LONDON — There are 8 million artifacts in the British Museum. But to commence his tale of existential jeopardy, risk expert Luke Kemp made a beeline for just two items housed in a single room. On a visit in early fall, beyond a series of first-floor galleries displaying sarcophagi from pharaonic Egypt, we stopped beside a scatter of human bones.
The exhibit comprised two of the 64 skeletons unearthed from the sands of Jebel Sahaba, in northern Sudan, in 1964. Believed to be over 13,000 years old, the bodies in this prehistoric cemetery were significant for what they revealed about how their owners died. Of those 64 skeletons, at least 38 showed signs of violent deaths: caved-in skulls, forearm bones with parry fractures from victims staving off blows, or other injuries. Whether a result of organized warfare, intercommunal conflict or even outright massacre, Jebel Sahaba is widely considered to be some of the earliest evidence of mass violence in the archaeological record.
According to Kemp, these shattered bones were a foreshadowing of another object in this room. Ten feet away, displayed at knee-height, was the Palette of Narmer. Hewn from a tapering tablet of grey-green siltstone, the item on display was an exact cast of the 5,000-year-old original — discovered by British archaeologists in 1898 — that now sits in Cairo’s Egyptian Museum.
At the center of the stone stands the giant figure of Narmer, the first king of Egypt. His left hand clasps the head of an enemy, presumed to be a rival ruler of the Western Delta. In his raised right hand he holds a mace. The image is thought to depict Narmer bludgeoning his greatest opponent to death, an act that solidified his sovereignty over all Egypt. Beneath his feet lie the contorted bodies of two other victims, while overhead a falcon presents Narmer with a ribbon, believed to represent the god Horus bestowing a gift of the Western Nile. “Here we have perfect historical evidence of what the social contract is. It’s written in blood,” Kemp told me. “This is the first depiction of how states are made.”
In the British Museum’s repository of ancient treasures and colonial loot, the palette is by no means a star attraction. For the half hour we spent in the room, few visitors gave it more than a passing glance. But to Kemp, its imagery “is the most important artwork in the world” — a blueprint for every city-state, nation and empire that has ever been carved out by force of arms, reified in stone and subsequently turned to dust.
Systematizing Collapse
When Kemp set out seven years ago to write his book about how societies rise and fall — and why he fears that our own is headed for disaster — one biblical event provided him with the perfect allegory: The story of the Battle of the Valley of Elah, recounted in 1 Samuel 17. Fought between the Israelites and the Philistines in the 11th century BCE, it’s a tale more commonly known by the names of its protagonists, David and Goliath.
Goliath, we are told, was a Philistine warrior standing “six cubits and a span,” or around 9 feet, 9 inches, clad in the alloy of copper and tin armor that would give his epoch its name: the Bronze Age. As the rival armies faced off across the valley, the giant stepped onto the battlefield and laid down a challenge that the conflict should be resolved in single combat.
For 40 days, Goliath goaded his enemy to nominate a champion, until a shepherd named David came forward from the Israelite ranks, strung a stone into his slingshot and catapulted it into Goliath’s brow, killing him at a stroke, and taking his head with the giant’s own sword. For centuries thereafter, the story of David and Goliath has served as a parable challenging the superiority of physical might. Even the most impressive entity has hidden frailties. A colossus can be felled by a single blow.
According to Kemp’s new book, “Goliath’s Curse,” it’s a lesson we would do well to heed. Early on, he dispenses with the word “civilization,” because in his telling, there is little that might be considered civil about how states are born and sustained. Instead, he argues that “Goliath” is a more apposite metaphor for the kind of exploitative, hierarchical systems that have grown to organize human society.
“‘Goliath’ is a more apposite metaphor for the kind of exploitative, hierarchical systems that have grown to organize human society.”
Like the Philistine warrior, the Goliath state is defined by its size; in time, centralized polities would evolve to dwarf the hunter-gatherer societies that prevailed for the first 300,000 years of Homo sapiens. Ostensibly, it is well-armored and intimidating, exerting power through the threat and exercise of violence. And, in kind with the biblical colossus, it is vulnerable: those characteristics that most project strength, like autocracy and social complexity, conceal hidden weaknesses. (A more modern allegory, Kemp writes, can be found in the early Star Wars movies, in which a moon-sized space station with the capacity to blow up a planet can be destroyed by a well-placed photon torpedo.)
Kemp is, of course, by no means the first scholar to try to chart this violence and vulnerability through the ages. The question of what causes societies to fail is arguably the ultimate mission of big-picture history, and a perennial cultural fixation. In the modern era, the historian Jared Diamond has found fame with his theories that collapse is usually a product of geographical determinism. The “Fall of Civilizations” podcast, hosted by the historian Paul Cooper, has over 220 million listens. Perusing a bookshop recently, I spotted a recent release, entitled “A Brief History of the End of the F*cking World,” among the bestsellers.
What distinguishes Kemp’s book from much of the canon is the consistencies he identifies in how different political entities evolved, and the circumstances that precipitated their fall. A panoramic synthesis of archaeology, psychology and evolutionary biology, “Goliath’s Curse” is, above all, an attempt to systematize collapse. Reviewers have hailed the book as a skeleton key to understanding societal precarity. Cooper has described it as “a masterpiece of data-driven collapsology.”
Moreover, it is a sobering insight into why our own globalized society feels like it is edging toward the precipice. That’s because, despite all the features that distinguish modern society from empires of the past, some rules hold true throughout the millennia.
Becoming ‘Dr. Doom’
In September, Kemp traveled down from Cambridge to meet me in London for the day. Given his subject, I half-expected a superannuated and eccentric individual, someone like Diamond with his trademark pilgrim-father beard and penchant for European chamber music. But Kemp, 35, would prove to be the antithesis of the anguished catastrophist. The man waiting for me on the concourse at King’s Cross was athletic, swarthily handsome and lantern-jawed. He’d signed off emails regarding our plans to meet with a puckish “Cheerio.”
Kemp’s background is also hardly stereotypical of the bookish scholar. He spent his early years in the dairy-farming town of Bega in New South Wales, Australia, where cattle outnumbered people three-to-one. It was “something of a broken home,” he told me. His father was an active member of the Hell’s Angels, involved in organized crime, a formative presence that would later germinate Kemp’s interest in power dynamics, the way violence is at once a lever for domination and for ruin.
Escaping to Canberra, after high school, Kemp read “interdisciplinary studies” at the Australian National University (ANU), where he found a mentor in the statistical climatologist Jeanette Lindsay. In 2009, it was Lindsay who persuaded him to join a student delegation heading to COP15 in Copenhagen, where Kemp found himself with a front row seat to what he calls “the paralysis of geopolitics.”
At one stage, during a symposium over measures to curb deforestation, he watched his own Australian delegation engage in endless circumlocutions to derail the debate. Representatives from wealthier countries, most notably America, had large teams that they could swap in and out of the floor, enabling them to filibuster vital, potentially existential questions to a deadlock. “If you’re from Tuvalu, you don’t have that privilege,” Kemp explained.
Afterward, Kemp became preoccupied by “a startling red thread” evident in so many spheres of international negotiation: the role of America as arbiter of, and all too often barrier to, multilateral cooperation. Kemp wrote his doctoral thesis on how pivotal issues — such as biodiversity loss, nuclear weapons and climate change — had grown captive to the whims of the world’s great superpower. Later, when he published a couple of academic articles on the same subject, “the ideas weren’t very popular,” he said. “Then Trump got elected, and suddenly the views skyrocketed.”
In 2018, Kemp relocated to the United Kingdom, landing a job as a research affiliate at Cambridge University’s “Centre for the Study of Existential Risk” (CSER, often articulated, in an inadvertent nod to a historical avatar of unalloyed power, to “Caesar”). His brother’s congratulatory present, a 3-D printed, hand-engraved mask of the Marvel character “Dr. Doom,” would prove prophetic. Years later, as Kemp began to publish his theories of societal collapse, colleagues at CSER began referring to him by the very same moniker.
“Goliath hierarchies select for assholes — or, to use Kemp’s preferred epithet, ‘dark triad’ personalities: people with high levels of psychopathy, narcissism and Machiavellianism.”
It was around this time that Kemp read “Against the Grain,” a revisionist history of nascent conurbations by James C. Scott. Kemp had always been an avid reader of history, but Scott’s thesis, which argued that the growth of centralized states “hadn’t been particularly emancipatory or even necessarily good for human wellbeing,” turned some of Kemp’s earlier assumptions about human nature on their head.
Such iconoclastic ideas — subsequently popularized in blockbuster works of non-fiction like Rutger Bregman’s “Humankind” (2019), and “The Dawn of Everything” (2021) by Graeber and Wengrove — would prompt years of research and rumination about the preconditions that enable states and empires to rise, and why they never last forever.
‘Hobbes’ Delusion’
“Goliath’s Curse” opens with a refutation of a 17th-century figure whose theories still cast a long shadow across all considerations of societal fragility. In “Leviathan” (1651), the English philosopher Thomas Hobbes proposed that the social contract was contingent on the stewardship of a central authority — a “Leviathan” designed to keep a lid on humanity’s basest instincts. Political scientists refer to this doctrine as “veneer theory.”
“Once civilization is peeled away, chaos spreads like brushfire,” Kemp surmises. “Whether it be in post-apocalyptic fiction, disaster movies or popular history books, collapse is often portrayed as a Hobbesian nightmare.”
For decades now, the predominant version of history has been beholden to this misanthropic worldview. Many of the most influential recent theories of collapse have echoed Hobbes’ grand theory with specific exemplars. Diamond has famously argued that the society on Rapa Nui, or Easter Island, unraveled due to self-inflicted ecocide before devolving into civil war. That interpretation, in which the islanders deforested the land in the service of ancestor worship, has since been held up as a species-wide admonition — evidence, as researchers John Flenley and Paul Bahn have written, that “humankind’s covetousness is boundless. Its selfishness appears to be genetically inborn.” In “The Better Angels of Our Nature” (2011), Steven Pinker estimated that 15% of Paleolithic people died of violent causes.
But Kemp was struck by a persistent “lack of empirics” undermining these hypotheses, an academic tendency to focus on a handful of “cherry-picked” and emotive case studies — often on islands, in isolated communities or atypical environments that failed to provide useful analogs for the modern world. Diamond’s theories about the demise of Rapa Nui — so often presented as a salutary cautionary tale —have since been debunked.
To further rebut such ideas, Kemp highlights a 2013 study by the anthropologists Jonathan Haas and Matthew Piscitelli of Chicago’s Field Museum. In what amounted to the most comprehensive survey of violence in prehistory, the authors analyzed almost 3,000 skeletons interred during the Paleolithic Era. Of the more than 400 sites in the survey, they identified just one instance of mass conflict: the bones of Jebel Sahaba. “The presumed universality of warfare in human history and ancestry may be satisfying to popular sentiment; however, such universality lacks empirical support,” Haas and Piscitelli wrote.
If there was any truth to the Hobbesian standpoint, the Paleolithic, with its absence of stratified social structures, should have been marked by mass panic and all-out war. Yet the hunter-gatherer period appears to have been a time of relative, if fragile, peace. Instead, conflict and mass violence seemed to be by-products of the very hierarchical organization that Hobbes and his antecedents essentialized. Cave art of armies wielding bows and swords dates only to around 10,000 years ago. “As soon as you start tugging on the threat of collapse, the entire tapestry of history unravels,” Kemp told me.
But if Hobbes was wrong about the human condition — if most people are averse to violence, if mass panic and mutual animosity are not the principal vectors of societal disintegration — what then explains the successive state failures in the historical record? Where or what, to mix metaphors, is Goliath’s Achilles’ heel?
What Fuels Goliath?
In seeking to disentangle a template of collapse from this historiography, Kemp turned to historical data, searching for traits of state emergence and disintegration shared by different polities. “When I see a pattern which needs to be explained, it becomes a fascination bordering upon obsession,” he told me.
A central pillar of his research was the Seshat Global History Databank, an open-source database incorporating more than 862 polities dating back to the early Neolithic. Named after the Egyptian goddess of wisdom, Seshat includes a range of metrics like the degree of centralization and the presence of different types of weaponry; it aggregates these to create nine “complexity characteristics” (CCs), including polity size, hierarchy, governmental framework and infrastructure.
“Wherever Goliath took hold, ‘arms races’ followed, as other status-seeking aspirants jostled for hegemony. And Goliaths were contagious.”
Using this and other sources, Kemp set out to collate his own novel dataset, this time focusing on the common features not of complexity, but of collapse. In keeping with Seshat’s old-god nomenclature, he dubbed it the “Mortality of States” index, shortened to “Moros”, after the Greek god of doom. Covering 300 states spanning the last five millennia, the resulting catalogue is, Kemp claims, “the most exhaustive list of state lifespans available today.”
To some extent, Kemp’s data told a story that has become received wisdom: As Earth thawed out from the last ice age, we entered the Holocene, a period of warmer temperatures and climatic stability. This shift laid the terrain for the first big inflection point: the advent of agriculture, which encouraged our previously itinerant species to settle in place, leading to greater population density and eventually proto-city-states. These early states rose and fell, often condemned by internal conflict, climatic shocks, disease or natural disasters. But gradually the organization of human societies trended toward higher levels of complexity, from the diffuse proto-city-states, through the birth of nations, then empires, to the globalized system of today. The violent paroxysms of the past were merely hiccups on a continuum toward increased sophistication and civility, and perhaps someday immortality. Such is the tale that is commonly framed as the arc of human progress.
But trawling through the data in more detail also revealed unexpected and recurrent patterns, leading Kemp to an early realization: states observably age. “For the first 200 years, they seem to become more vulnerable to terminating. And after 200 years, they stay at a high risk thereafter,” Kemp told me.
The other glaring commonality concerned the structure of these societies. “The common thread across all of them is not necessarily that they had writing or long-distance trade,” Kemp said. “Instead, it’s that they were organized into dominance hierarchies in which one person or one group gains hegemony through its ability to inflict violence on others.”
Kemp argues that dominance hierarchies arise due to the presence of three “Goliath fuels.” The first of these is “lootable resources,” assets that can be easily seen, stolen and stored. In this respect, the advent of agriculture was indisputably foundational. Cereal grains like wheat and rice could be taxed and stockpiled, giving rise to centralized authorities and, later, bureaucracies of the state.
The second Goliath fuel is “monopolizable weapons.” As weaponry evolved from flint to bronze, the expertise and relative scarcity of the source material required for early metallurgy meant that later weapons could be hoarded by powerful individuals or groups, giving those who controlled the supply chain a martial advantage over potential rivals.
The third criterion for Goliath evolution is “caged land,” territories with few exit options. Centralized power is predicated on barriers that hinder people from fleeing oppressive hierarchies.
In Kemp’s telling, every single political entity has grown from one of these seeds, or more commonly, a combination of all three. Bronze Age fiefdoms expanded at the tip of their metal weaponry. “Rome,” Kemp writes, “was an autocratic machine for turning grain into swords,” its vast armies sustained by crop imports from the Nile Valley, its endless military campaigns funded by the silver mines it controlled in Spain. In China, the Han dynasty circumscribed its territory with its Great Wall to the north, intended both to keep Xiongnu horseback raiders out and the citizenry in. Europe’s colonial empires were built, in Diamond’s famous summation, by “Guns, Germs and Steel.”
For millennia, the nature of forager societies kept these acquisitive impulses to some extent contained, Kemp argues. The evolutionary logic of hunting and gathering demanded cooperation and reciprocity, giving rise to “counter-dominance strategies”: teasing, shaming or exile. With the advent of Goliath polities, however, the “darker angels of our nature” were given free rein, yielding social arrangements “more like the dominance hierarchies of gorillas and chimpanzees.”
“Rather than a stepladder of progress,” Kemp writes, “this movement from civilization to Goliath is better described as evolutionary backsliding.” Moreover, Goliaths “contain the seeds of their own demise: they are cursed. This is why they have collapsed repeatedly throughout history.”
In Kemp’s narrative, our retrograde rush toward these vicious social structures has been less about consensus than the relentless ascent of the wrong sort of people. Goliath hierarchies select for assholes — or, to use Kemp’s preferred epithet, “dark triad” personalities: people with high levels of psychopathy, narcissism and Machiavellianism. Consequently, history has been shaped by pathological figures in the Narmer mold, dominance-seekers predisposed to aggression. Reinforced by exceptionalist and paranoid ideologies, these strongmen have used violence and patronage to secure their dominion, whether driven by a lust for power or to avenge a humiliation. Several of the rebellions that plagued dynastic China, Kemp points out, were spearheaded by aggrieved people who failed their civil service examinations.
“Whether societies collapsed through gradual depopulation, like Çatalhöyük, or abruptly, as with Teotihuacan’s conflagration, Kemp argues that the triggers were the same.”
Wherever Goliath took hold, “arms races” followed, as other status-seeking aspirants jostled for hegemony. And Goliaths were contagious. The growth of “one bellicose city-state” would often produce a domino effect, in which the threat of an ascendant Goliath would provoke other regional polities to turn to their own in-house authoritarian as a counterweight to the authoritarian next door.
In this way, humankind gravitated “from hunting and gathering to being hunted and gathered,” Kemp writes. Early states had little to distinguish them from “criminal gangs running protection rackets.” Many of the great men of history, who are often said to have bent society to their will, Kemp told me, are better thought of as “a rollcall of serial killers.”
The 1% View Of History
Back downstairs, on the British Museum’s ground floor, we walked into a long gallery off the central atrium containing dozens of megalithic totems from the great ages of antiquity. The giant granite bust of Rameses II sat beatific on a pediment, and visitors peered into a glass cabinet containing the Rosetta Stone. Kemp, slaloming through the crowds, murmured: “The 1% view of history made manifest.”
Along both walls of an adjacent corridor, we came upon a series of bas-reliefs from the neo-Assyrian city of Nimrud, in modern-day Iraq. Depicting scenes from the life of the Ashurnasirpal II, who ruled Nimrud in the 9th century BCE, the gypsum slabs were like an artistic expression of Kemp’s historical themes: Ashurnasirpal sitting on a throne before vassals bearing tribute; Ashurnasirpal surrounded by protective spirits; Ashurnasirpal’s army ramming the walls of an enemy city, rivals dragging themselves along the ground, backs perforated with arrows. The entire carving was overlaid with cuneiform script, transcribed onto signage below, with sporadic sentences translated into English: “great king, strong king, king of the universe. … Whose command disintegrates mountains and seas.”
Across the atrium, in a low-lit room containing a bequest from the Rothschild family’s antique collection, Kemp lingered over an assortment of small wooden altarpieces, with biblical scenes and iconography carved in minuscule, intricate detail. Elite status could be projected in the imposing size of a granite statue, he said. But it could just as well be archived in the countless hours spent chiseling the Last Supper into a fragment of boxwood.
It is, of course, inevitable that our sense of history is skewed by this elite bias, Kemp explained. While quotidian objects and utensils were typically made of perishable materials, the palaces and monuments of the governing class were designed to be beautiful, awe-inspiring and durable. In the hours that we spent on the upper floors, we spied just one relic of ordinary life: a 3,000-year-old wooden yoke from Cambridgeshire.
Likewise, early writing often evolved to reinforce the “1% view of history” and formalize modes of control. The predominance of this elite narrative has produced a cultural blind spot, obscuring the brutality and oppression that has forever been the lot of those living at the base of a pyramid, both figurative and actual.
From all this aristocratic residue, Kemp sought to extract a “people’s history of collapse” — some means of inferring what it was like to live through collapse for the average person, rather than the elites immortalized in scripture and stone.
The Curse Of Inequality
If Kemp’s research revealed that historical state formation appears to follow a pattern, so, too, did the forces that inexorably led toward their demise. To illustrate how the process works, Kemp provides the example of Çatalhöyük, a proto-city that arose on the Konya Plain in south-central Turkey around 9,000 years ago, one of thousands of “tells,” mounded remnants of aborted settlements found throughout the Near East.
Excavations of the site’s oldest layers suggest that early Çatalhöyük was notable for its lack of social differentiation. Crammed together in a dense fractal of similarly sized mud-brick dwellings, the settlement in this period exhibits no remnants of fortification and no signs of warfare. Analysis of male and female skeletons has shown that both sexes ate the same diet and performed the same work, indicating a remarkable degree of gender equity.
This social arrangement, which the Stanford archaeologist Ian Hodder has described as “aggressively egalitarian,” lasted for around 1,000 years. Then, in the middle of the 7th millennium BCE, the archaeological record starts to shift. House sizes begin to diverge; evidence of communal activity declines. Later skeletal remains show more evidence of osteoarthritis, possibly betraying higher levels of workload and bodily stress. Economists have estimated that the Gini coefficient, which measures disparities in household income, doubled in the space of three centuries — “a larger jump than moving from being as equal as the Netherlands to as lopsided as Brazil,” Kemp writes. Within a few centuries, the settlement was abandoned.
“In almost every case, [societal] decline or collapse was foreshadowed by increases in the appearance of proxies of inequality.”
The fate of Çatalhöyük established a template that almost every subsequent town, city-state and empire would mirror. Its trajectory resounds throughout the historical record and across continents. Similar patterns can be discerned from the remnants of the Jenne-Jeno in Mali, the Olmecs of Mesoamerica, the Tiwanaku in Titicaca, and the Cahokia in pre-Columbian North America.
Occasionally, the archaeological record suggests a fluctuation between equality and disparity and back again. In Teotihuacan, near today’s Mexico City, the erection of the Feathered Serpent Pyramid by an emergent priestly class in around 200 CE ushered in a period of ritual bloodletting. A more egalitarian chapter followed, during which the temple was razed, and the city’s wealth was rechanneled into urban renewal. Then the old oligarchy reasserted itself, and the entire settlement, beset by elite conflict or popular rebellion, was engulfed in flames.
Whether societies collapsed through gradual depopulation, like Çatalhöyük, or abruptly, as with Teotihuacan’s conflagration, Kemp argues that the triggers were the same. As Acemoğlu and Robinson explored in “Why Nations Fail” (2012), the correlation between inequality and state failure often rests on whether its institutions are inclusive, involving democratic decision-making and redistribution, or extractive: “designed to extract incomes and wealth from one subset of society to benefit a different subset.” Time and again, the historical record shows the same pattern repeating — of status competition and resource extraction spiraling until a tipping-point, often in the shape of a rebellion, or an external shock, like a major climate shift or natural disaster, which the elites, their decision-making fatally undermined by the imperative to maintain their grip on power, fail to navigate.
In almost every case, decline or collapse was foreshadowed by increases in the appearance of proxies of inequality. A rise in the presence of large communal pots indicates an upsurge in feasting. Deviation in the size of dwellings, preserved in the excavated footprints of early conurbations, is a measure of social stratification, as wealth accumulates among the elite. Graves of that same nobility become stuffed with burial goods. Great monuments, honoring political and religious leaders or the gods who were supposed to have anointed them, proliferate. Many of the most lucrative lootable resources throughout history have been materials that connote elevated social standing, an obsession with conspicuous consumption or “wastefully using resources,” that marked a break from the hunter-gatherer principle of taking only what was needed. (Kemp wears a reminder of the human compulsion to covet beauty as much as utility, an obsidian arrowhead, on his wrist.)
All the while, these signs of burgeoning inequality have tended to be twinborn with an increasing concentration of power, and its corollary: violence. War, often instigated for no more reason than the pursuit of glory and prestige, was just “the continuation of status competition by other means,” Kemp writes. On occasion, this violence would be manifested in the ultimate waste of all: human sacrifice, a practice custom-made to demonstrate the leadership’s exceptionality — above ordinary morality.
Better Off Stateless
As Kemp dug into the data in more detail, his research substantiated another startling paradox. Societal collapse, though invariably catastrophic for elites, has often proved to be a boon for the population at large.
Here again, Kemp found that the historiography is subject to pervasive and fallacious simplifications. In his book, he repudiates the 14th-century Tuscan scholar Petrarch, who promulgated the notion that the fall of classical Rome and Greece ushered in a “dark age” of cultural atrophy and barbarism. His was a reiteration of sentiments found in many earlier examples of “lamentation literature,” left behind on engraved tablets and sheaves of papyrus, which have depicted collapse as a Gomorran hellscape. One of Kemp’s favorites is the “Admonitions of Ipuwer,” which portrays the decline of Egypt’s Old Kingdom as a time of social breakdown, civil war and cannibalism. “But it actually spends a lot more time fretting about poor people becoming richer,” he said.
In reality, Kemp contends, Petrarch’s “rise-and-fall vision of history is spectacularly wrong.” For if collapse often engulfed ancient polities “like a brushfire,” the scorched earth left behind was often surprisingly fertile. Again, osteoarcheology, the study of ancient bones, gives the lie to the idea that moments of societal disintegration always spelled misery for the population at large.
Take human height, which archaeologists often turn to as a biophysical indicator of general health. “We can look at things like did they have cavities in their teeth, did they have bone lesions,” Kemp explained. “Skeletal remains are a good indicator of how much exercise people were getting, how good their diet was, whether there was lots of disease.”
“Societal collapse, though invariably catastrophic for elites, has often proved to be a boon for the population at large.”
Prior to the rise of Rome, for example, average heights in regions that would subsequently fall under its yoke were increasing. As the empire expanded, those gains stalled. By the end of the Western Empire, people were eight centimeters shorter than they would have been if the preceding trends had continued. “The old trope of the muscle-bound Germanic barbarian is somewhat true. To an Italian soldier, they would have seemed very large,” Kemp said. People in the Mediterranean only started to get taller again following Rome’s decline. (In a striking parenthesis, Kemp points out that the average male height today remains two centimeters shorter than that of our Paleolithic forebears.)
Elsewhere, too, collapse was not necessarily synonymous with popular immiseration. The demise of the extravagant Mycenaean civilization in Greece was pursued by a cultural efflorescence, paving the way for the proto-democracy of Athens. Collapse could be emancipatory, freeing the populace from instruments of state control such as taxes and forced labor. Even the Black Death, which killed as much as half of Europe’s population in the mid-14th century, became in time an economic leveler, slashing inequality and accelerating the decline of feudalism.
It’s a pattern that can still be discerned in modern contexts. In Somalia, the decade following the fall of the Barre regime in 1991 would see almost every single indicator of quality of life improve. “Maternal mortality drops by 30%, mortality by 24%, extreme poverty by 20%,” Kemp recounted from memory. Of course, there are endless caveats. But often, “people are better off stateless.”
Invariably, however, Goliaths re-emerged, stronger and more bureaucratically sophisticated than before. Colonial empires refined systems of extraction and dominance until their tentacles covered diffuse expanses of the globe. Kemp, never shy of metaphor, calls this the “rimless wheel,” a centripetal arrangement in which the core reaps benefits at the margins’ expense.
At times, such regimes were simply continuations of existing models of extraction. In 1521, when the Spanish conquistador Hernán Cortés unseated the Aztec ruler Moctezuma II, it was merely a case of “translatio imperi” — the handing over of empire. The European imperial projects in the Americas were an unforgivable stain, Kemp said. But, more often than not, they assumed the mantle from pre-existing hierarchies.
Endgame
In the afternoon, we walked north from the British Museum over to Coal Drops Yard, formerly a Victorian entrepôt for the import and distribution of coal, now a shiny vignette of urban regeneration. The morning rain had cleared, and Granary Square was full of tourists and office workers enjoying the late summer sun. Kids stripped to their underwear and played among low fountains; people chatted at public tables beneath a matrix of linden trees. Kemp and I found an empty table and sat down to talk about how it could all fall apart.
As “Goliath’s Curse” approaches its conclusion, the book betrays a sense of impending doom about our current moment. The final section, in which Kemp applies his schema to the present day, is entitled “Endgame,” after the stage in chess where only a few moves remain.
Today, we live in what Kemp calls the “Global Goliath,” a single interconnected polity. Its lootable resources are data, fossil fuels and the synthetic fertilizers derived from petrochemicals. Centuries of arms races have yielded an arsenal of monopolizable weapons like autonomous drones and thermonuclear warheads that are “50 trillion times more powerful than a bow and arrow.” The land — sectored into national borders, monitored by a “stalker complex” of mass surveillance systems and “digital trawl-nets” — is more caged than ever.
We have reached the apotheosis of the colonial age, a time when extractive institutions and administrative reach have been so perfected that they now span the globe. However, the resulting interdependencies and fetishes for unending growth have created an ever-growing catalog of “latent risks,” or accumulated hazards yet to be realized, and “tail risks,” or outcomes with a low probability but disastrous consequences. Kemp characterizes this predicament, in which the zenith of human achievement is also our moment of peak vulnerability, as a “rungless ladder.” The higher we go, the greater the fall.
“We have reached the apotheosis of the colonial age, a time when extractive institutions and administrative reach have been so perfected that they now span the globe.”
Under a series of apocalyptic subtitles — “Mors ex Machina,” “Evolutionary Suicide,” “A Hellish Earth” — Kemp enumerates the existential threats that have come to shape the widespread intuition, now playing out in our geopolitics, that globalized society is sprinting toward disaster. After the post-Cold War decades of non-proliferation, nuclear weapons stockpiles are now growing. The architects of artificial intelligence muse about its potential to wipe out humanity while simultaneously lobbying governments to obstruct regulation. Our densifying cities have become prospective breeding grounds for doomsday diseases. Anthropogenic climate change now threatens to shatter the stability of the Holocene, warming the planet at “an order of magnitude (tenfold) faster than the heating that triggered the world’s greatest mass extinction event, the Great Permian Dying, which wiped away 80–90% of life on earth 252 million years ago,” Kemp warns.
The culprits in this unfolding tragedy are not to be found among the ranks of common people. The free market has always been predicated on the concept of Homo economicus, a notional figure governed by dispassionate self-interest. But while most people don’t embody this paradigm, we are in thrall to political structures and corporations created in that image, with Dark Triad personalities at the wheel. “The best place to find a psychopath is in prison,” Kemp told me. “The second is in the boardroom.”
Now, deep into the Global Goliath’s senescence, several of the indicators that Kemp identifies as having historically presaged collapse — egalitarian backsliding, diminishing returns on extraction, the rise of oligarchy — are flashing red. Donning his risk analyst hat, Kemp arrives at the darkest possible prognosis: The most likely destination for our globalized society is “self-termination,” self-inflicted collapse on a hitherto unprecedented scale. Goliath is more powerful than ever, but it is on a collision course with David’s stone.
Lootable Silicon
All of this seemed hard to reconcile with the atmosphere of contented civility in Granary Square on this sunny September afternoon. I proposed that an advocate for global capitalism would doubtless view our current circumstances as evidence of the Global Goliath’s collective, trickle-down bounty.
“We should be thankful for a whole bunch of things that started, by and large, in the Industrial Revolution,” Kemp said. “Vaccines, the eradication of smallpox, low infant mortality and the fact that over 80% of the population is literate. These are genuine achievements to be celebrated.”
Kemp argued that most redistribution has been a product of “stands against domination”; for example, the formation of unions, public health movements and other campaigns for social justice. Meanwhile, underlying prosperity still depends on the rimless wheel: the hub exploiting the periphery. “If we were here 150 years ago, we’d be seeing child laborers working in these courtyards,” he said, gesturing at the former coal warehouses that are now an upmarket shopping mall and that once served as a nerve center of the fossil fuel industry that built the modern age.
The same dynamics hold sway today, albeit at a further remove. Just south of us, across the Regent’s Canal, sat the London headquarters of Google, a billion-dollar glass edifice. At first glance, Kemp gave the building an enthusiastic middle finger.
Later, he explained: “The people sitting in that building are probably having a pretty good time. They have lots of ping pong tables and Huel. But the cobalt that they’re using in their microchips is still often dug up by artisanal miners in the Democratic Republic of Congo, getting paid less than a couple of dollars a day.”
Like much of the oligarchic class, the boy-gods of Silicon Valley still cleave to Hobbesian myths to justify their grip on wealth and power. Their techno-Utopian convictions, encapsulated in Bill Gates’ mantra that “innovation is the real driver of progress,” are merely a secular iteration of the divine mandates that Goliaths once used to legitimize their rule. Promises of rewards in the afterlife have been supplanted by dreams of a technological singularity and interplanetary civilization.
Another plausible eventuality, which Kemp dubs the “Silicon Goliath,” is a future in which democracy and freedom are crushed beneath the heel of advanced algorithmic systems. He is already at work on his next book about the evolution of mass surveillance, an inquiry that he told me “is in many ways even more depressing.”
Slaying Goliath
Toward the end of “Goliath’s Curse,” Kemp imagines a scenario in which the decision of whether to detonate the Trinity atomic bomb test in New Mexico in 1945 was made not by a Department of War but by a “Trinity jury,” an assembly of randomly selected members of the public.
“Now several of the indicators that Kemp identifies as having historically presaged collapse — egalitarian backsliding, diminishing returns on extraction, the rise of oligarchy — are flashing red.”
In such a counterfactual, with the Nazis defeated, Japan already inches from surrender and Manhattan Project physicists warning of a non-zero possibility that the test could ignite the whole atmosphere and exterminate all life on Earth, Kemp contends that a more inclusive decision-making process would have changed the course of history. “If you had a random selection by lottery of 100 U.S. citizens and asked them, ‘Should we detonate the bomb?’ What decision do they come to? Almost certainly ‘No,’ he told me.
As Kemp sees it, the widespread adoption of such open democracy is the only viable route to escape the endgame. These citizen juries wouldn’t be free-for-alls, where the loudest or most outrageous voice wins, but deliberative procedures that necessitate juror exposure to expert, nonpartisan context.
Such assemblies wouldn’t be enough to “slay Goliath” on their own, Kemp told me. “Corporations and states … [must] pay for the environmental and social damages they cause … to make the economy honest again.” Per capita wealth, Kemp added, should be limited to a maximum of $10 million.
I challenged Kemp that this wish-list was beginning to sound like a Rousseauvian fever-dream. But seven years immersed in the worst excesses of human folly had left him in no mood for half-measures. “I’m not an anarcho-primitivist,” he said. There was no point trying to revivify our hunter-gatherer past. “We’d need multiple planet Earths!” Kemp conceded. And yet the urgency of our current circumstances demanded a radical departure from the existing status quo, and no less a shift in mindset.
His final demotic prescription, “Don’t be a dick,” was an injunction to everyone that our collective future depends as much on moral ambition as political revolution. Otherwise, Goliath won’t be just a Bible story. It could also be our epitaph.
I sent a shorter version of this newsletter as the last issue of the (the entire walk is) Between Two Mountains pop-up newsletter. This version is quite expanded. The archives of the B2M pop-up are available to SPECIAL PROJECTS members on the members’ site. And if you’d like to follow along in my digital footsteps, the GPX file of the walk is here.
This talk was given on October 14, 2025 at Kinference in Brooklyn, New York.
Spoiler alert: the last part of the talk covers plot points of the movie Spirited Away. Another warning is included right before the spoilers with a jump forward link to the spoiler-free conclusion.
I am so tired of hearing about AI. Unfortunately, this is a talk about AI.
I’m trying to figure out how to use generative AI as a designer without feeling like shit. I am fascinated with what it can do, impressed and repulsed by what it makes, and distrustful of its owners. I am deeply ambivalent about it all. The believers demand devotion, the critics demand abstinence, and to see AI as just another technology is to be a heretic twice over.
Today, I’d like to try to open things up a bit. I want to frame the technology more like an instrument, and get away from GenAI as an intelligence, an ideology, a tool, a crutch, or a weapon. I find the instrument framing more appealing as a person who has spent decades honing a set of skills. I want a way of working that relies on my capabilities and discernment rather than something so amorphous and transient as taste. (If taste exists in technology, it needs to be smuggled in.)
John Coltrane and J Dilla
Thinking of AI as an instrument recenters the focus on practice. Instruments require a performance that relies on technique—the horn makes the sound, but how and what you blow into it matters; the drum machine keeps time and plays the samples, but what you sample and how you swing on top of it becomes your signature.
In other words, instruments can surprise you with what they offer, but they are not automatic. In the end, they require a touch. You use a tool, but you play an instrument. It’s a more expansive way of doing, and the doing of it all is important, because that’s where you develop the instincts for excellence. There is no purpose to better machines if they do not also produce better humans.
I believe artists have more helpful things to say than the money guys about how to use and creatively misuse technology. So today, I want to share four artists, and I hope you’ll leave with some more flexibility in how to collaborate with the machine in your own work, creative or otherwise.
First, some background. Over the summer, the vibe around AI seemed to shift.
All those aggressive predictions about the destruction of knowledge work didn’t hit their six-month deadline. The much anticipated GPT-5 felt more like an incremental step up from GPT-4, signaling LLMs have probably moved past the revolution phase and into optimization and evaluating trade-offs. The fact that OpenAI is encouraging everyone to fantasize about hardware with the Jony Ive annoucement tells me the software side may not have enough headroom for the profits they need. Meanwhile, small, local models are good enough for a surprising number of use cases, while being cheaper, more private, and energy efficient.
Progress has slowed, projections are being walked back, and the science is starting to look more hazy. We may not be in AI winter, but I am hoping for an AI autumn. Autumn is amazing; the air cools, the mania of summer dissipates, things slow down. Right now, the changes in AI feel incremental enough to start laying down strategy. This means we can think about our approach with steadier footing instead of vacillating in response to the hype, whether it comes from the top as LinkedIn posts by prepper CEOs or from the bottom by the hustleheads on X.
The hype is expected—new tech runs on speculation. You can feel the residue of the last 30 years of booms. There is a sense that people missed their chance to get rich on the internet, on ecommerce, on the app store, on social media, on crypto, on meme stocks, on NVIDIA. The hype bubbles get inflated because individuals don’t want to miss their chance at another windfall, and companies don’t want to get displaced by any nascent technological shifts. The history of tech has calcified into stories of dramatic wins and unforeseen downfalls, and what results is a tech culture of near compulsory participation in prediction rather than creating value or serving needs.
The TV show The Wire had a phrase about how people navigate the risk of failure inside more traditional institutions: “You can’t lose if you don’t play.” In the tech world, the logic reverses. The drawbacks of a collective fallacy are smaller than not participating in the next innovation, so the rule becomes, “You only lose if you don’t play.”
In other words, the push to join the AI rush comes from the sense that it’s the only game in town right now, and everyone else is already playing. The hype gets louder, and the bubble gets bigger.
What surprised me wasn’t the AI hype, though, but the lack of solidarity that came with it. Faced with the story of AI labor displacement, our first instinct as technology workers wasn’t to protect one another, but to search for ways to use the tools to replace our collaborators.
The fractures fell neatly along disciplines: engineers using AI to wish away designers, designers wishing away engineers, product managers wishing away both. In this climate, AI becomes frenemy identification technology, another way to avoid working together. It’s always easier to grab a tool and bypass the mess of coordination, even if that means doing more and doing it alone. AI lowers the barrier to working outside your lane, and sure, that could mean more overlap between disciplines, but right now, we’re mostly boxing each other out or stepping on one another’s toes.
With collaboration already strained, it’s no surprise that we fall back on individual effort. But individual work, disconnected from the whole and accelerated by automation, only makes the turbulence worse and the course corrections more violent. The deeper problem is that companies still haven’t figured out how to mass-produce orientation, so workers get thrashed around in the speed and scale of the system.
When coordination breaks down, the fantasy of self-sufficiency rushes in to fill the gap. GenAI, after all, is built as an individual technology, whether it is expressed as a one-on-one chat, the fantasy of the one-person startup, or the sycophantic assistant set up to glaze you. It says “just get it done, no skills needed.”
All you need is a prompt, a dream, and some vibes. And, of course, we’ve chosen an individual to be the face of that.
“Money is everywhere but so is poetry. What we lack are the poets.”
Federico Fellini
Vibe coding has a mascot, and it is the music producer Rick Rubin. A bit of background on him before we get back to vibe coding.
Rubin got his start in the music industry as the co-founder of Def Jam Records. He helped bring hip-hop into the mainstream by producing records for LL Cool J, Run-D.M.C., and the Beastie Boys. Later, Rubin produced albums across genres for Jay-Z, Johnny Cash, Red Hot Chili Peppers, and plenty of others. He’s credited with shaping the sound of the last forty years without ever learning to play an instrument. In interviews, Rubin leans into the irony: he can’t operate the board, can’t engineer, can’t really play a guitar.
Rubin’s supposed lack of technical skill has made him an easy fit as the poster child for vibe coding, the notion that with the right sensibility and some prompting, you can steer technical output without ever knowing how to use the tools.
Anthropic played into this with The Way of Code, a project where Rubin rewrote the Tao Te Ching, an ancient Chinese philosophical text, to be about vibe coding.
It was a marketing stunt, but I read it as a proposal that the path to enlightenment is moving away from competency and towards the vibes.
Here’s one of the chapters from the site:
The Way of Code #47
Without going outside,
you may know the whole world.
Without looking through the window,
you may see the ways of heaven.
The further you travel, the less you know.
The more you know, the less you understand.
Therefore, The Vibe Coder
knows without going, sees without looking,
and accomplishes all
without doing a thing.
Ugh.
When I first saw this, I thought “Did anyone read this?” That’s always the question when encountering something that you think is god awful. But in the AI era, I found myself asking a follow-up question: “Did anyone write this?”
It’s a complicated knot. Art is interesting partly because someone put effort into making this thing, not all the others they could have made. But artists also know that the best ideas can appear whole, as if they made themselves. Effort isn’t what makes the work meaningful, yet it still feels like a small violation to release an AI’s output untouched.
I figured they used AI to write revised chapters of the Tao since the verses are shown beside some vibe coded visuals. But in interviews Rubin says he wrote them himself. I’ll be generous and say that it’s a joke that doesn’t land for me.
Like most ideas that spread like memes, “vibe coding” got carried way too far. Karpathy’s original post was modest: a fun hack for engineers, and “not too bad for weekend throwaway projects.” Lightweight, bounded, disposable work, perfect for prototyping, experiments, scripting, and personal software. If this is the scope, I’m on board, giddyup.
But once the phrase caught on, people started applying it everywhere, stretching a hack into a philosophy worthy enough to be inserted into ancient texts.
Like a lot of things with AI, it feels completely out of proportion. Maybe that’s the humor of it. But in the process, Rick Rubin became valorized as proof you don’t need skills. It’s as if wu wei, the principle of not forcing at the heart of the Tao, had been hollowed out and recast as dependency. Use the machine, no skills or knowledge needed.
Rubin obviously has skills and knowledge, but we get two Rick Rubins. There’s the one in the studio, unable to play guitar, but gifted at guiding artists and clarifying what they want. And then there’s the cartoon Rubin, who leans into his lack of ability in interviews, writes bad poems, and poses as the guru. Maybe that split is what happens when the work becomes too abstract from execution.
Lately, I’ve been thinking about my use of AI as a kind of spatial relationship. Where do I stand in relation to the machine—above it, beside it, under it? Each position carries a different kind of power dynamic. To be above is to steer, beside is to collaborate, below is to serve.
In the Rubin vibe arrangement, you’re under the machine and dependent on it. Without skills, the model’s limits become your own, no different from anyone else typing wishes into a text box. Take what the AI gives without question and you’re not producing, you’re consuming. Eventually, that passivity gets used against you. We’ve seen it before: streaming services flattened art into algorithmic averages and background noise, newsfeeds rewired attention toward outrage because engagement meant growth. Each time, the machine attracted us with individual customization while quietly taking control of the terms. The same will happen with AI if used for passive consumption, even if that consumption is dressed up as execution.
So while vibe coding may be useful for short-term work, it’s not a suitable approach for anything intended to last longer than a tub of yogurt. Time saved is not strength gained, so I went looking for other examples about how to work with the machine.
“It could be uplifting to watch a person energetically building a beautiful coffin. And depressing to watch someone sloppily and carelessly make the worst birthday cake ever.”
George Saunders
I’ve spent the last couple of months digging into Brian Eno’s work and music.
Eno began as a keyboardist in the band Roxy Music in the early ’70s, helping to popularize the use of synthesizers in pop music. Then he went solo, and later built a career producing some of the most important bands of the last fifty years—David Bowie, Devo, Talking Heads, U2, and, more recently, collaborating with Fred Again.
Eno is rarely the virtuoso; instead, he’s the collaborator, the systems thinker, the one who turns the studio into a laboratory. What makes Eno especially relevant for me is his work beyond the songs. His impact as an artist has mostly been to define new forms, supply vocabulary, and arrange the vibe.
Rather than a pre-arranged composition, his music drifts and breathes, built from overlapping loops of different lengths that phase in and out of sync with one another, creating slow variations he can’t fully predict.
Eno called it “a river of sound,” always the same and never the same. In that sense, ambient music was his first experiment with what he’d later name generative art: systems that grow and change within carefully set constraints.
Since the ’80s, Eno has been interested in using software and systems that produce music and visual art. In the last decade, he carried this idea into the phone with generative apps for music-making.
Bloom lets you seed melodies by tapping dots that ripple outward.
Scape has you place abstract elements into a pictorial landscape to create a landscape of sound
And Reflection dissolves the boundaries of a Brian Eno album by shipping generative software that allows endless music.
I believe that if you’re looking for a music producer to give some inspiration on how to work with machines, you’d have better luck and more ideas to consider with Brian Eno than Rick Rubin. The difference, to me, comes down to placement.
If working in the Rubin style puts you under the machine, Eno works beside it. He sets up a system with his inputs and samples, then listens, selects, and continues to shape. Eno often says while making music he feels like a gardener: planting loops and textures, then watching them sprout into something unexpected with the potential to become incredibly beautiful with a little bit of care and pruning. The machine may produce material, but the job of shaping it into something meaningful still rests with him. It is creativity as cultivation.
The value of the machine’s output depends on how we see it, and our interpretation often has little to do with its technical perfection. A flawless, virtuoso output from GenAI can feel lifeless, while something raw or broken might have something interesting about it. That’s why I like to write bad and contradictory prompts, because they feel like they are more aligned with how these models actually function. The models aren’t deterministic; we don’t fully understand how their associations form or why certain patterns appear. So why not let them drift into ambiguity and see what happens? I wouldn’t want an irregular AI in my bank app, but in a creative workflow, hallucinating feels like the point of it all.
In a sense, Brian Eno was tinkering with prompts long before we had the word. In 1975, he and Peter Schmidt created Oblique Strategies, a deck of cards meant to shake artists out of their habits. Each carried a short phrase to be interpreted and followed: “Honor thy error as a hidden intention.” “Use an old idea.” “Work at a different speed.” These weren’t instructions so much as provocations—small reframings that opened space for something unexpected to emerge.
That, to me, feels like a better model for prompting in creative work, whether the first act of execution belongs to a person or a machine. A good prompt doesn’t need to function like a blueprint. They can also behave like a horoscope or a fortune.
The results from ambiguous prompts can be weird, illogical, and incredibly stimulating, because the system is not scared of being wrong. The good stuff is at the technological edge, even if it’s a bit shit. Eno put it beautifully in 1995:
“Whatever you find weird, ugly, uncomfortable, and nasty about a new medium will surely become its signature. CD distortion, the jitteriness of digital video, the crap sound of 8-bit, all of these will be cherished and emulated as soon as they can be avoided.”
Time has proven him right. This idea explains the return of vinyl, the revival of iPods, Gen Z filming parties on old digital camcorders, or why 8mm film is nostalgic for Boomers in the same way the blown-out photos from the first iPhone make us Millennials sentimental. And if I had to make a guess, we will miss the awkward trickle-in of words seen in modern LLMs.
In other words, every new technology promises better clarity, yet its essence is determined by the noise it produces. The friction of limits is what gives a technology its character, so when a system becomes too smooth, too all-encompassing, or too accommodating, it stops having a signature at all. Here’s Eno again from earlier this year:
“I can see from the little acquaintance that I have with using AI programs to make music, that what you spend nearly all your time doing is trying to stop the system becoming mind numbingly mediocre. You really feel the pull of the averaging effect of AI, given that what you are receiving is a kind of averaged out distillation of stuff from a lot of different sources.”
An average email or line of code is fine. Average art isn’t. To make something alive with AI, we have to resist its pull towards average by working beside it, shaping what it gives, and listening for what’s missing. Sometimes what’s needed is a good, old-fashioned mistake or two.
Another answer is not to cultivate the machine’s output but to compose through it, treating the model as material, choosing the inputs and shaping the rules. In other words, not working beside the machine, but stepping inside it.
“By the late twentieth century, our time, a mythic time, we are all chimeras, theorized and fabricated hybrids of machine and organism.”
Donna Haraway
If Brian Eno shows us what it means to work beside the machine and cultivate its outputs, Holly Herndon and Mat Dryhurst show us what it means to move into it.
Herndon and Dryhurst are visual artists and musicians who, for over a decade, have worked with AI and treated digital systems as instruments for rethinking music itself. They’ve built tools and trained models that show what AI can offer creative practice as a means of integration, extension, and amplification of the artist.
They’ve founded Spawning, which they call the consent layer for AI, letting artists decide whether to opt in or out of model training. Alongside it, they launched a beta for Public Diffusion, a foundation model trained entirely on 30 million public-domain images, designed to make generative work copyright-safe.
xhairymutantx
Then there’s xhairymutantx from 2024, shown at the Whitney Biennial, a project where they trained a text-to-image model on photos of Holly, then opened it to the public, inviting others to create with it and explore how identity becomes warped and extended inside generative systems.
And most recently, they created The Call, a choral AI project involving choirs across the UK, where recorded voices become a shared dataset, and the resulting models are folded into a spatial audio installation performing generative choral arrangements. Serpentine Gallery acts as steward of a data trust that governs usage, ensuring the participating choirs are paid and have agency over how their voices are used.
The compositions are inspired by Medieval music, Renaissance music, and hymns. The vocal performances were made from a combination of the choral data and Holly’s own song data. Let’s listen for a minute:
One of the things you notice when encountering Herndon and Dryhurst’s work is that they are just as concerned with the administrative structures needed to serve artists as they are with the creative potentials of new technology. They say that all media is training data, so their work wrestles with the implications of what media generation at scale means for artists. The AI model and the economic model don’t need to come packaged together. Both can be areas for innovation, and severing the implied connection may be a requirement for ethical AI.
What emerges from this approach isn’t just art, but a reimagining of the conditions that make art possible. When the system is designed to respect artists, scale becomes a tool rather than a threat. It opens up new questions. What new forms appear at that scale? The story can be rethought; instead of fewer people making the same amount, what about the same number of people making stranger, more abundant, more connected work?
As the tools evolve, the metaphors we use to understand them must also be updated. Herndon and Dryhurst describe this next phase as the move from sampling to spawning. Sampling was the logic of the 20th century. You took a slice of a record—a James Brown breakbeat, a horn stab from a jazz LP—and folded it into a new track. It was transformative, but you knew the source, and you could trace the lineage.
Spawning is different. Instead of lifting fragments, you train a model on an artist’s entire body of work and generate new material in their style. Clear lineage, but fuzzy origins. Sampling dealt in citation. Spawning touches the DNA. This distinction matters because spawning raises the stakes in ways that sampling never did. When your work trains a model, what’s taken isn’t a note or a beat, but the sensibility and perspective of your practice.
Sampling sparked arguments about ownership and credit; spawning resets the terms. The internet challenged copyright by creating infinite distribution of perfect copies. With AI, what happens when infinite distribution is hooked up to infinite imitation?
You’re so wonderful
Too good to be true
You make me, make me
Make me hungry for you
Why Can’t I Be You? by The Cure
I opened this talk with a tone-deaf marketing moment from an AI company, so it feels fitting to end with another. Last March, OpenAI launched image generation in ChatGPT and encouraged people to upload selfies and turn themselves into anime characters.
The results were unmistakable: the soft light, round faces, and watercolor skies of Hayao Miyazaki and Studio Ghibli. Within hours, social media flooded with Ghibli-fied images—sweet, strange, and grotesque.
The irony cut deep. Miyazaki’s films are animated by hand. They require over 150,000 drawings per movie, and now his studio’s style was being used to sell the very shortcuts and automations he’s spent a career standing against. The model ate Miyazaki’s work along with everything else, so OpenAI can puppet his style when it suits their needs. Last year, they took Scarlett Johansson’s voice to promote text to speech. Next year, they’ll line up a new artist.
Enough has been said about the viral moment at this point, so instead of staying with the style discussion, I want to value Miyazaki’s art and think about the substance of his movies.
To finish the talk, I’d like to take a look at Spirited Away, a film that wrestles with identity and imitation, appetite and satisfaction. While there’s always a risk in reading too much allegory into art, it seems fair to seek wisdom in what moves us.
Spirited Away begins with a crossing. Driving through the countryside, Chihiro, a ten-year-old girl, and her parents pass a torii gate, a marker used to signify temples.
They park and wander into what looks like an abandoned amusement park. Chihiro and her parents don’t realize it yet, but the torii gate they passed marked a threshold. They’ve entered the spirit realm.
When they find an unattended food stall, her parents sit down and eat, promising to pay later. But the food isn’t for sale; it was prepared as an offering for the gods. They take what isn’t theirs, and in doing so, Chihiro’s parents fail the moral test.
As punishment, they are turned into pigs. The parallel to AI is hard to miss: eating without consent, where uncontrolled appetite destroys any awareness of the intangible dimensions of what’s being consumed, etc. etc. etc.
To save her parents, Chihiro is told to take a job in the nearby bathhouse. It’s a greedy place that runs on pleasure and appetite; ostentatious and opulent in every way. It is managed by Yubaba, a witch who binds her workers to service by stealing their names.
Chihiro is renamed Sen (translation: thousand, literally a number), and learns that if she forgets her true name completely, she will never find her way back home. The lesson is clear: in a place ruled by appetite, everything can be eaten, even who you are.A similar pattern happens with the models: once they eat your work, they take your name off of it to use your labor for their own purposes.
The movie later returns to hunger, this time in its spiritual form. Chihiro meets No Face, a quiet spirit waiting outside the bathhouse in the rain. Out of kindness, she lets him in to take shelter, not realizing what she’s invited inside.
No Face is mimicry embodied. Once inside the bathhouse, he observes and absorbs its values. He tries to find ways to repay Chihiro for letting him into the bathhouse, but she turns down his offerings.
Later that evening, a greedy frog who works at the bathhouse sneaks back to the Big Tub to see if any gold was left behind by another customer. While searching, he comes across No Face.
No Face lures him closer by materializing gold from his hands because he sees that the frog wants it. The frog is captivated, and steps forward. He’s entranced. The frog gets closer,
and No Face eats him. He can now speak in the frog’s voice, because consuming allows him to mimic and command. But No Face is still hungry.
Another employee comes along after hearing some noises. No Face uses the frog’s voice to issue commands to the next employee in line. He’s says he’s hungry. Wake everyone up. It’s time to eat. So he produces more gold, gets more food, and his appetite continues to escalate.
It’s not hard to see the parallel, is it? An insatiable force, feeding endlessly to imitate, provoking action with vast sums of money?
No Face’s hunger hooks into the bathhouse’s greed. Word spreads that a wealthy guest has arrived, and the place erupts into a frenzy to serve him.
Cooks parade steaming platters through the halls, servants sing and dance for his attention, even Yubaba eventually takes notice. Everyone wants a piece of No Face, but no amount of food or flattery can satisfy him. What No Face wants is to see Chihiro.
There’s something poignant about how he keeps trying to give her things, because that’s what he’s learned gets people’s attention.
But Chihiro keeps refusing, which both frustrates him and feeds his obsession with her.
No Face’s consumption eventually comes to a head. He eats two more employees and wrecks the joint, so Chihiro is forced to confront him.
No Face does not understand that what he’s really seeking—genuine connection—can’t be bought or consumed. He mistakes his spiritual hunger for a physical one, trying to feed an emptiness that food can’t fulfill.
When Chihiro refuses No Face’s gifts for the last time, she offers up herself to be eaten. But first, she wants No Face to eat the medicine she had been saving to restore her parents.
No Face eats it simply because he eats everything. And suddenly, the intoxication of the bathhouse begins to break.
The bathhouse workers finally see his gold as fake, No Face vomits up everyone he ate, and he quickly returns to his mute, shadow thin form.
I suppose the lesson is that you can’t get enough of what you don’t need.
On her way out of the bathhouse, Chihiro sees No Face brought back down to size. She realizes what he needs is connection, and lets him follow her as she leaves.
No Face joins Chihiro on the train to go meet Yubaba’s gentle twin Zeniba in the swamp.
Zeniba is Yubaba’s opposite. Warm, less ostentatious. At Zeniba’s cottage, something shifts. After knocking, No Face hesitates at the door, but she holds it open for him to enter. This time, he’s invited in.
There No Face receives something that finally satisfies him. He can eat normally now that he has a modest place to belong without the performance of abundance.
No Face finally finds peace by spinning thread beside Zeniba, turning his attention to small and steady work to make the materials for a collaboration.
The lesson for AI might be similar. Its danger comes because it operates inside systems with no sense of “enough.” AI needs boundaries, and so do we. The question isn’t just “what can this machine do?” but “what should it serve?” and, most importantly, “when should we stop?”
Listen, I’m not naive. I know how little room there is to move inside these systems. It’s 2025, and I’m tired. I don’t believe words like these will change much. The people who could change things aren’t listening, and the incentives are too strong to keep the machine running.
So I look for something smaller: incremental progress, unexpected benefits, minor redirections, small refusals. I am not sure how to feel about it, which is why I am trying to be articulate in my confusion. I want to carve out a small creative place for myself in everything that is happening. Is it possible?
The optimist in me remembers Chihiro, the girl who brought the devouring spirit back down to size with her refusal. The pragmatist in me remembers Miyazaki, the artist who made her, still caught in the machines he tried to resist. The realist in me knows that whatever I see in them is coming from me.
There’s a quote by the philosopher and writer Simone Weil that I keep with me. She says:
“We have to endure the discordance between imagination and fact. It is better to say, ‘I am suffering,’ than to say, ‘This landscape is ugly.’”
Simone Weil
In other words, it is better to stay with your own experience instead of projecting it outward.
Spirited Away was released in 2001. It has nothing to do with AI. If I gave this talk at a different time, No Face could have been crypto, could have been Google, could have been social media. Allegory has its limits. Simone Weil might say that what all these interpretations have in common is that I’m the one making them.
I used to like technology. The only reason it frustrates me is because I secretly believe it can satisfy me. Perhaps it once did, but the machine is not enough. Is that technology’s failure or my own growth? There are better things to suffer for.
Maybe writing this is my own version of finding some small work to do at Zeniba’s cottage. The wheel spins and I put another few lines down on the page. And I think, “the machine may know everything, but at least I know where to stop.”
It sounds like something from the Tao. Which leads me back to Rubin’s adaptation. I was unfair earlier. Not every verse he wrote is bad. If you dig, there is one hidden in the middle of the stack:
The Way of Code #53
The great way is easy,
yet people search for shortcuts.
Notice when balance is lost:
When rich speculators prosper
while farmers lose their land.
When an elite class imposes regulations
while working people have nowhere to turn.
When politicians fund fraudulent fixes
for imaginary catastrophic events.
All of this is arrogance and corruption.
And it is not in keeping with Nature’s Way.
Way over in a seldom-updated corner of the internet, you can find the Trampoline History Blog, written by a woman named Dagmar Munn, whose father George Nissen developed the modern trampoline.
George didn’t invent the first ever trampoline, but he popularized trampolines for sports and recreation from the 1950s through the 1980s, and came up with new designs and ways to use them. The stories Dagmar tells on her blog are a fascinating history of something I knew very little about. Many of the stories are told through the words of her husband Ron Munn, who worked with her father in the trampoline business.
The blog is a bit of a rabbit hole, with every click revealing something new and interesting.
Here are some things you’ll find at the Trampoline History Blog:
The Great Pyramid
One of the most incredible stories on the blog tells how George and Ron tried to get a trampoline up to the top of one of the Great Pyramids in Egypt.
The trampoline set up at the pyramids’ base in 1977
At some point in its history, the very top of the Great Pyramid of Khufu was taken off, supposedly by Napoleon if Ridley Scott is to be believed. Whatever happened, it left the top of the pyramid with a surface flat enough to set a trampoline on.
In 1977, since Ron and George were going to be in Egypt for a trampoline event anyway, they wondered if they could get one up to the top of the pyramid. They planned to climb to the top and arranged for a helicopter to lower the trampoline to them. But at the last minute, the pilot felt the weather wasn’t cooperative enough and bailed out.
So instead, they carried a mini-trampoline to the top of the pyramid, where 63 year old George Nissen did flips on the mini-trampoline.
That’s pretty amazing. But that wasn’t the dream. The dream was a full-size trampoline. So over several months they plotted how they could get a trampoline up to the top. Finally, a return trip to Egypt presented an opportunity.
Inspired by how the pyramid had been built one stone at a time, they came up with a way to bring a full size trampoline to the top of the pyramid in pieces, and assemble it there.
I’m pretty sure this isn’t allowed anymore
The plan worked.
Ron later wrote, “this attempt was certainly the first somersault ever turned on a trampoline atop the Great Pyramid of Khufu!”
You can read the whole story, written by Ron and with all its twists and turns and many great photos, in a five-part series on Dagmar’s blog.
Someone should make that story into a movie.
Trampoline Centers
When I’ve taken my kids to birthday parties at trampoline centers, I’ve actually been kind of jealous because they look so fun. But there was a short time in the 1950s where “Jump Centers” sprang up all over the country. They were so popular that even gas station owners were starting to set up trampolines at their gas stations! I guess that way your kids could get some pent up energy out while you’re filling up on a road trip? They should really bring that back.
Jump Centers were pretty photogenic, so tons of photos exist, and you can find a lot of great pictures on the blog. The craze even made the cover of LIFE magazine!
Vintage Videos
The blog has an associated YouTube channel with lots of vintage videos from trampoline’s emergence in popularity. The oldest one is from 1937 and shows high diver Larry Griswold incorporating a trampoline and a trapeze into his diving routine!
And this clip from 1949 shows a great 31-bounce routine from a trampoline competition:
Here’s a bit of context from the video’s description:
Prior to 1948, early trampoline competition rules allowed each bouncer a full 2-minutes, which could be used for skills, intermediary bounces, or even dismounting the trampoline to consult with his coach before remounting and completing the time period.
By the time the 1949 NCAA Trampoline Championships were held, the rules changes to limited competitors to a 31-bounce routine.
Edsel “Ed” Buchanan, hailing from Amarillo, Texas, was the first to connect all this skills together in what was called SWINGTIME. Representing the University of Michigan men’s gymnastics team, Ed won the NCAA title on trampoline in 1949, 1950 and 1951.
Company Newsletters
Dagmar has a second blog where she’s archived a history of the Nissen Company’s internal newsletters, the Nissen News, from 1957 to 1980. It’s a quaint bit of ephemera preserved for the world to see, and it makes Nissen look like a fun place to work.
Which reminds me: Congratulations to George in accounting’s son Danny, who won first place in the “Big Wheel” division at the Merchant’s National Bank bike races in 1977 in Cedar Rapids, Iowa.
Every mop-headed Danny had a Big Wheel
Dagmar Herself
Dagmar, who runs the blog and its spin-off blogs, was a jumper herself. She and others in her family used to perform as a trampoline act called The Nissens. Today, she writes about how she lives with ALS on the website ALS News Today.
And amazingly, she still jumps! In fact, she uses a trampoline as part of her ALS therapy. She recently posted this on Facebook:
Keep bouncing, Dagmar!
You know, one thing I don’t really talk about here often is my archive. I think of my writing as being primarily for my regular readers and secondarily for someone who discovers Ironic Sans for the first time and goes down a rabbit hole to see what else I’ve written about. So I try to write mostly evergreen content that could be enjoyed any time.
I’m thinking of that today because trampolines have made me think of this website’s “bounce rate” – the percentage of people who come to just one page and then leave. I don’t know what the bounce rate is, but it’s probably pretty high. Or low? Whichever one is worse.
Of course, if you get my newsletter in your Inbox, or follow the RSS feed, bounce rate doesn’t apply since you don’t visit the site to begin with (although you really should browse the archive if you never have). But if you’re new around here and visiting the website, maybe you want to look around. You’re sure to find something you like.
And that’s it for another newsletter! Thanks as always for reading. See you next time!
One thing that often surprises my friends and family is how tech-avoidant I am. I don't have the latest gadget, I talk about dumb TVs, and Siri isn't activated on my iPhone. The only thing left is to go to the kitchen, take a sheet of tin foil, and mold it into a hat.
To put it simply, I avoid tech when I can.
The main reason for my skepticism is that I don't like tracking technology. I can't stop it, I can't avoid it entirely, but I will try as much as I can.
Take electric cars, for example. I get excited to see new models rolling out. But over-the-air updates freak me out. Why? Because I'm not the one in control of them.
Modern cars now receive software updates wirelessly, similar to smartphones. These over-the-air updates can modify everything from infotainment systems to critical driving functions like powertrain systems, brakes, and advanced driver assistance systems. While this technology offers convenience, it also introduces security concerns, hackers could potentially gain remote access to vehicle systems. The possibility for a hostile take over went from 0 to 1.
I buy things from Amazon. It's extremely convenient. But I don't feel comfortable having a microphone constantly listening. They may say that they don't listen to conversations, but you can't respond to a command without listening. It does use some trigger words to activate, but they still occasionally accidentally activate and start recording.
Amazon acknowledges that it employs thousands of people worldwide to listen to Alexa voice recordings and transcribe them to improve the AI's capabilities. In 2023, the FTC fined Amazon $31 million for violating children's privacy laws by keeping kids' Alexa voice recordings indefinitely and undermining parents' deletion requests.
The same thing with Siri. Apple likes to brag about their privacy features, but they still paid $95 million in a Siri eavesdropping settlement.
Vizio TVs take screenshots from 11 million smart TVs and sell viewing data to third parties without users' knowledge or consent. The data is bundled with personal information including sex, age, income, marital status, household size, education level, and home value, then sold to advertisers. The FTC fined Vizio $2.2 million in 2017, but by then the damage was done.
This technology isn't limited to Vizio. Most smart TV manufacturers use similar tracking. ACR can analyze exactly what's on your screen regardless of source, meaning your TV knows when you're playing video games, watching Blu-rays, or even casting home movies from your phone.
In 2023, Tesla faced a class action lawsuit after reports revealed that employees shared private photos and videos from customer vehicle cameras between 2019 and 2022. The content included private footage from inside customers' garages. One video that circulated among employees showed a Tesla hitting a child on a bike.
Tesla's privacy notice states that "camera recordings remain anonymous and are not linked to you or your vehicle," yet employees clearly had access to identify and share specific footage.
Amazon links every Alexa interaction to your account and uses the data to profile you for targeted advertising. While Vizio was ordered to delete the data it collected, the court couldn't force third parties who purchased the data to delete it. Once your data is out there, you've lost control of it forever.
For me, a technological device that I own should belong to me, and me only. But for some reason, as soon as we add the internet to any device, it stops belonging to us.
The promise of smart technology is convenience and innovation. The reality is surveillance and monetization. Our viewing habits, conversations, and driving patterns are products being sold without our meaningful consent. I love tech, and I love solving problems. But as long as I don't have control of the devices I use, I'll remain a tech skeptic. One who works from the inside, hoping to build better solutions.
The industry needs people who question these practices, who push back against normalized surveillance, and who remember that technology should serve users, not exploit them. Until then, I'll keep my TV dumb, my Siri disabled, and be the annoying family member who won't join your facebook group.
I'm worried that they put co-pilot in Excel because Excel is the beast that drives our entire economy and do you know who has tamed that beast?
Brenda.
Who is Brenda?
She is a mid-level employee in every finance department, in every business across this stupid nation and the Excel goddess herself descended from the heavens, kissed Brenda on her forehead and the sweat from Brenda's brow is what allows us to do capitalism. [...]
She's gonna birth that formula for a financial report and then she's gonna send that financial report to a higher up and he's gonna need to make a change to the report and normally he would have sent it back to Brenda but he's like oh I have AI and AI is probably like smarter than Brenda and then the AI is gonna fuck it up real bad and he won't be able to recognize it because he doesn't understand Excel because AI hallucinates.
Worse, Brenda understands the business in ways that the executive and the AI cannot - because she sees the whole, as she's in a support role. The executive might be asking the wrong question, and the AI will congratulate them on what an excellent question it is, then give an answer which is technically correct but actually wrong. Or worse, illegal. Oh crap, we're headed towards multiple accounting scandals because of this, aren't we?
Log in
