer network. Developed under the US Department of Defense, the network drew on the early cybernetic research of Warren McCulloch and Walter Pitts, whose 1943 paper on artificial neural networks was among the first efforts to connect computational logic to brain structure. Similar to how neurons in the brain form connections with other neurons to transmit information, the ARPANET was accordingly designed as a distributed network where computers can communicate directly with each other through interconnected nodes without going through a center. In the context of the ever-imminent threat of a nuclear attack, the strength of such a network was that it allowed information to flow even if certain nodes were taken offline, just like how the brain adapts and reroutes signals when certain areas are damaged.
ARPANET logical map, 1977.
However, as a project realized through centralized funding and government oversight, the ARPANET is more accurately described as a distributed network fully encased by the state, allowing the latter to consolidate its power through the former in a manner that state-socialist experiments in building national computational networks—from the Soviet OGAS to Chile’s Cybersyn under the Salvador Allende government—had envisioned but never attained.[19] The transformation of the network into the freely accessible and notionally unbounded space that is today’s internet only came later in the 1980s through the interconnection of networks across multiple government organizations and academic institutions in the US. As has become more evident over time, what accompanied this expansion was a cracking apart of the political order from which the network originated, and it was within this tentative fissure that the biological metaphor underlying the network’s implementation—the network as a brain contained within a body—became profoundly decorporealized as it became subjected to the logic of a more encompassing network: the market.
In the 1990s, against the backdrop of a bubbling ideological cocktail of techno-utopianism and market fundamentalism centered on the Bay Area in California, the publicly funded open-protocol architecture of the internet was deregulated and ceded to a handful of private companies, transforming the network into a truly public good and opening it up for commercial exploitation at the same time. Given the contemporaneous alternatives such as proposed public utility models and citizen-run networks, the eventual establishment of the internet as a privately owned “public space” was far from the inevitable result of a natural isomorphism between the market and the internet’s TCP/IP-based backbone.[20] Instead, if the trippy post-Gutenberg typesetting of Wired magazine—the “monthly bible” of Californian ideologues—serve as any indication, it was the historical convergence of a lapsed hippie radicalism and the deregulationist Right by way of a shared antiestablishment ethos that gave rise to cyber-libertarian dreams of networked technologies as a means for abolishing the state altogether, sowing the seeds for the internet as we use it today.[21]
In this double movement of the nation entering the network just as the network enters the market, what accordingly transpired was not simply a change of guards but also a radical refiguration of the imaginary that gave form to the network. As media historian Benjamin Peters suggests, when neoliberal proponents of the internet envision the transformation of the nation—or indeed the entire planet—into a network, the analogy upon which they lean is not of a body gaining a brain in the form of the network than of it becoming one in its entirety, with the bodies that make up the nation in turn becoming neurons firing at each other within this disembodied brain. Departing from earlier and contemporary imaginaries of the networked state—from Rajaratnam’s global city to Singapore’s ongoing Smart Nation initiative—the brain here is not the command center of a vast nervous system extending toward the furthest reaches of an organic whole. Rather, it is a theoretically limitless matrix of equally electrified nodes, each the address of an individual assumed to be less a body engaged in physical labor than a purely cognitive entity free to enter into peer-to-peer relations with another such entity.[22]
Certainly, even the most rudimentary AI chatbot would tell you that this vulgar McLuhanism cannot be further from what the internet is today: less a marketplace of ideas than a fragmented attention economy of microtargeted ads and compulsive doomscrolls, less an electronic agora than a communicative purgatory trapping us in positive feedback loops of simulated outrage, less an emancipatory tool for a new generation of digital nomads than an oligarchic concentration of “free,” data-rich platforms built on casualized and poorly remunerated labor.
More significantly, as the architecture of control layered onto the internet begins taking on a planetary scale, what we’ve been witnessing in this largely unregulated political frontier is the emergence of what can be called stately networks—networks, specifically cloud platforms, that begin to instantiate new forms of sovereignty without entirely replicating the state’s traditional functions. As Benjamin Bratton has theorized, this unprecedented political formation cannot be understood through prevailing discourses on “technologies of governance” because it is generated by the very computational network that it attempts to govern: not the state as machine or the state machine, but the machine as the state. As we can see with every new regulation enacted to keep up with the latest mutation of the technostructure, this new political order, far from being modeled on a centuries-old nation-state framework, is already transgressing its legal and conceptual boundaries while transforming the very idea and practice of statehood itself.[23]
Here, Bratton’s now-ubiquitous concept of the Stack intervenes to provide a distinctly vertical topography to describe the global system that governs our everyday interactions with the looped planar topologies of today’s computational networks. Comprised of six interdependent layers— Earth, Cloud, City, Address, Interface, and User—this hierarchically ordered “accidental megastructure” inscribes new political constellations primarily through the outsized weight of cloud companies like Alphabet and Meta, whose platforms collectively gather a population of citizen-users so massive and distributed as to outstrip the most sophisticated state surveillance efforts. Yet, to the extent that such companies control some but not all layers of the Stack, whatever forms of sovereignty produced out of their verticalized totalities also cannot impose upon the Stack a singular political logic, but must necessarily contend with other existing and new jurisdictional claims—including those of the state, which can likewise scale up and down the Stack—in the process generating a spectrum of compound political worlds that, by Bratton’s estimation, cannot “finally resolve into a consensus sovereignty of last instance (or last resort).”[24]
To this end, the difficulty in holding cloud companies accountable for their platforms’ social harms arises not simply from a network effect displacing culpability onto an amorphous and jurisdictionally meaningless neuron-user crowd. Rather, it is the result of the materialization of that effect through the vertical body of the Stack whereby an input from a given user must necessarily pass through all its layers—from TikTok video to 5G antenna to server farm to cobalt mine—on its way down and then up again to another user, in the process navigating an entire rugged terrain of competing and interlocking juridico-discursive zones and political no-man’s-lands.
It’s clear then that the challenge posed by stately networks lies in their radical break from twentieth-century modes of capital accumulation, upon which governments, for the most part, have honed the tools of statecraft that have hitherto served them well. But does this emergence necessarily imply, as Bratton suggests, that the nation-state system is, by way of an inverse relation, headed towards obsolescence? Looking at the waves of deficit spending, tariff showdown, and industrial reshoring that have taken place since The Stack was published, my sense is that this treatise both overgeneralizes the state form and underestimates its resilience.[25] There are, of course, strong reasons to believe that the state form is incapable of reining in the excesses of cloud capitalism. With each new hearing to address the latest scandal to hit one or more major cloud companies, we are seeing a curious reversal of the neural network’s distributed logic, as the CEOs of these companies use the force of their personalities to reassure the public of the ethical standards to which they—not the state—hold themselves accountable. This heightened identification of such companies with their leaders has further inspired much analysis suggesting that this new class of “cloudalists” in fact portends the end of capitalism and the rise of a technofeudalism built on rent rather than profit—reminiscent of the power once wielded by feudal lords, except over digital infrastructure rather than land.[26]
Yet, what is less appreciated in such critiques is how such platforms can only survive by virtue of their “feudal” entitlements as bequeathed to them by the state, whose position remains secured by its still-unchallenged monopoly on violence and its hallowed status as the institutional form of the nation in all its affective moorings. This parasitical relationship was on full display at Trump’s second inauguration, where the heads of the major US digital fiefdoms gathered to pay tribute to their presumed sovereign in exchange for tax cuts and deregulation (which is nothing more than a further enclosure of our digital commons). Given that Trump had won the election on an unapologetic agenda of economic nationalism and US withdrawal from the global system, this crowning image signifies less a case of odd bedfellows than of the reinscription of the technostructure within the nation-state’s semiopolitical (but not necessarily economic) domain.
Criticisms of Trumpian populism as more gesture than policy thus completely miss the point, for it is exactly in the gesture’s affective affordances that one secures a represencing of the nation through the state form over the immediate but semantically hollow connections of the network— or a restoration of semiotic over noetic power at a time when the redistribution of the social along verticalized informational routes has rendered impossible the return of the social to itself (as both commons and common sense). That the faithful here will inevitably find out that mere presence cannot substitute for political representation is less revealing of the contemporary limits of state power than of the poverty of our current democratic mechanisms for broadening access to that power.[27]
In fact, shifting our gaze outside the Western hemisphere would show how state power has been augmented amid the proliferation of computational networks. While this phenomenon ostensibly reflects the limits of horizontal politics within authoritarian regimes, the determinant in each case is more accurately the extent to which the state is able to appropriate or dispense with networks. Here, our most obvious example is China’s ascent into what is arguably the most expansive networked state of our era—a development that has notably occurred through the exclusion of networks deemed pernicious. What’s more concerning, however, is the increasing use of complete internet shutdowns by governments seeking to disable entire dissident networks. From what appears to be its earliest deployment in 2007 amid the Saffron Revolution in Myanmar to its more prominent use by the Egyptian government during the Arab Spring or more recent applications in Rakhine and Chin states and in Jammu and Kashmir, what has emerged as a peculiar irony is how an entire network can grind to a halt through the activation of a proverbial kill switch precisely because of its electronic interconnectedness. In another turn to our unending McLuhanite drama, it is a rebel group in the rural periphery, not least one in pursuit of its own state-building project, that recognizes most intimately the inherent stateliness of the internet—that it is only through the centralized governance of its physical mechanisms and programmatic protocols that the internet can be materialized as a distributed network. This same understanding also animates persistent secessionist fantasies of a very different kind among Silicon Valley’s most ardent libertarians, for whom the point is not to exit the state form that they cannot do without but to have the option to choose from a marketplace of states. Effectively delinking the nation from the state form, this techno-anarcho utopia that has been articulated by the likes of Peter Thiel and Balaji Srinivasan seeks the proliferation of states in tandem with networks, enabling the citizen- user to shop between different master algorithms and to hop to the next jurisdiction (or to create an entirely new one) to escape a rogue regime.[28]
As it stands, the political order of the day, however historically contingent its emergence, is caught up between two competing and sometimes interlocking imaginaries seeking to inscribe their design principles upon it. On one side is the “body with a brain” that gives form to the networked state; on the other, the “brain without a body,” whose de facto political constitution is the stately network. Despite their differing ideas of how exactly a network is (or is not) incorporated, both profess a continued faith in networks: the latter unconditionally, the former on the premise that even the pathologies engendered by networks are best addressed through a more judicious deployment of networks themselves.[29] In truth, seeing how readily we turn to machines today to seek the best possible way of keeping alive a system that is driving us toward the absolute worst outcome of planetary catastrophe, the user of tomorrow looks just like the protagonist of a late-capitalist time loop, cycling up and down the Stack in pursuit of the least hazardous way of navigating its fragmented sovereignties and ideological minefields.
But if there’s any way out of this conundrum, it is by first reckoning not merely with the multiple interconnected strata that comprise the technostructure that we inhabit today but also with the layers of history that inhabit us through our networked interactions with each other. As we’ve seen, the Stack we have is, in fact, a haunted stack of imperial misdesign, failed intelligences, strategic missteps, and unrealized convergences. Each is a moment telling us how things could very well have been otherwise: that the trajectory that led us from a 1955 painting hanging in the National Gallery Singapore to the spectral traces of its likeness dispersed within the neural networks of the Google Cloud is not technological but historical. Funnily enough, the actual distance between the model and its digital copies might be far less breathtaking than the history that connects them. In 2013, Google opened its first Southeast Asian data center in Singapore, drawn by the city’s global connectivity and tax incentives. In the loop between the networked state and the stately network, the ghosts of history give proof to a different kind of resonance.
Ho Rui An is an artist and writer working in the intersections of contemporary art, cinema, performance and theory. Through lectures, essays and films, his research examines the relations between labor, technology and capital across different systems of governance in a global age.
Ho Rui An. Photo courtesy of Eike Alkenhorst
[1] Susan Caruthers, Winning Hearts and Minds: British Governments, the Media and Colonial Counter-insurgency 1944–1960 (London: Leicester University Press, 1995), 87–95.
[2] Tan Teng Phee, Behind Barbed Wire: Chinese New Villages During the Malayan Emergency, 1948–1960 (Petaling Jaya: Strategic Information and Research Development Centre, 2020), 141.
[3] Caruthers, Winning Hearts and Minds, 94.
[4] Villagers were told to mark their questionnaire with a symbol so they could be identified for a reward if their information led to a capture. See Karl Hack, The Malayan Emergency: Revolution and Counterinsurgency at the End of Empire (Cambridge: Cambridge University Press, 2022), 294–296.
[5] Tan, Behind Barbed Wire, 74.
[6] For a firsthand account of the Vietnamese strategic hamlets, see Robert Thompson, Defeating Communist Insurgency: Experiences from Malaya and Vietnam (London: Chatto & Windus, 1966).
[7] Marshall McLuhan, The Gutenberg Galaxy: The Making of Typographic Man (Toronto: University of Toronto Press, 1962), 45.
[8] Ginger Nolan, The Neocolonialism of the Global Village (Minneapolis: University of Minnesota Press, 2018).
[9] McLuhan, Gutenberg Galaxy, 31. See also J. C. Carothers, The African Mind in Health and Disease (Geneva: World Health Organization, 1953).
[10] Marshall McLuhan, Understanding Media: The Extensions of Man (New York: McGraw-Hill, 1964), 297.
[11] McLuhan, Gutenberg Galaxy, 38.
[12] Leonard W. Doob, Communication in Africa: A Search for Boundaries (New Haven, London: Yale University Press, 1961), 27, 346.
[13] This description of US foreign policy as a “Global Plan” comes from Yanis Varoufakis. See Yanis Varoufakis, The Global Minotaur: America, Europe and the Future of the Global Economy (London: Zed Books, 2011), 57–89.
[14] Gregory Clancey, “Towards a Spatial History of Emergency: Notes from Singapore,” Beyond Description: Singapore Space Historicity, eds. Ryan Bishop, John Philips, and Yeo Wei-Wei (London: Routledge, 2004), 31.
[15] S. Rajaratnam, “Singapore: Global City,” Singapore Press Club, Singapore, February 6, 1972. See also Saskia Sassen, The Global City: New York, London, Tokyo (Princeton, NJ: Princetown University Press, 1991).
[16] Rajaratnam, ibid.
[17] Marshall McLuhan, Letters of Marshall McLuhan, eds. Matie Molinaro et al. (London: Oxford University Press, 1987), 278. See also Olga Touloumi, “Globalizing the Village: Development Media, Jaqueline Tyrwhitt, and the United Nations in India,” Architecture in Development: Systems and the Emergence of the Global South (Oxford, New York: Routledge, 2022), 259–277.
[18] For an in-depth analysis of on-demand work platforms, seeMary L. Gray and Siddharth Suri, Ghost Work: How to Stop Silicon Vallery from Building a New Global Underclass (Boston, MA: Houghton Mifflin Harcourt, 2019).
[19] On the Soviet OGAS, see Benjamin Peters, How Not to Network a Nation: The Uneasy History of the Soviet Internet (Cambridge, MA: MIT Press, 2016). On Chile’s Cybersyn, see Eden Medina, Cybernetic Revolutionaries: Technology and Politics in Allende’s Chile (Cambridge, MA: MIT Press, 2014).
[20] Geert Lovink, Dark Fiber: Tracking Critical Internet Culture (Cambridge, MA: MIT Press, 2002), 42–67.
[21] Richard Barbrook and Andy Cameron, “The Californian Ideology,” Science as Culture 6, no. 1 (1996): 44–72.
[22] Peters, How Not to Network a Nation, 55.
[23] Benjamin H. Bratton, The Stack: On Software and Sovereignty (Cambridge, MA: MIT Press, 2016), 8.
[24] Ibid., 5, 250.
[25] Bratton partially addresses this in his later work on “hemispherical stacks”. See Benjamin H. Bratton, “On Hemispherical Stacks: Notes on Multipolar Geopolitics and Planetary-Scale Computation,” As We May Think: Feedforward: The 6th Guangzhou Triennial, ed. Wang Shaoqiang (Guangzhou: Guangdong Museum of Art), 77–85.
[26] Yanis Varoufakis, Technofeudalism: What Killed Capitalism (Brooklyn: Melville House, 2024). See also Cedric Durand, How Silicon Valley Unleashed Techno-feudalism: The Making of the Digital Economy, trans. David Broder (London: Verso, 2024).
[27] On the difference between represencing and representation, see Rosalind C. Morris, “Populist Politics in Asian Networks: Positions for Rethinking the Question of Political Subjectivity,” positions 20, no.1 (2012): 37–65.
[28] Quinn Slobadian, Crack-Up Capitalism: Market Radicals and the Dream of a World Without Democracy (New York: Metropolitan Books, 2021), 2, 211.
[29] Orit Halpern and Robert Mitchell call this the “smartness mandate.” See Orit Halpern and Robert Mitchell, The Smartness Mandate (Cambridge, MA: MIT Press, 2022).
Fellow Sooyoung Leam
