So far, however, civilization has not been destroyed. It seems the demand for techno-apocalypses is much greater than the supply. What’s going on here? Why are so many people always convinced that technology is on the cusp of destroying civilization?

Of course, each individual prediction of doom has its own internal reasoning which should be evaluated apart from the broader trend. However the sheer number of widely expected techno-apocalypses, and the similarities in how the ideas are spread throughout society suggest a common pattern at work, separate from the question of how plausible any particular apocalypse scenario might be.

To understand this, the first thing we have to look at is when and where has this been happening. Is this a general phenomenon that humans do in all times and all places, or something that happens in all technological civilizations, or is it something specific to modern Western civilization?

It would be strange if there were popular predictions of techno-apocalypse before rapid technological change became such a visible force. This is indeed what we find. Before the modern era, the closest match is popular “millenarian” movements in Christian societies. These rhyme a little bit with modern techno-apocalypse, and the psychological effects on the believers are remarkably similar to the effects of modern techno-apocalypse beliefs, but they’re not really what we’re looking for. For one thing, millenarian beliefs are about society being radically transformed into a permanent utopia rather than being destroyed, and for another, the transforming force is Christ bringing about the Last Judgment rather than technology. We’ll have to keep looking.

Do we see techno-apocalypticism emerge as soon as rapid technological advance sets in? Actually, no! If we look around the Industrial Revolution, starting around the 1770s or so, there is nothing of the sort. Only a few cutting-edge intellectuals started to realize how important technological progress would be. Ben Franklin watched the prototype hot air balloons and immediately realized that air power would someday transform war, but rather than feeling anxious about the destruction it would cause, he hoped it might “[c]onvinc[e] sovereigns of the folly of wars.” This futurism was unusual even among intellectuals like Franklin, and none of it filtered down to popular discourse—understandably, because industrial technology was not yet transforming the regular person’s daily life.

It was not until the Second Industrial Revolution, starting roughly around the 1860s, that technologies like trains and mass production and electric lights rapidly intruded on urban people’s daily lives, and regular people began to perceive technological progress as a major force in the world. Yet, while there was popular discourse about technological progress, we still do not see techno-apocalyptic anxieties. The only case that looks a little bit similar was the “coal question,” the idea that available supplies of coal—at the time the only industrial power source in use—would eventually run out and industrial civilization would go with it. The idea was common knowledge in intellectual circles, but made approximately no impression on the public. Intellectuals occasionally brought up the “coal question” during popular debates about high coal prices, but never with the palpable anxiety of 20th century writers talking about “peak oil” and the like, and it never achieved any traction beyond the sort of philosophical futurist who in 2026 has opinions about the “simulation hypothesis” and the “Fermi paradox.”

The Apocalypse of 1914

What was the first techno-apocalypse that achieved popular traction? It first emerged right after the First World War and abruptly dominated public discourse. Nearly every thinker and pundit anywhere in Western civilization suddenly worried that war with ever more powerful weapons would destroy civilization.

It’s no mystery where this idea comes from! Industrial war, with trucks bringing men and supplies to the front en masse, and machine guns mowing them down by the hundred, caused death and destruction on a scale no one had seen in living memory. While the per capita death toll from the World War was on par with earlier great power super-wars like the Seven Years’ War or the Napoleonic Wars, the World War affected people more profoundly.

Many people at the time believed that humanity was progressing beyond such things, so they saw the World War as a demolition of their worldview rather than as the normal, if tragic, course of history. That the war appeared so senseless, even by the low standards of international war, contributed further to the emotional impact. At the time, most commentators attributed the outbreak of the war to nationalistic pride of the most narrow-minded and unstrategic sort, and today most historians offer passive-voiced stories about tangled alliances and balances of power which suggest the whole thing just kind of happened by mistake. Of the two, I think the “short-sighted pride” explanation is actually closer to the mark.

And, of course, the rapid introduction of so many new weapons made it easy for people to reason, correctly, that future wars would see the creation of more and more destructive weapons, and further to extrapolate that this would eventually destroy civilization. This discourse was absolutely everywhere, even in subjects with no immediate relation to the topic, in a pattern that will be familiar to us from, for instance, the climate apocalypse discourse of the 2010s. To give one example, from T. A. Rickard’s 1932 Man And Metals:

Civilization obviously is menaced by the misuse of the very products that were essential to its advancement. By aid of metals man made the tools by which he emerged from savagery and with which he constructed the machines that have given a bigger scope and a wider meaning to human life. From the dawn the digging of ore has played a leading part in the drama of humanity and no one has more cause than the miner to deplore the misuse of the products of his skillful toil. The sword was made before the plowshare, the spear was fashioned before the chisel. The maleficent use of metals has preceded the beneficent use of them. The perversity of mankind has turned a blessing into a curse. Shall we mend our ways or go with the Gadarene swine down the steep slope of perdition?

 

The history of mining, like all other history, thunders a warning. Those that live by the sword shall die by the sword. The Assyrian trampled upon the Egyptian, the Persian on the Assyrian, the Greek on the Persian, the Roman on the Greek. As they did in ancient times, so we, more civilized, as we deem ourselves, have done in later times. History is a philosophy that teaches by example. We have more examples than our predecessors; shall we heed them no better, more particularly the latest of them, which brought us to the very brink of perdition; or shall we too join the great discard of those that were weighed and found wanting? The finger of history, like that of Daniel before Belshazzar, bids us beware lest we too go the way of Nineveh, and our civilization, like its many proud forerunners, be destroyed by the forces it created but could not curb, by a demon it might invoke but could not exorcise.

Two things are worth noting in this passage. First, Rickard does not specify a particular weapon which will destroy civilization, only the general trend of destruction. Some men predicted that indiscriminate bombing of cities would be the weapon to destroy civilization, and in hindsight we must give them credit for foreseeing the shape of terror bombing even if they greatly overestimated its strategic effects. But this was the exception, and most discourse about modern weapons destroying civilization followed Rickard’s more general reasoning.

Second, the destruction of civilization is held out as an open question. Can civilization pull back from the brink? Can we figure out some way not to march together off the cliff? This was a live possibility. Unlike the “coal question,” these views did not stay isolated among intellectuals and futurists. This view spawned popular movements and drew statesmen to its banner. They created the League of Nations and the disarmament movement achieved the unprecedented Washington Naval Treaty, which limited the construction of warships among the major powers.

The science fiction author Olaf Stapledon grappled with the spiritual angst of the apocalypse he felt bearing down on civilization in his 1930 novel Last and First Men. In this book, the world’s great powers fight a series of wars with ever more destructive weapons. Eventually, the whole population of Europe is annihilated with poison gas. To avert an even more destructive war, the remaining nations unite in a single World State. This endures until coal supplies run out, at which point civilization disintegrates and man reverts to savage tribes. A hundred thousand years later, a new civilization emerges. 

During a rebellion, their advanced technology causes a massive worldwide explosion that turns the planet into a lifeless volcanic wasteland and kills the whole human species, except a few dozen survivors who settle in now-tropical Siberia. Over millions of years, the planet slowly becomes inhabitable once again, and their descendants evolve into the posthuman “Second Men.” New species rise and fall and repeat the mistakes of the past. Finally, the Eighteenth Men—the Last Men—achieve an enlightened utopian society. This, too, begins to decay after a stellar disaster occurs and the Last Men realize that their civilization will eventually be destroyed by the Sun. In the introduction to the novel, Stapledon writes:

We all desire the future to turn out more happily than I have figured it. In particular we desire our present civilization to advance steadily toward some kind of Utopia. The thought that it may decay and collapse, and that all its spiritual treasure may be lost irrevocably, is repugnant to us. Yet this must be faced as at least a possibility. […] May this not happen! May the League of Nations, or some more strictly cosmopolitan authority, win through before it is too late! Yet let us find room in our minds and in our hearts for the thought that the whole enterprise of our race may be after all but a minor and unsuccessful episode in a vaster drama, which also perhaps may be tragic.

We can see statesmen grappling with the practical side of this problem in a 1932 disarmament debate in the British parliament. Frederick Seymour Cocks, like Rickard, fears the rising power of new weapons in aggregate: 

If it fails now, the only alternative will be to arm. The nations will arm, and will embark upon another armed race, by land, by sea, and by air. Following upon that, as closely as a man is pursued by his own shadow, will come war, under the waters, in the air and on the land, until civilisation cracks beneath the strain and the Vesuvius of revolution opens out beneath our feet. 

Clement Atlee, who would become Prime Minister in 1945, fears the bombing of cities in particular:

I believe that what we have to do is to try to build up a constructive internationalism, and I believe that the most fruitful suggestion which has been made in this regard is that relating to the internationalising of civil aviation … I do not believe that you have a defence against air warfare at the present time. I do not believe that you can restrain air forces as long as you have nationalised civil aviation, and I believe that, unless air warfare is restrained, civilisation will be wiped out. 

Similar calls for world government over the supposedly apocalyptic technologies have also been a common feature in every subsequent techno-apocalypse prediction, be they caused by nuclear weapons or artificial intelligence. This debate is most famous for a speech by former Prime Minister of the United Kingdom Stanley Baldwin:

[T]here is, as has been most truly said, no way of complete disarmament except the abolition of flying. Now that, again, is impossible. We have never known mankind go back on a new invention. It might be a good thing for this world, as I have heard some of the most distinguished men in the Air Service say, if man had never learned to fly. But he has learned to fly, and there is no more important question, not only before this House, but before every man, woman and child in Europe, than: “What are we going to do with this power now we have got it?” … This is a question for the younger men far more than it is for us. They are the men who fly in the air. Future generations will fly in the air more and more. Few of my colleagues around me here, probably, will see another great war.… If the conscience of the young men should ever come to feel with regard to this one instrument that it is evil and should go, the thing will be done; but if they do not feel like that—well, as I say, the future is in their hands. But when the next war comes, and European civilisation is wiped out, as it will be and by no force more than by that force, then do not let them lay the blame upon the old men. Let them remember that they, they principally or they alone, are responsible for the terrors that have fallen upon the earth.

And then the next war came. The two decades of desperate searching for a way out ended in fire and rubble. London was bombed, and Dresden, and Tokyo. The logic of industrial production was turned to the mass extermination of civilians. The promised new weapons arrived, most notably the nuclear bomb.

It was not as bad as Baldwin and his colleagues expected. Civilization was not actually wiped out. It took about a decade for Tokyo, Hiroshima, and Nagasaki to recover to population levels higher than their prewar starting point. But the war was destructive enough that the doomsayers felt vindicated anyway. For most thinkers, it was no longer a question of whether man will destroy himself, but when.

This felt vindication is not well-grounded historically. World War II killed more people than any conflict in human history because industrial technology can support an enormous population. Apart from modern agricultural technology making everything happen at a larger scale, the destruction of World War II was well within historical precedent. As far as the fraction of the European population it killed, World War II was about as deadly as the Thirty Years’ War, fought with pikes and muskets in the 17th century. Both killed somewhere around seven percent of Europe, with deaths closer to forty percent in the hardest-hit regions. Both left the worst-off areas “post-apocalyptic” in the colloquial sense, but came nowhere close to destroying civilization in the way Rickard, Stapledon, and Baldwin feared. 

In World War II, the advanced superweapons were strategically critical, the gas chambers and trainyards were morally horrifying, but almost all of the actual killing was done with bullets and artillery shells in ways that Napoleon would have found perfectly familiar. Probably less than ten percent of the war’s deaths came from the vaunted new weapons like tanks and aircraft. The reality is that industrial armies, with guns and bombs and concentration camps, have been less thorough than Mongols with horses and bows, who killed more people in Baghdad with their own hands than the Americans managed to kill by dropping atomic bombs from thirty thousand feet.

But for people who had imagined their technological superiority also made them morally superior to past civilizations, and who had long forgotten the last time their great powers fought a total war targeting civilians en masse, World War II felt like a near-apocalypse which came within a hair’s breadth of destroying civilization entirely. The illusion of rational progress “advanc[ing] steadily toward some kind of Utopia” was wounded by the First World War, and killed by the Second.

After the First World War, more people expected that some unknown future technology might destroy civilization at some unknowable future date. After the Second World War, more people expected that a specific identified technology would destroy civilization within their own lifetime, although which particular technology is supposedly on the cusp of destroying us has changed over time.

The Parade of Prophecies

The first technology which the public believed was about to destroy civilization was, of course, the nuclear bomb. In popular imagination, it was the sheer explosive force of destroying cities that would end civilization. Cooler heads realized blowing up major cities would merely kill an enormous number of people, but was not actually enough to end the world. Then, in 1950, the great nuclear physicist Leo Szilard popularized the idea of nuclear “cobalt bombs” deliberately designed to spread radioactive fallout which remains lethal for years rather than the usual days. This led to widespread fears that a nuclear war would permanently irradiate the Earth and destroy all life. Such a scenario was not well-grounded; weather spreads nuclear fallout throughout the atmosphere very unevenly, such that covering the entire globe is effectively impossible. No cobalt bombs were ever built, and popular fear of global irradiation persisted for a decade or two. Eventually it faded from the world’s attention, although the idea of a permanently radioactive nuclear wasteland remains vaguely in the cultural imagination, mostly because of depictions in fiction.

In the 1980s, the radiation poisoning apocalypse was succeeded by popular fear of “nuclear winter,” the idea that large-scale nuclear war would loft soot into the stratosphere, higher than the rain which could bring it back to Earth, and so persist for years while blocking sunlight, cooling the Earth, disrupting agriculture, and causing massive famines that kill billions, thereby ending civilization or even causing human extinction. This was based on speculative computer models of cities burning in enormous firestorms and further speculative models of how soot behaves and persists in the stratosphere. These models were invented largely by activist-scientists whose publicly declared aim was to terrify world leaders into believing nuclear weapons were a weapon too terrible to use and, unsurprisingly, their models have not held up well to empirical verification. 

In the 1991 Gulf War, massive oil well fires did not produce the predicted lofting effect; and in 2017, when smoke from Canadian wildfires was lofted to the stratosphere, it dissipated much more quickly than predicted by the models used to forecast nuclear winter. By now, the public anxiety about nuclear winter has mostly faded away, partly because of the problems with the climate models, but mostly because the end of the Cold War and the activist campaigns for nuclear disarmament meant that there was no longer much media raising the issue in people’s minds.

While popular fears persist, there is currently no widely-accepted case that nuclear bombs will destroy civilization among experts. Perhaps a new model will arise to fill this niche in the next decade or two. The idea that nuclear bombs could kill a hundred million people, and it would be just another devastating war that gets recorded in the history books while people rebuild, seems unintuitive and even morally offensive to us. Something like that deserves the narrative weight of an apocalypse.

Of course, nuclear weapons are not the only technology that was supposed to have destroyed the world by now. The normal operation of industry at scale was going to cause the apocalypse as well. Like with nuclear weapons, the specifics have changed as individual mechanisms were disproved or simply fell out of fashion, because many people have a deeper spiritual conviction that large-scale industry ought to end the world. The 1968 publication of The Population Bomb by zoologist Paul Ehrlich popularized the theory that overpopulation, enabled by industrial-era advances in agriculture, sanitation, and medicine, would increase the population above the carrying capacity of the Earth and lead to mass starvation and collapse within the 1970s. Ehrlich would later be one of the chief organizers of the campaign to develop and spread the nuclear winter theory as well.

In 1972, The Limits To Growth book made the case for two additional mechanisms of industrial doom: resource depletion, as all possible sources of industrial inputs like aluminum and chromium are exhausted; and pollution, as exponentially increasing emissions of industrial waste poison the world and make it uninhabitable. The latter of these was especially influential, as fictional depictions of barren landscapes blighted by toxic sludge made the idea popular among the general public. In fact, the normal operation of industry was perfectly able to solve these problems. Food, metals, minerals, and fossil fuels are all far more abundant today than they were when these dire predictions were made, and this trend shows no sign of stopping. 

The actual limits of matter and energy available to human civilizations are incomparably vaster than the doomsayers imagined, as far beyond modern civilization as modern civilization is beyond a primitive tribe worrying about using up all the flint for their arrowheads. Cleaning and preventing pollution proved even more tractable. Filtering and treating industrial waste required research and development to create air scrubbers, waste-to-energy incinerators, clay-lined landfills, and other such technologies. Deploying these across the industrial stack was reasonably expensive, but well within the means of an industrial society, and most wealthy countries mandated these improvements throughout the 1970s. The problems are now mostly solved. The cities are no longer choked by smog, the rivers are clean again, chlorofluorocarbons have been replaced with non-ozone-depleting substitutes, and acid rain is a thing of the past.

Much as the fears of nuclear apocalypse switched from radiation to nuclear winter when a new justification was needed, the ecological fears switched from pollution to global warming. In the late 1980s, ecological projections that carbon emissions would raise global temperatures by several degrees celsius and cause substantial disruptions to local climates and human life broke into mainstream awareness, where they were soon spun by activists, reporters, and fiction authors into widely-believed but scientifically baseless fears that warming would make the world literally uninhabitable. A series of dire predictions about coastal cities being submerged by melting polar ice have failed to materialize on schedule. 

As atmospheric carbon capture technology is developed and rolled out in coming decades, it seems plausible that industrial technology may solve global warming and prevent even the moderate problems which would come from warming of several degrees celsius, but this remains somewhat speculative. If not, the climate disruptions will fall far short of the “world literally on fire” messaging popular in the 2010s. In the last couple of years, it seems as if the wind has gone out of the sails of the global warming apocalypse narrative, and few people remain emotionally invested the way they were even in 2021. Like the other fears of technological doom, it is not that the core arguments were explicitly refuted in the public mind, but rather that people eventually got tired of waiting for an apocalypse that never arrived.

The latest popular doomsday scenario is artificial intelligence, which exploded into public consciousness after the creation of compelling generative artificial intelligence chatbots, especially ChatGPT in 2022. In addition to the inchoate fears of economic obsolescence and the rise of new magnates disrupting the political balance of power that accompany every major new technology, there is also fear that the AI itself will soon exterminate humanity. The most influential and organized activists advancing these views trace to the work of Eliezer Yudkowsky, where he posits a sufficiently intelligent artificial mind would, by default, exterminate humanity as it pursues its own inhuman goals, unless the extremely difficult research problem of aligning its goals exactly to human goals succeeds completely on the first try, in which case the AI would instead create a permanent utopia.

What these “AI Doomers” provide for this narrative is not so much an argument that superhuman AI is dangerous. Futurists and philosophers have speculated about artificial minds supplanting humanity since before the electronic computer was invented. The idea that intelligent nonhuman minds are in natural competition with humanity is obvious and very common, and has motivated stories of robot uprisings since R.U.R. (Rossum’s Universal Robots) in 1920. 

Rather, the modern “AI Doomers” provide arguments and intellectual authority for the claim that artificial general intelligence is happening soon, that computer minds pursuing their own interests without the need for human direction are a scientifically respectable possibility, rather than a fiction trope like time travel or alien invaders, that GPT and its derivatives are one of the last steps on the path towards building Man’s successor. There have been past waves of enthusiasm when specialists in artificial intelligence believed they were close to building general intelligence only to be disappointed as the field plunged into “AI winter,” but these did not achieve adoption beyond technical specialists and inveterate futurists. Yudkowsky’s predictions of superintelligence’s imminent creation were the first to achieve wider reach, which ironically inspired Sam Altman, Elon Musk, Dario Amodei, and other businessmen, researchers, and investors to create the AI labs that are now advancing the state of the art.

There have also been plenty of techno-apocalyptic scenarios which remained relatively niche and did not spark visceral public anxiety—bioengineered pandemics, electromagnetic pulse weapons, nanotechnological self-replicating “gray goo,” exotic physics disasters from experimental particle accelerators, the “clathrate gun hypothesis” that moderate global warming would set off a runaway feedback loop and overcook the planet, and more. Partly because these do not fit as neatly into the ideology of industrial civilization being destroyed by its own hubris, and partly because of which causes the most competent activists have chosen to push, these have not achieved the scale or the emotional resonance of the most successful apocalypse narratives.

The Hypothetical AI Apocalypse

In each case that achieves popularity, the pattern is mostly the same. First, intellectuals describe a phenomenon that, if taken to its maximum conceivable extreme, would destroy civilization or the entire human species. Usually this speculation is logically sound, within the limits of its clearly-defined assumptions. There exists some threshold of global warming that would cause human extinction, even if 5 °C would not suffice; dusting the entire Earth with radioactive cobalt-60 would kill all humans outside of shelters if it could somehow be arranged; running out of coal in the 1860s would have shut down industrial society before it learned how to make solar panels or fission reactors. This work attracts the attention of intellectuals and futuristic types, but not the general public. 

Second, some of the intellectuals make a public case painting a vivid picture of how the apocalypse could happen within a decade or two in order to engage the popular imagination. This uses looser arguments with poorly justified or unfounded assumptions, although the intellectually rigorous people who first develop the looser arguments—such as in The Limits To Growth—specify that it’s just a model, or an illustration, or one way that things could plausibly go, and “of that day and hour no one knows.” They are careful to make their assumptions explicit and highlight which steps of the argument they place less weight on. Some of the intellectuals come to viscerally believe that the apocalypse will be soon, for reasons which are only partially based on their explicit arguments, and partly due to subterranean matters of ideology and psychological pressures. Even so, the most careful and scrupulous are aware of what their arguments do not prove, and avoid saying that their worst fears will happen, speaking only of possible scenarios and models worth considering—which incidentally means they are never publicly proven wrong when the fears do not materialize.

For now, such qualifications are the norm in predictions of AI doom from the movement’s intellectual core. For example, in 2020, when Ajeya Cotra published a lengthy analysis of when “transformative AI” (TAI) is supposed to arrive based on claimed analogies to biological intelligence, the report assumed AI development would follow what it admitted was a “relatively unrealistic path to TAI because it is simpler to analyze.” Or in 2025, when the authors of the influential popularizer “AI 2027” received criticism from people who believed the title meant they were predicting the AI apocalypse would come in 2027, they were quick to place an addendum at the top: 

to prevent misunderstandings: we don’t know exactly when AGI will be built. 2027 was our modal (most likely) year at the time of publication, our medians were somewhat longer. Specifically, our medians ranged from 2028 to 2032. When AI 2027 was first published we explained this in Footnote 1 as above, but to make our views more clear we have added a clarification to the foreword text.

Yudkowsky himself has gone even further and denounced the entire practice of making up “AI timelines” for when the superhuman AI will arrive and presumably exterminate humanity, and he makes a point of not giving specific dates.

In every case, from interwar fears of runaway militarism to today’s fears of superintelligent AI takeover, the shuffle works by making an unjustified leap from the claim that the apocalypse is possible in principle to the claim that it’s happening soon. Practical difficulties are skated over or assumed away. Imagined scenarios and wild assumptions are fed into great quantitative models to produce very serious reports. The principle of “garbage in, garbage out” means that these have no value whatsoever for predicting the future. Strip away the leaps of logic and the illustrative fictions and the load-bearing assumptions made to “simplify analysis” and the calculations of irrelevant numbers, and the core remains that a coordinated cadre of well-studied experts just kind of feel a vibe that it’s happening soon. 

But the trappings of mathematized analysis can be very persuasive anyway. If you grab a pile of numbers and put them in a graph, then most readers will be dazzled by the appearance of mathematical rigor—that’s a lot of calculations!—and only a minority will ask, “Hold on, where exactly did this graph come from? What’s the argument that these numbers have any causal relationship at all with predicting when the world is gonna end?” For example, prominent AI Doomers like Katja Grace put a great deal of effort into collating software researchers’ predictions about when superhuman AI will arrive. But of course there is no reason to believe that such a big pile of predictions can tell us anything at all about the reality of such an event. There has been a great deal of research into when aggregating forecasts is a useful exercise and when it is not, and as Karger, Atanasov, and Tetlock argue, predictions of events like the arrival of superintelligent AI are not informative. 

Such esoteric points may be important for the most engaged and sophisticated intellectuals, but they have little impact on how the public receives an apocalypse narrative. The fact that intellectually respectable technical experts lend their names to the cause is important, while the details of their opinions are lost in a long game of telephone. Activists and journalists and fiction authors spread the scenario, altering and simplifying it as they do. The intellectuals’ caveats are left by the wayside, sometimes because the popularizers only understand the vivid story and do not understand an argument structure filled with careful conditionals, and often because they want to tell a more sensational story, as in environmentalist disaster media like Waterworld (1995) or Wall-E (2008). Many people believe this story. They might experience fear and anxiety, or donate money to professional activists, or forgo having children, or stop saving for retirement, or respond in any number of ways. Popular fear of the AI apocalypse is only recently mainstream. It remains to be seen how widely it will spread.

People give an apocalypse about twenty or thirty years to arrive. After that, most lose interest, leaving behind only a small residue of ideological holdouts, and a wake of science fiction settings that feel increasingly fantastical rather than urgent. There is no big finale to the debate, there is no explicit acknowledgement that the apocalypse is cancelled. People simply wander away from the subject. This seems to be less because the core arguments get refuted with reason and evidence, and more because people slowly lose faith after enough of the activists’ predictions about worldwide famines in the 1970s, or cities flooded by 2015, do not come to pass.

In the realm of pure logic, the failure of the activists’ vivid pictures may have little bearing on the abstract case for doom advanced by the more rigorous thinkers, but few people make the distinction—with some justice, as the highbrow intellectuals are generally happy to lend their intellectual authority to the popularizations. Perhaps the most notorious example is the Bulletin of the Atomic Scientists, founded by Albert Einstein and Manhattan Project scientists, which created the Doomsday Clock—initially set at seven minutes to “midnight” in 1947—to publicize the imminent nuclear apocalypse by laundering the scientists’ subjective feeling of doom into an annual media spectacle. The Doomsday Clock first cited the risk of AI destroying humanity in 2024.

It’s possible that industrial technology will destroy civilization at some point. But the technology of 2026 does not seem especially close. However, the basic observation formulated after World War I—that technology’s ability to destroy increases over time—remains sound. Surely we are closer today than in 1918. This observation does not tell us whether we should fear a techno-apocalypse in five years, or fifty years, or five hundred years. It does suggest it’s worth paying attention to individual apocalypse predictions and evaluating them on their technical merits.

But as we consider them, we must remember that there is great popular and political demand for techno-apocalypse stories. Since World War I and especially after World War II, the mainstream view in our civilization is that Man’s nature is to destroy himself with technology. There have been many attempts to fill in the details with specific apocalypse scenarios. Many tales of extinction and collapse become accepted among the vanguard intellectuals and the public even though the argument for doom has gaping holes. There is a huge audience for playing up reasonably large problems as though they are the literal end of the world. There is a huge audience for talking about speculative future disasters as though the default course is for them to happen very soon. Before we believe the latest prophecy of doom, we must remember to hold out for a full argument which covers all of these steps. We cannot settle for most of an argument with critical steps handwaved away, no matter how many fancy graphs or prestigious names are attached.

The most reputable displays of leisure were aristocratic in Veblen’s time. Veblen uses examples like hunting for sport, speaking Latin and Greek, and learning refined manners to demonstrate “good breeding.” By now, all of this is hopelessly old-fashioned. But the leisure class is far older than these aristocratic values and aesthetics, and did not cease to exist just because that ideology collapsed. Today, the leisure class has adopted the new ideology, which we can roughly call “social activism,” and performs its conspicuous leisure in accordance with these newer values and aesthetics.

The Leisure Class of Today

When we talk about “the leisure class” today, we do not mean people who spend all day watching TikTok or playing video games or listening to true crime podcasts. We are talking about people who engage in conspicuous leisure. By conspicuous, we mean that they show off their exemption from unworthy labor through accomplishments which those without leisure cannot match, for want of time or money or energy. They spend their time and effort in “honorific” pursuits which place them above the base necessity of directly producing wealth.

A meatpacker illegally working twelve-hour shifts can watch Breaking Bad when he comes home, so watching Breaking Bad is just ordinary leisure, and having opinions about Breaking Bad does not demonstrate conspicuous leisure. But only a man of means and distinction can take three-week vacations to go scuba diving in exotic locations—and upload the selfies to social media—so this becomes a mark of honor. In the language of today’s economists, what Veblen calls “conspicuous” might be phrased as “suitable for costly signaling.” Conspicuous leisure often includes mastery of subtle and exacting speech codes, and adherence to precise forms of manners, carriage, and behavior, all of which requires careful study and training within the social milieu of the reputable elites.

This is why class expression is not the same thing as wealth. Many anthropologists of the modern United States, including Paul Fussell in his masterful Class: A Guide Through the American Status System, have observed that there is a large class of wealthy businessmen who own and operate valuable companies, yet do not code as part of the upper class. This is because they make their money through base production. They own and operate farms, or car dealerships, or construction businesses, or the like. They spend their energy in the work of creating wealth—not “creating wealth” in the sense of amassing dollars and stock options and intangible claims on other people’s labor, but “creating wealth” in the sense of manipulating physical objects, the food and cars and houses that people want to acquire with their dollars at the end of all the negotiating and fundraising and politicking about how the dollars will be distributed. In popular perception, perhaps even instinctively, this work of creating tangible wealth is considered inherently base. The entire point of conspicuous leisure is to prove that one is above such concerns, so no matter how much money a physical business operator may make this way, he cannot fully become one of the rarefied gentlemen.

Of course, a gentleman of means has far more wealth than he can spend on his own leisure, so he also needs other ways to make his station visible. One of these is conspicuous consumption, the most famous concept from Veblen’s book. This is when people buy expensive objects, not mainly because they think the physical Louis Vuitton handbag is so much better than another handbag or because the Lamborghini is so much better than another car, but to display their ownership of the object to others.

Conspicuous consumption has become less important in recent generations, although it is far from dead. As industrial mass production has made physical objects cheaper and widely available, they have become a poor way to distinguish the gentleman from the throng. When fine clothes were out of reach for most people, they were an extremely important way for the wealthy to distinguish themselves, and you could tell rich from poor at a glance. In 2025, it’s possible to buy a decent ballgown on a minimum wage salary, so there is little point in wearing ballgowns. Multimillionaires might as well wear a t-shirt and sneakers. We have seen less dramatic versions of the same trend with objects like fine tableware, furniture, televisions, and even diamonds. 

Where conspicuous consumption of manufactured goods remains an effective social tactic, now it is often a matter of purchasing expensive brand names at deliberately inflated prices, rather than purchasing objects that necessarily require a great deal of labor to make, like a 19th-century ballgown. A billionaire can buy a brand-name handbag, and a grocery store clerk can buy a “counterfeit” handbag, which most people find indistinguishable.

The sharpest point in the decline of conspicuous consumption was the rise of the “counterculture” in the 1960s and especially the 1970s. This ideology eschewed material status symbols, often using Veblen’s words to denounce them as crass and spiritually polluted. Instead, they turned back to conspicuous leisure—following rock bands on tour across the United States, backpacking the “hippie trail” across southwest Asia, cultivating mystic awareness, devoting themselves to radical activism, and a dozen other means of demonstrating their remove from labor and physical production. By the 21st century, when the counterculture had been fully recuperated into the mainstream culture, marketers spoke of the value of “experiences” over “things” in order to sell conspicuous leisure to the middle class.

There is another way for a gentleman to display his wealth beyond what he can expend on his own leisure, even more important than conspicuous consumption. This is vicarious leisure, that is, maintaining others to engage in nonproductive activities which redound to the honor of the master. Historically, the most basic form of vicarious leisure was maintaining a wife to engage in reputable leisure rather than household production, but it could reach much greater scale than that. In the Dark Ages, men like Hrothgar would throw massive feasts for his followers every night in his mead-hall and give gold bracelets to his favorites. In the Middle Ages, a lord would maintain a court full of retainers and knights and astrologers and jesters. As Adam Smith tells the story in The Wealth of Nations, it was only with the late medieval rise of craft production and long-distance trade that conspicuous consumption of luxury goods could absorb the money that had once gone to the vicarious leisure of maintaining a court.

By the time Veblen published in 1899, the middle class had come to the fore, so the leisure class had grown to include many people of moderate wealth. At that point, he claims the main form of conspicuous leisure was employing household servants to maintain standards of exacting cleanliness far beyond the point required by hygiene and good health, “not so much for the individual behoof of the head of the household as for the reputability of the household taken as a corporate unit.”

We see a glimpse of this culture in the novel Peter Pan, written around the same time as The Theory of the Leisure Class. In the magical world of the story, the Darling family’s nanny is a dog, because they cannot afford to pay a human servant, to the mortification of the patriarch. “No nursery could possibly have been conducted more correctly, and Mr. Darling knew it, yet he sometimes wondered uneasily whether the neighbours talked. He had his position in the city to consider.” In his eyes, a nanny who merely takes excellent care of the children is inadequate if she does not also conspicuously demonstrate his household’s reputability—which is to say its wealth.

Household servants have since fallen out of favor, partly because the idea of the household as a “corporate unit” does not fit with today’s ideology, but largely because industrial mass production of household appliances—especially the washing machine—has made it so anyone can keep a household to high standards, not just the wealthy. As household servants hauling wood have been replaced by gas ovens and electric heaters activated at the press of a button, the leisure class redirected this money.

The most obvious of the new forms of conspicuous leisure is the college degree, which is long since mandatory as a mark of reputability. Even if the actual coursework has become a sick joke, and perfect grades no longer indicate any special skill or intelligence, a degree still represents four years of foregone earnings in an expensive environment. In fact, as college courses have become ever more useless, this arguably makes college a better mode of conspicuous leisure, for much the same reason that dog breeders now focus on breeding sickly show dogs rather than breeding productive working dogs as they once did.

In addition, today’s Mr. Darlings frequently use international travel as a mode of conspicuous leisure to distinguish themselves from the lower class. A trip to Berlin, for an American, or a trip to New York, for a German, serves as a mark of reputability and sophistication. Those who have never left their home country often feel that this makes them unworthy in comparison. The middle class goes to great pains to display their travel on social media or find excuses to mention it in conversation. More than once, on a first date, I have been explicitly asked, “Do you travel?” in the tones that a Jane Austen character might use to inquire about a suitor’s income.

Activism Cements the Aristocracy’s Position

According to Veblen, the inherently honorific vocations which the leisure class naturally pursues are “government, war, sports, and devout observances.” Today’s elite ideology does not attach much honor to war or sports, unlike the aristocratic ideology of Veblen’s day. So how does today’s leisure class use its wealth to pursue government and devout observances?

Today, the ideology that awards honor to leisure class elites is social activism. The leisure class can use this ideology to justify their position in society and to legitimize their pursuit of naturally honorific vocations. As a result, the social activist ideology has dominated moral discourse, intellectual institutions, and mass media for generations.

A good description of the psychology at play was written by one of its major practitioners. In 1997, the business magnate Michael Bloomberg coauthored an autobiography containing an unusually frank discussion of the motivations for his philanthropy. Bloomberg later entered politics and, in 2019, published a second edition of his autobiography with much of this material removed. Bloomberg was a lifelong supporter of the Democratic Party at the time he first published the book. Soon after, he switched to the Republican Party as part of a successful campaign to become mayor of New York City, then switched to Independent for his third term as mayor, and later returned to the Democrats in an unsuccessful campaign to become President of the United States. According to Bloomberg, emphasis added:

 “The reality of great wealth is that you can’t spend it and you can’t take it with you. All you can do is give it to other individuals (with large gift or inheritance taxes to pay), or give it to philanthropic organizations (usually with large income tax credits to receive). The issues left to your discretion are only to whom, how much, and when to give. … The real financial legacy I’m leaving my kids is much more powerful. They will be the key trustees of our family’s foundation and, as such, will possess great influence. For the rest of their lives, along with their mother and a handful of my closest friends, they’ll distribute large grants to worthy institutions and creative individuals needing support. In their hands will be the ability to channel cultural development, further scientific and medical research, shape the political process, mold our youth, and support their religious organizations. … What greater satisfaction could we possibly get than watching ourselves do great things for humanity? Not only great things, but things we, not someone else, think should be done.”

Philanthropically-funded activist groups—Bloomberg alone has given $21 billion to various nonprofits—have been behind most of the major changes in governance in living memory. Megadonors like Bloomberg, Bill Gates, Laurene Powell Jobs, George Soros, Pierre Omidyar, or MacKenzie Scott Bezos select which causes to push and which to pass over, and society is reshaped towards the ideologies and visions that get funded. The activist’s most important role is in laying the moral and intellectual foundations for transforming society, but they also groom leadership cadres to take political office; prosecute lawsuits to alter and enforce the practices of governing bureaucracies; and directly write laws which are passed verbatim by allied legislators. Because activists govern, in the most literal sense, activism is inherently honorable.

And while the comparison between activist ideology and religion is sometimes taken too far, it is inarguable that much of the psychological need for devout observances is now channeled through activist rites like parades, public murals and monuments, and the canonization of saints, while endowments to academia serve the same psychological purpose as the endowments of monasteries seen in medieval Christianity, Buddhism, Hinduism, and Taoism. This is another source of honor. As a result, activism is an ideal method for people like Bloomberg to display their wealth through vicarious leisure, maintaining other people to engage in leisure, which makes the maintainer more honorable. He writes:

“Every year, my father received a publication listing contributors to his favorite charity. During dinner, he would look down each page of the book for familiar names and remark on the size gift made by people he knew, or the complete absence of other names from the list. … Philanthropy dominates the social lives of the wealthy in big U.S. cities. … The style section of our city newspapers chronicles which celebrities attend which philanthropic dinners each night; the most celebrated are honorees there, partially for their past achievements, but also for their current fund-raising abilities. Executives and socialites solicit each other for their favorite organizations. They attend events where they bestow small tokens of appreciation on one another after suitably flattering speeches. Fun evenings for fine causes.”

The most honor attaches to the megadonors who maintain the activists in their tens of thousands and can direct them to new causes as they will. But just as honor once attached to an individual knight in the household of a great lord, so honor now attaches to an individual activist at a Bloomberg-funded NGO. Some of the leisure class are very well paid for their activism, but many more happily take low salaries and maintain their station with an inheritance or through a high-income spouse. Others make do without. Veblen writes:

“Wherever the canon of conspicuous leisure has a chance undisturbed to work out its tendency, there will therefore emerge a secondary, and in a sense spurious, leisure class—abjectly poor and living in a precarious life of want and discomfort, but morally unable to stoop to gainful pursuits. The decayed gentleman and the lady who has seen better days are by no means unfamiliar phenomena even now. This pervading sense of the indignity of the slightest manual labour is familiar to all civilized peoples, as well as to peoples of a less advanced pecuniary culture. In persons of a delicate sensibility who have long been habituated to gentle manners, the sense of the shamefulness of manual labour may become so strong that, at a critical juncture, it will even set aside the instinct of self-preservation.”

Of course, this is a familiar character to us today. As college degrees are printed by the truckload, Veblen’s “spurious leisure class” has become the subject of a great deal of commentary, now called the “downwardly mobile” or “overproduced elites.” Reams have been written about how their economic position draws them to radical activism, especially the populist brand of radical activism championed by organizations like the Democratic Socialists of America. Most of these people—admittedly not all—would be able to make more money if they learned to code or went to trade school or some such career, but they prefer to be paid in honor rather than money. While some critics try to paint this preference as “economically irrational,” there is little basis for this critique. De gustibus non est disputandum.

None of this means that the new leisure class is insincere or cynical. The leisure class of 1899 mostly believed in its aristocratic ideology. The leisure class today are generally earnest believers in the activist ideology they preach too. Whatever ideology the leisure class someday finds useful to adopt after the activist ideology eventually fades away, our descendants will believe it with their whole hearts.

The old aristocratic ideology, long fraying, ultimately collapsed in World War I as its proponents finally lost faith in their vision of the world. It took decades for the activist ideology to cohere into the form we recognize today and become the most reputable vehicle for vicarious leisure. Megadonors like Vanderbilt, Rockefeller, and Ford had pioneered these new methods of vicarious leisure. One major reason the activist ideology was able to fill the void is that its proponents did the best job of building out a full playbook of how to engage in honorific leisure, from the billionaire philanthropist honored by the vicarious leisure of the nonprofits they maintain and fêted at galas and in newspapers, to the conspicuous leisure of the heiress whose modest trust fund enables her to take a low-paying job for a cause she believes in, all the way to the spurious leisure class living in hip flophouses while they scrabble for the cause. 

Future ideologies will no doubt provide scalable playbooks for these roles as well. It remains to be seen whether the current activist ideology will succumb to its current troubles and dissolve as the last of the Baby Boomers propping it up pass away, or whether their heirs will adapt it into a new form functional enough to endure for generations more, but no ideology lasts forever. Just as the seed that eventually grew into the activist ideology’s funding networks was formed by megadonors like Vanderbilt and Carnegie in the last decades of the aristocratic ideology’s heyday, alternative funding networks formed in the coming decade may also prove to be world-historically significant.

Archaeology as we know it—a tradition of knowledge among a dedicated body of scholars in dialogue with each other and more or less synced on goals and methods—dates from roughly the 1700s. When Western archaeologists first scoured the world to survey ancient sites, very often they found the locals had been vaguely aware of the ancient ruins nearby, but never bothered to investigate more deeply than boys wandering through for a curious afternoon. Since then, archaeology has gone global. Western culture and methods have been widely imitated by other civilizations, often with funding and training from the West, and archaeology has been part of that package.

Aside from modern Western and Western-derived archaeologists, I know of only two other cases where there was a serious tradition of knowledge studying ancient sites or artifacts to try to understand the past. Both of these were aimed at relatively narrow reconstructions of religious practices, rather than trying to understand everything about past societies like our own archaeologists. I’m sure there were also scattered individuals who stared at an old Roman road or an overgrown Khmer temple and tried to imagine how the people had lived, but that’s very different from a successful tradition of knowledge.

Rites of the Ancients

Before the West, Mesopotamian civilization came closest to archaeology as we understand it under the Neo-Babylonians in the 6th century B.C. They excavated ancient Mesopotamian temples which had been ruined for over a thousand years and studied them in detail, trying to infer their floor plans and furnishings so they could reconstruct them exactly as they had been originally. Kings like Nebuchadnezzar and Nabonidus left inscriptions boasting that their temples were exact restorations of the ancient works, often on the same sites, and incorporating bricks and idols from the ancient temple. This almost certainly required methods which today’s archaeologists would recognize, such as identifying likely-looking mounds as buried megastructures, mobilizing laborers to excavate them, preserving ancient artifacts as they were dug up, and trying to piece the remains into a picture of what the ancients had been trying to do.

Famously, this project also produced the earliest known museum so far, which is best described in the archaeologist Sir Leonard Woolley’s book Excavations at Ur. On the compound containing the newly reconstructed Ziggurat of Ur, the king’s daughter Ennigaldi presided as high priestess. In one room near her temple’s school, Woolley unearthed a collection of several unrelated artifacts which were already ancient in Ennigaldi’s time, as well as a “museum label” which announced its text was copied “from bricks found in the ruins of Ur, which while searching for the ground-plan [of the temple] the Governor of Ur found, and I saw and wrote out for the marvel of beholders”. The bricks in royal constructions were stamped with the king’s name and other metadata, which has been a great help to both Neo-Babylonian and Western excavators. I’m sure the old kings would be pleased that their names are still preserved. The room also contained a few ancient artifacts unrelated to construction, like the head of a mace, so there was at least some attention to preserving antiques apart from just restoring the temples. We have so little data that we can’t say confidently what else they did or didn’t study, although it seems to me like the temple restoration project was probably the driving force, with a little bit of antiquarianism done opportunistically along the way as old sites were excavated.

This proto-archaeological tradition lasted for several generations, but seems to have ended during Ennigaldi’s lifetime. The Book of Daniel recounts the story of Ennigaldi’s brother, the crown prince Belshazzar, who received the prophecy of the “writing on the wall” foretelling the doom of the Neo-Babylonian Empire. Belshazzar and his father Nabonidus were conquered by the Achaemenid Persian Empire under Cyrus the Great. Naturally, the Persians did not continue the Neo-Babylonian effort to revive Babylonian religion as a centralizing imperial project, and the expensive temple constructions were stopped. Rather than impose their own foreign religion, the Persians let the different subject peoples practice their own religions, most famously letting the Jews return to Judah and thereby ending the Babylonian Captivity. I know of no evidence of any excavations continuing under the Persians.

The next case I’m aware of was 1500 years later, in China under the Song Dynasty. No civilization in recorded history has studied ancient texts as devotedly as Medieval Chinese Civilization. In addition, some scholars closely studied ancient artifacts, mostly bronze vessels over a thousand years old used for the rites and rituals which were central to Ancient Chinese Civilization, and stone stelae which were centuries old. A few antiquarians assembled substantial collections, and some emperors maintained larger collections still. Their studies focused mostly on the inscriptions left in these artifacts, and they published rubbings and catalogues. This work was used to reverse-engineer the ancient rituals and revive them at the imperial court, complete with reconstructed ancient-style bronze vessels and bells. Popular art drew inspiration from the revived ancient styles as well.

Li Ji, who brought Western archaeology to China in the early 20th century, critiqued the Song antiquarians as limited and unscientific by comparison with modern methods, narrowly focused on inscriptions and elaborate art objects. The Song antiquarian tradition eventually died out, perhaps because of the Jurchen and Mongol conquests. The Jurchen invasion destroyed the collection and research notes of Zhao Mingcheng and Li Qingzhao, at the very least.

There may have been other archaeological traditions which have since been lost. If the Carthaginians or the Shang had anything like this, we probably wouldn’t know about it. Within our surviving records, however, these two are the only precursors similar to the Western tradition. This grew out of the Renaissance, when European scholars revived the widespread study of ancient Roman and Greek texts. A handful also devoted themselves to collecting and cataloguing the artifacts, and sometimes digging up ruins in search of more. These early modern antiquarians were initially scattered, preparadigmatic individuals, more in dialogue with the numerous scholars studying the old texts than with each other. The field gradually cohered over the centuries, informed by the development of Western science and empiricism. While there is no sharp dividing line, it’s probably fair to say that archaeology had emerged as a distinct field with its own methods and goals by the late 1700s.

European antiquarians were first interested in Greek and Roman artifacts, as an extension of Renaissance scholars studying Greek and Roman texts. Later on this interest grew to include non-Classical sites and artifacts in their home countries, such as Stonehenge. Then as Western powers dominated the world and sent political and ideological expeditions to distant lands, archaeologists went to sites mentioned in canonical texts like the Bible and the Iliad which had passed into the territory of other civilizations. From there it was a small step to extend archaeology to the entire world.

Why Our Archaeology Won’t Endure

While archaeology is now practiced around the world, we shouldn’t overstate our society’s dedication to it. Serious archaeological work occurs only where it is politically and ideologically convenient. When the government approves of the work, such as the British Museum’s excavations buttressing the prestige of the empire, or recent Turkish backing for the Göbekli Tepe excavations to boost tourism revenue, a great deal is possible. But without the active support or at least the tacit permission of the state, archaeology remains small-scale and marginal, if not suppressed. This is most obvious under governments which actively destroy past artifacts for ideological reasons, such as the Taliban, the Islamic State, or Australia.

In Italy, stifling “legal mechanisms are straightforwardly to blame for throttling archeological discovery”, to the extent that “Many artifacts end up on the black market … or are even simply destroyed or hidden away.” Less dramatically, when a friend of mine led an expedition to find ancient sites in the Amazon, his partner at a Brazilian university advised him not to get the government involved, because they don’t like finds which could interfere with resource extraction. In the United States, the ​​Native American Graves Protection and Repatriation Act gives federally recognized Native American tribes substantial power to veto or censor archaeological investigations, which is often used to block research that holds the possibility of contradicting a tribe’s mythology or complicating their claim to land.

Further east, excavations of Mississippi Mound Builder sites are only occasionally suppressed by the Mississippi Band of Choctaw Indians, who have been granted de facto jurisdiction as one of the few surviving tribes from the region; nevertheless these excavations remain small-scale and niche, in large part because the possibility of finding more evidence that there were once sophisticated native civilizations in North America would be very inconvenient for the civic mythology of both European-descended and Native-descended Americans. The largest investigations of these enormous and poorly-understood mounds were conducted during the Great Depression, when the Works Progress Administration was looking for any excuse to mobilize workers and would gladly hire two hundred men to excavate and catalogue a single site, whereas an excavation today will consist of maybe half a dozen graduate students.

Now that archaeology has been codified and established widely, even if imperfectly, will it persist forever? I doubt it. When a civilization adopts an unusual practice which does not confer a large, immediate material advantage, then that practice can persist for a long time, sometimes until the civilization itself collapses. It usually doesn’t persist into future civilizations, however. The Ancient Chinese rites were lost with the fall of Ancient Chinese Civilization, which is why Medieval Chinese antiquarians and Modern Chinese archaeologists are forced to guess about those rites based on partial descriptions in texts and the artifacts which happened to survive. Archaeology is much less central to our society than Ancient Chinese rites were to theirs. A closer comparison to archaeology’s importance might be Medieval Western anchorites, a religious practice that was extremely important to its practitioners and housed for centuries within the dominant intellectual institution, the Catholic Church, but was essentially abandoned by the modern era.

Among bookish people, a common joke is to talk about present society in the tones of a distant, far-future archaeologist looking back on our own society’s remains. I first encountered this as a lesson in my middle school history classes. The idea appeared at least as early as 1952, in Clifford Simak’s Hugo Award-winning novel City, and was popular on Tumblr back when I hung around in the 2010s. I’m sure this isn’t meant to be taken too literally. Even if it’s funny to describe a drive-through restaurant in the ritualistic tones of academese, that doesn’t mean serious thinkers really believe future archaeologists will continue using our exact academic forms of speech. Nevertheless it seems to me that most people making this joke are accepting the core claim, that future scholars will study our civilization’s remains with all the methods we use to study our forebears. That’s a strong claim, in need of a strong justification.

Human civilizations have varied tremendously, and will continue changing for as long as anything vaguely recognizable as human remains alive. Our own civilization is not the final form of Man sitting at the End of History. There are many examples of practices that used to be much more common than archaeology, and were much more load-bearing to the civilizations which practiced them, but have entirely died out. Over the next millennia, the successors to our civilization will abandon many things which we take for granted, for both good reasons and bad.

For example, human sacrifice used to be ubiquitous. If you tried to tell a city-dweller that maybe one day people would just stop, they’d look at you like you’d grown a second head. Then, over the course of the first millennium B.C., human sacrifice was abandoned by every major Old World civilization. Animal sacrifice persisted for longer, but is also basically gone. How many Roman priests ever considered that might happen?

Of course, human civilization is not infinitely malleable. At the extreme, if humans are still around in two thousand years, then it seems extremely likely that there will still be hierarchies and private property and ideologies, and things more or less like government and money, as we have observed in every known civilization. For practices which are relatively unusual, where we have many examples of civilizations expanding and enduring without them, we cannot assume they will persist without a very strong reason.

Some newer practices, such as industrial mass production, provide such overwhelming and immediate advantages that they will probably persist in some form, unless and until they are supplanted by something even more powerful. An older example is minted coins of precious metal. These were invented in the 7th century B.C., and replaced commodities valued by weight as the main form of money. This endured for over two thousand years; by now, perhaps coined money has permanently been obsoleted by fiat money, or perhaps our experiment with fiat money will eventually be remembered as a temporary and aberrant detour which ultimately returned to minted coins, like Medieval China’s centuries-long use of fiat money that was aborted by the Ming dynasty. It is too soon to tell. But however that shakes out, trading barley and loose silver by weight will not become the dominant form of money again.

Some people tell me that archaeology is like industrial production or better forms of money, and it will persist because it provides material advantages. Unfortunately, archaeology does not come close to meeting this bar. The material technology we gain is mostly a matter of curiosities, since our own technology has advanced far beyond past civilizations, just as our successors will very likely have better technology than we do. It is possible, and in fact fairly common, to reverse-engineer social technology from the distant past and use it as inspiration for creating new functional institutions. However, every case I know of this happening drew almost entirely from studying ancient texts, and little or nothing from archaeology.

Different Civilizations Pursue Different Projects

Archaeology is mostly done out of spiritual motivations, fueled by the drive to understand Man’s place in the universe and by the quest for Truth for its own sake. We do not study the Antikythera mechanism in order to use its once-advanced technology ourselves; we study it because we want to understand where we came from, with the framing and values which our scientific-materialist ideology puts on that question. As a believer in this ideology, I try to answer these questions for their own sake, but the material benefits are distant and intangible, more akin to how anchoritism supported Medieval Western ideology. Ideologies provide direction, cohesion, and spiritual power, and are not to be taken lightly even if all this is hard to quantify. However, practices which support one ideology will not necessarily support a later one. A Phoenician might argue that infant sacrifice would persist indefinitely because it supports the social structure, and this was true of his society, but it is not true of ours. In addition to the ideological motivation, archaeology can be a field for impressive achievement which demonstrates a state’s capacity, much like the Pyramids of Giza or the Apollo Moon landings. The form a state chooses for its demonstrative achievements is an expression of its ideology. Even today, when archaeology conflicts with state ideology, it is often suppressed.

Some people tell me that archaeology will persist because the basic idea of closely studying ancient artifacts is simple, and now that it’s been figured out, people are not going to forget. There are ideas which are very simple and very obviously useful, and so were quickly adopted once articulated and never lost after that, but nevertheless were not developed until late in human history. Examples include phonetic alphabets, the numeral zero, calendars with numbered years, and Cartesian coordinates.

However, while the basic idea of studying artifacts is simple, doing so productively is difficult and complex. The methods of excavating sites and structures, recording finds in a way that others can understand, preserving and storing artifacts, detecting fakes, dating sites and artifacts, and inferring the shape of past societies from the artifacts recovered are all difficult to do well. Developing these methods took generations of refinement. This is not a simple idea which can be adopted by anyone who hears the concept explained clearly. This is a live tradition of knowledge, handed down through immense effort to teach the next generation. Individual scholars in isolation, without an entire field to draw on, have historically achieved pretty limited results, as we saw with the Song and Renaissance antiquarians.

Back in Babylon, the basic idea of restoring the temples was apparently remembered long after the practice ceased, but that wasn’t enough to revive it. When Alexander the Great conquered the city after two centuries of Persian rule, Arrian tells us that he intended to rebuild the main temple, and Diodorus Siculus tells us that this was a suggestion from the Babylonian priests. However, the project failed and the temple was not rebuilt. The execution of a difficult project requires far more than just the idea.

Like any tradition of knowledge, archaeology can and eventually will die out. What we learned to do once, we can learn to do again, and in many fields such as astronomy or portraiture, we see traditions of knowledge get lost and rederived many times throughout history. If future civilizations care as much as we do, they will be able to figure out archaeological methods at least as good as ours. But if they don’t care enough to devote a great many full-time specialists then the mere idea of studying old artifacts won’t get them very far.

This cycle has already happened at least once. The Neo-Babylonian proto-archaeological tradition died, and for eons we know of nothing else that came close, even though their successors remembered the idea for a while. Time passed, Rome rose and fell, the Medieval Europeans built their castles and cathedrals, in Early Modern times they settled the New World—and only then, 2300 years after it was lost, was archaeology independently reinvented. If and when it is lost again, archaeology will probably be rederived, if you don’t mind waiting a millennium or three. By then, most of the artifacts that could be used to reconstruct our own society will be lost, and only a few isolated shards will remain. The future traditions will have very different goals and ends from our archaeologists, and think of their work very differently, but probably the systematic analysis of ancient artifacts will eventually be redeveloped to a high level—and then lost again, and so forth. In any given future year there probably won’t be a live archaeological tradition.

Some people tell me that archaeology will persist because now we are rich enough to fund projects like this, whereas before industrial technology, societies could not afford to do so. Greater wealth will make it somewhat easier, and better technology will continue widening the scope of what’s possible to recover, but I think this will have only a small effect on how much people care to apply their wealth and technology to this particular task. There are lots of things that people can do with their wealth, and for universal desires like food or clothes, it’s pretty safe to assume that people will produce more as they get richer. For more specific or more esoteric things which other civilizations care less about, this assumption does not hold. We can assume that richer civilizations will generally produce more entertainment, but we can’t assume that richer civilizations will produce more superhero movies, for the same reason that a 19th-century American would be incorrect in assuming that a richer civilization would memorize more poems. Up through the early 20th century it was common for governments or even wealthy individuals to produce beautiful buildings and public spaces for the public. Today, this practice—which is less fundamental than food but more fundamental than archaeology—is mostly extinct, even though we are far richer than in the days of Georges-Eugène Haussmann or Andrew Carnegie.

Nebuchadnezzar was rich enough to investigate and reconstruct ancient temples all the way back in the Iron Age. Alexander was rich enough to continue the work, although he failed for other reasons. Any other king capable of mobilizing scholars and laborers had the wealth to study ancient sites, if they cared to. Our greater wealth means our study of ancient artifacts occurs on a much larger scale, employing far more people than Nebuchadnezzar could, but this is not the reason we care enough to actually do it. The Song dynasty was also much richer than Nebuchadnezzar, and their society was more scholastic in general, yet their study of ancient artifacts in particular seems somewhat behind the Neo-Babylonians’.

The Aztecs were a very poor civilization with extremely primitive technology, even lacking metalworking and the wheel. They nevertheless maintained some very expensive practices, including mass human sacrifice, constructing monumental pyramids, and keeping a massive royal zoo. The Aztecs were conquered by the Spanish, who imposed a new civilization that was much richer and far more technologically advanced, yet they discontinued all of these practices until Mexico became independent in the early 19th century. A zoo was eventually reestablished in 1924, not because Mexicans became richer and naturally conceived a desire to spend their wealth on exotic animals, but because public zoos had recently become a cultural marker of city pride and sophistication as an outgrowth of Western scientific-materialist zoology. To this day human sacrifice and monumental construction have not been reestablished, even though modern Mexican society is incalculably richer than the Aztecs. The Great Pyramid of Cholula, which predates the Aztecs and was already abandoned when the Spanish arrived, is much larger than any monument built in Mexico with industrial technology. It will probably outlast all monuments ever built by industrial civilizations anywhere on Earth.

Over the very long run, archaeology will probably follow a trajectory similar to the building of monumental pyramids. Different civilizations, widely separated in time and space, independently decided that building pyramids was very important and established sophisticated engineering practices for the purpose. Most civilizations did not, and find the whole thing baffling and foreign, even if they can see some exotic romance in it. When a pyramid-building civilization collapses, its successor usually does not continue the practice. This is true even if the successor civilization, like our own, is far richer. The systematic study of ancient artifacts is much rarer historically than the construction of monumental pyramids. This will probably remain true in the distant future.

It is a tragedy, then, that our current economic policy does exactly the opposite. We are vastly poorer because, instead of supporting “infant industries” until they can stand on their own feet, the U.S. government has spent trillions of dollars to keep the most senile and sclerotic businesses on life support. This keeps millions of talented people and tens of trillions of dollars worth of plant and equipment locked up in decrepit enterprises run by mediocrities who specialize in preserving the status quo, or by outright incompetents running their businesses into the ground.

In the words of the economist Joseph Schumpeter, “[T]he problem that is usually being visualized is how capitalism administers existing structures, whereas the relevant problem is how it creates and destroys them.” If the government were willing to let big businesses die a natural death when their time comes, then these resources would be captured by better-managed firms, and our industrial and technological growth would proceed faster. In 2024, the largest 0.1% of businesses, those with over 500 employees, accounted for 54% of private sector jobs in the U.S. Firms with over 10,000 employees accounted for 18% of U.S. manufacturing jobs. Our prosperity, and especially our grandchildren’s prosperity, depends critically on how much of this is in the hands of companies like Boeing and how much is in the hands of companies like SpaceX.

The U.S. Government Keeps Zombie Companies Alive

If you listen to your schoolteachers and economics textbooks and the pundits on television and on Twitter, then you will learn that we live in a capitalist free market society. This means, we are told, that economic activity is a Darwinian contest of all against all. The most effective and efficient firms provide better products at lower prices, taking business away from their inferior competitors. Firms which cannot keep pace with the march of progress wither to dust and are blown away on the wind. In this way, the ruthless logic of the competitive market provides consumers with the best products at the best prices, driving improved living standards, technological development, and economic growth.

This has been roughly true in some times and some places—more on this later. But if you look at the present economy of the U.S. or any major nation, things seem very different now. Major companies simply do not collapse. If bankruptcy threatens, then the government will bail them out at any cost. If there’s a big disaster, then an individual executive might be fired, or not—more likely not. But either way, the company itself will persist. No matter how inefficient a company like General Motors or Boeing might become, no matter how much better and cheaper its competitors make their products, the state will not allow a major incumbent to die.

This is done to avoid the economic and political pain of companies failing. If the collapse of a major company were permitted, hundreds of thousands of people would lose their jobs. In the wake of the 2008 crash, General Motors went bankrupt. If it had collapsed and ceased operations, its 88,000 employees in the U.S. would have lost their jobs, along with probably a similar number of employees at companies directly supplying GM. This, in turn, would have caused devastating ripple effects as former GM employees stopped patronizing restaurants, buying homes, upgrading cell phones, and generally cut back spending.

Of course, these ripple effects are exactly what an economic depression consists of, and GM’s bankruptcy was itself triggered by ripple effects from the collapse of mortgage-backed securities. Rather than let the dominoes keep falling, the U.S. Treasury lent $50 billion to carry GM through bankruptcy and restructuring and preserved about three-quarters of the company’s U.S.-based jobs. Without this and similar bailouts, the 2008 depression would have been far worse than it was, at least in the short run.

In the long run, however, this retards U.S. industry below its potential by preventing badly-needed industrial restructuring. With the support of the explicit government loans and the implicit guarantee of an infinite federal backstop, GM has returned to profitable operations, but they are not actually very good at advancing American industry. Their best-selling cars are all variations of designs that are decades old at least, like the Chevrolet Silverado pickup truck, GM’s most popular product by volume, the first generation of which was sold in 1999.

The major advances in automaking technology—electric vehicles, self-driving vehicles, and, to a lesser extent, manufacturing process improvements like casting the chassis out of two large pieces rather than dozens of small pieces—are being driven by younger upstart companies without state backing. GM’s own entry into these advanced arenas has been lackluster at best. In 2024, electric vehicles accounted for only 4.2% of GM’s sales. Its foray into self-driving cars got underway with the 2016 acquisition of Cruise, an external startup company building self-driving taxis. Cruise began operating autonomous taxis in San Francisco, but their license to operate was revoked in 2023 after a Cruise vehicle struck a pedestrian, and the company attempted to withhold data on the accident from California officials. Cruise never recovered, and in 2024 GM shuttered the project.

An area related to this, but distinct, is the support of basic research, which can serve as a seed for industry. The U.S. excels at this, and while there is always room for improvement, we should mostly keep doing what we’re doing. Most famously, the Defense Advanced Research Projects Agency (DARPA) and the National Science Foundation housed and supported the projects which eventually became the internet. Or to return to the self-driving car example, the technology got started with DARPA challenges from 2004-2007, which incubated the teams of engineers that would go on to develop the first self-driving cars by offering cash prizes of up to a couple million dollars. That’s incredible bang for the buck, but not very many bucks. Most of the cost of research and development of self-driving cars has been borne by Google, which has probably spent billions on Waymo so far. The tasks of foundational R&D and of economy-scale economic policy are separate. The U.S. government is far better at supporting the former than the latter.

Prosperity Comes From Technology, Not From Full Employment

Much of the justification for our current policy of bailing out companies like GM was laid by the Keynesian economists, who take their name from the school’s founder, John Maynard Keynes. His magnum opus, The General Theory of Employment, Interest and Money, gives an algebraic model which argues that if technology and “technique” are held constant, then business cycles lead to depressions where employment and economic output are low, leading to unemployment and poverty with no purpose. This forms the main intellectual justification for the now-ubiquitous practice of massive “stimulus” spending in the face of economic depressions, where the government spends trillions of dollars on anything and everything in order to preserve employment and mitigate the slump, although the main political justification is simply that it wins votes from the money’s recipients.

Of course, technological improvement is by far the most important economic force in the world today, and any theory which assumes it away in order to simplify the algebra will be useless as a guide to steering a modern economy. Keynes repeatedly specifies that his theory is not concerned with technological development and focuses only on maintaining full employment assuming a static technology level. His theory’s recommendations do indeed help towards that goal.

However, the reason we are prosperous today is not because our forefathers maintained full employment. We are prosperous today because our forefathers developed and deployed better and better technology—cars, antibiotics, forge presses, container ships, cryogenic storage to transport liquefied natural gas, and thousands more. A society with full employment and the technology that existed in Keynes’s lifetime is vastly poorer than a society with high unemployment and today’s technology. Very few people would prefer being gainfully employed in the boom years of the late 19th century Gilded Age over struggling to find work in a sluggish 2020s economy—say, in Canada or Portugal.

In large part, this is because of the greatly expanded social spending and transfer payments, made possible by the gigantic increase in goods produced, which make it so that an unemployed American today consumes far more goods than a hardworking laborer or landowning farmer from the Gilded Age while living in bigger, better housing. Of course, the reason that Gilded Age society did not provide this is not because they were too unenlightened but because they were too poor. Most estimates place the U.S. per capita GDP in the year 1900 at around $10,000 in 2024 dollars.

While comparing GDP across centuries is necessarily a shoddy and incomplete affair due to poor data and the limitations of the concept of “GDP,” it is nevertheless a useful starting point. In 2021, the  Medicaid program alone spent $7,593 per enrollee, or about $9,000 in 2024 dollars. This one social program provides the poor with roughly the entire income of an average Gilded Age American, to say nothing of the plethora of other social services and transfer payments which America is wealthy enough to afford today.

If we take seriously that technological and industrial growth is the source of our prosperity, the source of the goods which are paid to workers for their effort and given to the poor out of charity, then the main economic questions become those of technological and industrial growth. These were best modeled by Joseph Schumpeter, another of the 20th century’s great economists. He described the process of “creative destruction” which creates new industries and raises standards of living. The act of creating the radio industry destroys the telegraph industry, creating the lightbulb destroys the gaslamp, creating the supermarket destroys the butcher shop, and creating the ridehailing app destroys the taxi.

Businessmen today euphemize this process as “disruption” rather than “destruction,” but it does destroy industries and livelihoods. This naturally provokes political opposition from incumbents, of course. Schumpeter calls economic history a “history of revolutions”—and revolutions are always contested. America’s great advantage, the unique source of its wealth and power for eighty years, is that it is the only place where the prevailing culture and the state do not strangle novel industries at the request of the old. Uncle Sam might not let Boeing die, but he will not kill SpaceX to protect them, unlike his counterparts elsewhere in the West.

The new products and production processes which surpass and destroy the old do not come from theoretical “perfect competition” among near-identical firms offering near-identical products, perhaps with some incremental improvements. Rather, in Schumpeter’s theory, these advances come from “competition from the new commodity, the new technology, the new source of supply, the new type of organization … competition which commands a decisive cost or quality advantage and which strikes not at the margins of the profits and the outputs of the existing firms but at their foundations and their very lives.” As Peter Thiel’s influential Zero to One echoed seventy years later, such a decisive advantage—a “moat,” in today’s business language—naturally leads the company to become a near-monopoly because other firms cannot replicate the advantage for years or even decades. For example, today ASML’s moat is the highly skilled and specialized engineers who create its extreme ultraviolet lithography machines, which no one in the world can replicate, and Amazon’s moat is its extremely effective logistics processes and staff.

This dynamic is important not just because the new replaces the old but also because the fear of being replaced can drive incumbents to preemptively improve their own products. The constant, implicit threat from the “perennial gale of creative destruction” means that even corporate giants cannot rest on their de facto monopolies. This mechanism still operates in businesses that lack a blank check from the government, leading to occasional moves such as when Netflix pioneered streaming video and creatively destroyed its own DVD-by-mail service, or when Amazon adopted the strategy of serving as a “marketplace” for third-party sellers directly competing with Amazon’s own offerings, who now account for over 60% of Amazon’s sales. More frequently than supplanting their own core products, functional conglomerates often pursue creative destruction in other fields, such as Google’s foray into self-driving cars with Waymo, or Amazon’s creation of Amazon Web Services cloud computing.

Industrial Greatness Requires Economic Depressions

A clear understanding, however, must bear in mind the serious costs of creative destruction. The most obvious is the jobs of the workers in industries that are creatively destroyed, who must now cope with gradually declining prospects forever or face the wrenching financial and psychological strain of entering a completely new field. While it is eventually better for the economy as a whole, that is little consolation in the moment.

Even more importantly, unleashing creative destruction’s full potential for rapid progress also causes massive economic depressions. It is no coincidence that the period when depressions were most frequent and most destructive, about 1850 to 1945, was also the fastest period of technological progress in human history, witnessing the rapid introduction of electricity, flight, antibiotics, automobiles, cheap steel, artificial nitrogen fertilizer, radio, and television, and countless other transformative technologies which completely overturned patterns of daily life at a pace and scale which eclipses even today’s rapid technological advances.

The Great Depression of 1929 is popularly remembered as a unique catastrophe, but it was actually the last of a series of economic crashes more frequent and more devastating than anything in living memory. Even the term “Great Depression” was originally used for what is now called the Panic of 1873 until the term was repurposed for more recent American mythmaking. Since then, FDR’s New Deal policies cut the rate of major depressions in half, if not further, and the U.S. has seen only two depressions of historical significance, in 1973 and in 2008.

There are two reasons why unfettered creative destruction leads to depressions. The first reason is the weaker of the two, but is still visible today. The bonanzas that come with big economic improvements often lead to wild hype cycles and inflated expectations, which must eventually come crashing back to Earth. The profusion of hustlers, get-rich-quick schemes, half-informed speculators, and earnest projects which simply aren’t very good, all surrounding a core of extremely valuable businesses applying the new technology, is a pattern we have all seen. Eventually the wheat is separated from the chaff, ill-founded companies and financial schemes collapse, and an economic downturn follows. Meanwhile, the core technology which prompted the boom and bust continues to grow and mature with only minor interruptions from the storm and fury of the surrounding business cycle.

The internet dot-com bubble around the year 2000 is the clearest of several cases in living memory. This also happened many times in our ancestors’ days, most notably with railroads. This is of limited importance, however, because an economy always contains local collapses which, if the underlying conditions are right and the local collapse is big enough, can set off a chain reaction and depress the economy as a whole. Underlying conditions matter more than an additional source of booms and busts.

The other, far more fundamental reason that these are connected is that depressions do not destroy businesses at random. Those which go bankrupt are disproportionately the ones which were most precarious and marginal in good times, whereas those which normally are comfortably profitable and well-run are more likely to weather the storm. The effect is similar to the pop science idea that predators make a herd healthier by disproportionately removing the sickly and selecting the population for those with better health, although it is a much stronger effect because when an inefficient firm collapses, the resources it uses—such as engineers, laborers, real estate, and machine tools—are not devoured by lions, but are eventually redeployed either to old businesses which survived the depression or to newly-created businesses. This is an ugly process, involving mass layoffs, desperate scrambling, emergency sales of assets at fire-sale prices, and everything else that economists euphemize as “transaction costs,” but the end result is more of our industrial resources controlled by men like Thomas Edison and Elon Musk, and fewer resources controlled by men like James McNerney and Jack Welch.

One error of most economic theories, including Keynes’s, is that they assume firms making similar products using similar assets are more or less interchangeable, that economic productivity is a sort of fluid which can seamlessly flow from one container to another until efficient markets have leveled them off at the same height. In fact, there are immense differences between firms. The difference between Boeing and SpaceX today, or the difference between Henry Ford and his initial competitors like Buick and Oldsmobile, is not a matter of marginally higher profits which can be modeled in financial and algebraic terms. It is a fundamentally different industrial logic—one that, in extreme cases, can do more cheaply what the competitor cannot do  at any price. These advantages can persist for years, even decades, but eventually they diffuse throughout society and are surpassed in turn. Henry Ford’s assembly lines would be considered crude and poorly organized today.

There is a good deal of talk about the psychological disciplining force of market competition and how having “skin in the game” causes people to make better decisions. This effect is real but should not be overstated. There are many, many people who will not make excellent decisions no matter what pressure they’re under. The main way the market produces progress is not by psychological force but by selection, by shrinking or destroying the institutions which do not measure up and reallocating their staff, equipment, and other assets to those which overperform. When government policy keeps inefficient megacorporations alive, it ipso facto retards industrial and economic progress. This is why the best government policies do not single out individual firms to be kept alive by arbitrary bailouts but intervene on the market broadly and let those who are most competent win the resulting contest.

If we want our descendants in 2125 to surpass our living standards the way we surpass our ancestors from 1925, then we will have to permit economic transformations at the scale that our ancestors did, including bankruptcies, job losses, and the cascading depressions that result. The individual pain of depressions does not have to be quite so severe as it once was. Because we are richer, we can and do spend vastly more on welfare, but this should be directed at individuals rather than at megacorporations. But there will always be some pain.

If we do our jobs, then during depressions, our grandchildren will still complain that they can only afford a 4,000 square-foot house, that they can no longer justify taking a one-hour trip from Austin to New York City every weekend and must cut their visits home back to once a month, and, of course, that they must abandon the career they poured their heart and soul into to find another. If they do their jobs, then their grandchildren, our great-great-grandchildren, will complain about maintaining their Moon estates, how long it takes a middle-class family to save up for their own geoengineered private island, or who knows what strange new wonders. But if we want to see this new wealth created, whether the unpredictable dreams of the distant future or just building out proven prototypes so we can all have unerring robots drive our electric cars for us ten or fifteen years from now, then we must accept the short-term costs.

Ironically, by spreading the belief that superintelligent AI is achievable and supremely powerful, these “AI Doomers,” as they came to be called, inspired the creation of OpenAI and other leading artificial intelligence labs whose technology they argue will destroy us all. Despite this, they have continued nearly the same advocacy strategy, and are now in the process of persuading Western governments that superintelligent AI is achievable and supremely powerful. To this end, they have created organized and well-funded movements to lobby for regulation, and their members are staffing key positions in the U.S. and British governments.

Their basic argument is that more intelligent beings can outcompete less intelligent beings, just as humans outcompeted mastodons or saber-toothed tigers or neanderthals. Computers are already ahead of humans in some narrow areas, and we are on track to create a superintelligent artificial general intelligence (AGI) which can think as broadly and creatively in any domain as the smartest humans. “Artificial general intelligence” is not a technical term, and is used differently by different groups to mean everything from “an effectively omniscient computer which can act independently, invent unthinkably powerful new technologies, and outwit the combined brainpower of humanity” to “software which can substitute for most white-collar workers” to “chatbots which usually don’t hallucinate.” 

AI Doomers are concerned with the former scenario, where computer systems outreason, outcompete, and doom humanity to extinction. The AI Doomers are only one of several factions that oppose AI and seek to cripple it via weaponized regulation. There are also factions concerned about “misinformation” and “algorithmic bias,” which in practice means they think chatbots must be censored to prevent them from saying anything politically inconvenient. Hollywood unions oppose generative AI for the same reason that the longshoremen’s union opposes automating American ports and insists on requiring as much inefficient human labor as possible. Many moralists seek to limit “AI slop” for the same reasons that moralists opposed previous new media like video games, television, comic books, and novels—and I can at least empathize with this last group’s motives, as I wasted much of my teenage years reading indistinguishable novels in exactly the way that 19th century moralists warned against. In any case, the AI Doomers vary in their attitudes towards these factions. Some AI Doomers denounce them as Luddites, some favor alliances of convenience, and many stand in between.

Most members of the “AI Doomer” coalition initially called themselves by the name of “AI safety” advocates. However, this name was soon co-opted by these other factions with concerns smaller than human extinction. The AI Doomer coalition has far more intellectual authority than AI’s other opponents, with the most sophisticated arguments and endorsements from socially-recognized scientific and intellectual elites, so these other coalitions continually try to appropriate and wield the intellectual authority gathered by the AI Doomer coalition. Rather than risk being misunderstood, or fighting a public battle over the name, the AI Doomer coalition abandoned the name “AI safety” and rebranded itself to “AI alignment.” Once again, this name was co-opted by outsiders and abandoned by its original membership. Eliezer Yudkowsky coined the term “AI Notkilleveryoneism” in an attempt to establish a name that could not be co-opted, but unsurprisingly it failed to catch on among those it was intended to describe.

Today, the coalition’s members do not agree on any name for themselves. “AI Doomers,” the only widely understood name for them, was coined by their rhetorical opponents and is considered somewhat offensive by many of those it refers to, although some have adopted it themselves for lack of a better alternative. While I regret being rude, this essay will refer to them as “AI Doomers” in the absence of any other clear, short name.

Whatever name they go by, the AI Doomers believe the day computers take over is not far off, perhaps as soon as three to five years from now, and probably not longer than a few decades. When it happens, the superintelligence will achieve whatever goals have been programmed into it. If those goals are aligned exactly to human values, then it can build a flourishing world beyond our most optimistic hopes. But such goal alignment does not happen by default, and will be extremely difficult to achieve, if its creators even bother to try. If the computer’s goals are unaligned, as is far more likely, then it will eliminate humanity in the course of remaking the world as its programming demands. This is a rough sketch, and the argument is described more fully in works like Eliezer Yudkowsky’s essays and Nick Bostrom’s Superintelligence.

This argument relies on several premises: that superintelligent artificial general intelligence is philosophically possible, and practical to build; that a superintelligence would be more or less all-powerful from a mere human perspective; that superintelligence would be “unfriendly” to humanity by default; that superintelligence can be “aligned” to human values by a very difficult engineering program; that superintelligence can be built by current research and development methods; and that recent chatbot-style AI technologies are a major step forward on the path to superintelligence. Whether those premises are true has been debated extensively, and I don’t have anything useful to add to that discussion which I haven’t said before. My own opinion is that these various premises range from “pretty likely but not proven” to “very unlikely but not disproven.”

Even assuming all of this, the political strategy of the AI Doomer coalition is hopelessly confused and cannot possibly work. They seek to establish onerous regulations on for-profit AI companies in order to slow down AI research—or forcibly halt research entirely, euphemized as “Pause AI,” although most of the coalition sees the latter policy as desirable but impractical to achieve. They imagine that slowing or halting development will necessarily lead to “prioritizing a lot of care over moving at maximal speed” and wiser decisions about technology being made. This is false, and frankly very silly, and it’s always left extremely vague because the proponents of this view cannot articulate any mechanism or reason why going slower would result in more “care” and better decisions, with the sole exception of Yudkowsky’s plan to wait indefinitely for unrelated breakthroughs in human intelligence enhancement.

But more immediately than that, if AI Doomer lobbyists and activists like the Center for AI Safety, the Institute for AI Policy and Strategy, Americans for Responsible Innovation, Palisade Research, the Safe AI Forum, Pause AI, and many similar organizations succeed in convincing the U.S. government that AI is the key to the future of all humanity and is too dangerous to be left to private companies, the U.S. government will not simply regulate AI to a halt. Instead, the U.S. government will do what it has done every time it’s been convinced of the importance of a powerful new technology in the past hundred years: it will drive research and development for military purposes. This is the same mistake the AI Doomers made a decade ago, when they convinced software entrepreneurs that AI is the key to the future and so inspired them to make the greatest breakthroughs in AI of my lifetime. The AI Doomers make these mistakes because their worldview includes many assumptions, sometimes articulated and sometimes tacit, which don’t hold up to scrutiny.

There is No Reason to Think Superintelligence is Coming Soon

AI Doomers believe that superintelligent AGI is coming very soon, generally within five or ten years. They engage each other in a grand debate over “timelines” to such an AGI, despite the lack of any solid grounding for this debate. They rarely advance concrete arguments for any particular timelines, and when they do, the arguments are not sound. More often, they justify this position by pointing to the loose vibes and intuitions of technologists and scholars—or rather, the intuitions of those technologists and scholars who happen to agree with them, since there is certainly no consensus in the field. 

There is no particular reason to think this type of reasoning can accurately predict breakthroughs for any technology, and it has not accurately predicted past breakthroughs, so there is no substance to any of these “timeline” predictions. The upshot is that, while there are internally-consistent philosophical arguments that AGI is possible, there is no good argument for when it might be built. Some of the AI Doomers object that they can dodge the necessity of justifying their views on timelines by stating their timelines in the form of a probability distribution, but of course this does not work, since such a probability distribution is nothing but a way of expressing a particular viewpoint, which must still have been produced by reasons—perhaps good reasons, perhaps bad reasons.

It’s worth noting that the sharpest AI Doomers, those closest to the spring of the intellectual waterfall, are not strongly wedded to “short timelines.” Among this crowd, the dominant view is that AGI takeover is fairly likely to happen soon, and this is a sufficient reason for extreme action, but that the strong confidence and drive towards quantification among the mass of followers is unjustified. Eliezer Yudkowsky, the father of the movement, has spoken against “AI timelines,” among other notables.

As an aside, this is a common split in the beliefs of utopian-apocalyptic movements. The leading thinkers hold that the world will be irrevocably transformed sometime, which might come soon or late, and are acutely aware that it might not be imminent. Among the early Christians, the Apostles themselves were careful to specify that no one could know exactly when Christ would return, but emphasized that it might happen any day. It was the mass of their followers who expected the Second Coming in their own lifetime. Similarly, in the 19th and 20th centuries, Marx himself and the leading communist intellectuals claimed only that communism was eventually inevitable, and it was their followers who confidently expected to see the end of capitalism themselves.

The leading intellectuals of the 20th century Gaian environmentalists, in works like The Limits to Growth, were careful to specify that their mathematical models showing the imminent collapse of industrial society were mere illustrations and possibilities of an eventual exponential trend, but enough of their followers took them at face value to implement China’s one-child policy and India’s coercive sterilization program, while in freer nations many families voluntarily had fewer children. Of course these psychological patterns in how people and movements react to utopian-apocalyptic eschatologies do not directly bear on whether any particular utopian-apocalyptic claim is true or false, but these patterns are worth understanding anyway.

Society’s Response to Artificial General Intelligence Does Not Improve Over Time 

AI Doomer groups agitate for an “AI Pause,” or other methods of slowing AI development and “buying time.” The argument they give is that humanity is doomed by default because research to make AI more powerful progresses faster than research to align AI with human values. The current structure of society makes this all but inevitable, so if we are to avoid doom, then new policies are necessary, although no one has yet articulated what policies would produce an aligned AGI. Regardless, if the critical moment occurs in, say, 2032 instead of 2027, that gives more time for the world in general, and the U.S. in particular, to get its act together rather than following the natural incentives to carelessly build an unaligned superintelligence.

This is a strange and unjustified assumption! I would not say that the U.S. has been getting its act together, lately. In fact it seems like it’s becoming more and more of a mess as time goes on. It seems clear that an Obama administration would’ve done a better job at navigating the deployment of a new world-changing technology than a Biden administration, and an Eisenhower administration would’ve done a better job than an Obama administration. In ten years we’ll probably look back and long for the relative functionality of the Biden administration and lament that our government is now less able to deal with complex problems. The overall decline in competence is no longer a controversial position, so I won’t relitigate that at length. The upshot is that, so long as this goes on, kicking the can down the road by five or ten years makes things worse.

Personally, I don’t think this is especially worrying with regard to AGI, because I don’t see any reason to think AGI is coming soon—though I’m pretty worried about it for other reasons! The functionality of any social system waxes and wanes, with brief sharp spikes where new functional institutional templates are designed and propagated, then long periods of gradual decay, until the next functional form is created internally or imposed from without. Or, if you don’t want to get into how all the cycles work, then hopefully you’ll at least agree with the more general claim that sometimes a society gets more functional and sometimes it gets less functional, and lately ours has been getting less functional.

Assuming superintelligent computers are coming at all, I see no good reason to expect they will come during the current cycle rather than the next one, or two cycles from now, or five, or twenty. It certainly does matter whether AGI comes during a peak of functionality or a trough of incompetence, but without a justified model of when it’s coming—and I will repeat ad nauseam that nobody has one of those—interventions to change its timing are fumbling in the dark. An arbitrary delay could push the invention of AGI into a less competent period as easily as it could push it into a more competent period. Figuring out whether a delay is good or bad requires specific knowledge of both AI timelines and of changes in institutional functionality. “AI Pause” proponents have neither.

But if I were somehow convinced that AGI was coming in the current cycle—and again, there is no good reason to think this—then I would want it to come sooner, before the decay metastasizes even further.

Trying to make it more likely that the superintelligence comes at a peak of institutional functionality makes sense as far as it goes, but rather than trying to make the presumed-inevitable creation of AGI come at a particular moment, this can be done better by improving the health of the relevant institutions in general. Government representatives and bureaucrats capable of understanding the fields they regulate, universities purged of rampant fraud and quackery, technology corporations led by ascended engineers who understand their own production process rather than being driven into the ground by financiers—if AGI happens to come in the near future, any of these would make it far more likely that society responds competently. And if it happens to be further away than that, or to be unachievable for some as-yet-unknown reason, then improving our core institutions will have been a much better use of a career than kneecapping beneficial technology over fears that never materialize, like the overpopulation degrowthers or the anti-nuclear-power activists. If everything depends on our competence at an unknown critical moment, then it is better to build competence in our own time just in case it comes soon, rather than blindly try to match a completely unknown timeline for the arrival of AGI to an at-best-vaguely-guessable timeline for institutional functionality.

Unless, of course, you plan to take power personally, and need more time to do it. From that perspective, slowing things down becomes coherent.

Your AI Safety Coalition is Not Special 

The remaining justification AI Doomers give for why the societal response to AGI improves over time is that, over time, the AI Doomer coalition will gain power, and that, once in power, it will be uniquely able to make sure the superintelligent computer creates a utopia rather than exterminating all mankind.

The justification for expecting the AI Doomers to gain in power is straightforward and solid. When an organized group of young ideologues of the right skills and background sets their mind on gaining power and then works at it diligently for decades, they frequently succeed. And in fact the AI Doomers have done very well at taking power. Most notably, the household of husband and wife Holden Karnofsky and Daniela Amodei has overseen the disbursement of billions of dollars of charitable donations through GiveWell and the Open Philanthropy Project, including approximately all funding for all Effective Altruist organizations, and hundreds of millions for AI R&D and regulation, and separately controls Anthropic, a leading AI developer which created the Claude series of chatbots.

Elsewhere, Jason Matheny, a longtime Effective Altruist who researched existential risk at Nick Bostrom’s Future of Humanity Institute, has risen through the U.S. security state and currently leads the RAND Corporation. In 2023, the Open Philanthropy Project gave RAND $10.5 million for “AI governance” work, “to be spent at the discretion of RAND president and CEO Jason Matheny.” He was briefly a trustee of Anthropic’s Public Benefit Trust from its 2023 announcement until he stepped down months later “to preempt any potential conflicts of interest that might arise with RAND Corporation’s policy-related initiatives.” Many other activists are also doing increasingly well, many of them through the plethora of small Open Philanthropy-funded “AI policy” organizations that provide activists with a stable career track, which Open Philanthropy directed about $15 million to in 2024 alone. These organizations have an increasingly large hand in writing AI legislation, especially for individual U.S. states, and likely soon for European nations.

While their trajectory to power seems solid, the assumption that, once in power, this coalition will be especially able to steer society to develop an aligned AGI, or at least prevent the creation of unaligned AGI, is much more tenuous. One might justify such a claim, firstly by arguing that they have a strong track record of steering society thus far; or secondly, by arguing that the policies they advocate for are especially good; or thirdly, by arguing that they are intrinsically disposed to make better decisions. None of these are true.

Firstly, the AI Doomers’ track record at steering the course of AI development is abysmal. Over two decades of advocacy, they have caused exactly the fate they claim they want to prevent, via a series of own goals which can only be compared to environmentalist opposition to nuclear power causing the construction of coal-burning power plants. Who has accelerated AI development more than the AI Doomers who speak in dire tones of the importance of slowing it down?

Of the recent wave of AI companies, the earliest one, DeepMind, relied on the Rationalists for its early funding. The first investor, Peter Thiel, was a donor to Eliezer Yudkowsky’s Singularity Institute for Artificial Intelligence (SIAI, but now MIRI, the Machine Intelligence Research Institute) who met DeepMind’s founder at an SIAI event. Jaan Tallinn, the most important Rationalist donor, was also a critical early investor. Later, OpenAI was founded by a team that was explicitly inspired by Yudkowsky’s arguments about the power and possibility of computer intelligence, and more immediately by a conference held by the Future of Life Institute (FLI), an Open Philanthropy-funded organization created to popularize Yudkowsky’s views on AI.

At this conference, Elon Musk and Ilya Sutskever signed an open letter on the importance of AI safety, which would prove to be the first of several similar letters; months later, the pair were among OpenAI’s cofounders. Both later left OpenAI. Musk left in 2018, allegedly due to conflict of interest with his AI development at Tesla Motors, and Sutskever was forced out months after his participation in the AI Doomers’ failed attempt to eject CEO Sam Altman in 2023. Leading AI Doomers have since said that they regret their role in launching these AI labs.

In 2017, the Open Philanthropy Project directed $30 million to OpenAI, which was a nonprofit organization at the time. In return, Karnofsky received a seat on OpenAI’s board. In 2021, the brother and sister team of Daniela and Dario Amodei, Karnofsky’s wife and his roommate respectively, left OpenAI to found Anthropic, a rival AI company, citing Rationalist AI alignment ideas as justification. Anthropic soon received a $500 million investment from FTX, whose CEO Sam Bankman-Fried was the second-largest Effective Altruist donor until his arrest for fraud. Karnofsky resigned from OpenAI’s board due to the conflict of interest, and his seat was taken by Helen Toner, an inner-circle Effective Altruist and former Karnofsky employee, who later participated in—and may have instigated—the failed attempt to eject Altman.

The AI Doomers have also built a recruiting pipeline which sources, inspires, and sometimes trains a substantial fraction of the top research talent for all of these AI labs. This was done with the full and explicit backing of Rationalist and especially Effective Altruist leaders from 2015, when the leaders of AI labs were invited to speak and recruit at EA Global conferences, to 2022, when leading AI Doomers began publicly speaking against these labs. Since then the policy has become contentious, but it has not fully stopped.

That AI advocacy would accelerate AI development wasn’t some sort of unforeseeable surprise out of left field. It was entirely predictable. In fact many people predicted it. I was one of them, and my position was not considered novel or surprising at the time. In 2013, as the Rationalist campaign to raise awareness of AGI risk was taking off, I asked a couple of top Rationalist leaders—including the then-head of MIRI—why they thought their awareness campaign would help, given that it could very well inspire people to accelerate the research they spoke against. 

Both were clearly familiar with the “arms race” argument already, both said that inspiring development and making things worse was an entirely plausible outcome, neither gave any sort of argument that the benefits would outweigh the risks or struck me as especially interested in the question. In hindsight, the only flaw in the “arms race” argument is that it was too grandiose. We were imagining scenarios where “AI safety” researchers deliberately sprint towards AGI because of Space Race-level competition with credible foreign projects, not because of profit-driven competition between different American companies.

Will the same “raising awareness” strategy have different results now that they’re lobbying legislators instead of tech magnates? Of course not. If they succeed in persuading the U.S. government that AI is too powerful to leave unregulated, then the U.S. will respond the way it does to every powerful new technology—by funding and organizing its research and development. If and when the U.S. government takes charge of AI research, expect Berkeley Rationalists to start working for military labs that are explicitly working towards AGI while mouthing platitudes about “safety,” just like they did with all the previous AGI labs.

As for advocating specific policies, the AI Doomer coalition does not have a developed policy platform which they agree on. Its members vehemently disagree on how to solve the problems they organize around. Some support legal requirements that AI models must meet defined “safety benchmarks”, while others denounce this as “safetywashing” which creates a false and counterproductive impression that current techniques are fine without addressing the unique dangers of future superintelligent systems. Some AI Doomers support strong export controls and other policies to ensure Western countries maintain their lead in AI technology, especially over China, while others support conciliation and the creation of an international treaty and monitoring regime. 

Mostly they are bound together by the belief that Something Must Be Done, rather than any particular idea of what to do. Historically, this frequently prompts movements to push half-baked programs which satisfy members’ anxieties but do not achieve their strategic goals, and have a host of unintended consequences. It leaves a movement vulnerable to hijacking by opportunists who do not share their goals, and in fact this has happened to AI Doomers repeatedly—not just with the AI labs siphoning off researchers and funding, but also political opportunists, who have coopted the name and credibility of “AI safety” so thoroughly that the AI Doomers refused to be associated with the term they themselves coined. 

Because the AI Doomers lack a party line or any ability to police adherence, opportunists have been very good at wielding the resources assembled by AI Doomers, who have provided much of the funding and research talent for the AI research labs that they ostensibly oppose, and laid much of the intellectual and propaganda groundwork for the political factions who they treat as sometime-foes sometime-allies. The only policy that AI Doomers mostly agree on is that AI development should be slowed down somehow, in order to “buy time.” As discussed above, the reasons for this don’t hold together, except to justify their own quest for power—and like every political coalition in the history of the human species, they certainly agree about that. But unless one believes that the U.S. government of 2035 is likely to reverse its decline and be better able to react effectively to social problems than the U.S. government of 2025, the only remaining reason why slowing down AI development would be more likely to result in friendly AGI would be if the AI Doomers themselves simply have better character which makes them better able solve the problem even apart from any specific policies they might or might not pursue.

This was once a popular belief among the Rationalists. However, it has fallen by the wayside many years ago. 

The Old Movement’s Discarded Dreams

As the name suggests, the Rationalist community was founded on the promise of “refining the art of human rationality.” Readers today might be surprised that Yudkowsky’s essays, which created the group, have artificial intelligence as only a secondary topic, and are primarily about developing mental techniques for unlocking the full potential of human thought. This was the main focus of the movement in its early years, and Yudkowsky’s writing does indeed describe some useful mental practices which a diligent and introspective reader can practice and adopt. The hope was that the movement’s research into mental practices and insight into the laws of ideal thought would make them extremely competent in all walks of life. 

This belief was explicit in Yudkowsky’s Sequences essays, and inspired his followers to establish the Center For Applied Rationality (CFAR) to develop further techniques. As long as the Rationalists still mostly believed this would work, there was a justification for believing that their own people would govern better, especially on this topic where they had most thoroughly documented the irrationalities which ostensibly made it difficult for outsiders to act rightly. To make a long story short, the Rationalist training programs failed to deliver the promised upgrades, so the movement rationally updated their views in accordance with the evidence, and this belief was gradually walked back and abandoned. 

By about 2017, the movement had thoroughly discarded its early dream of serving as a vanguard party wielding the perfected techniques of reason, and of “raising the sanity waterline” by distributing these techniques to the population at large. CFAR’s original cadre of true believers left the organization and were replaced by less ambitious staff. Today, a minority of the Rationalist old guard still hold to the belief in the power of trained rationality techniques in their personal lives, but it no longer plays a part in Rationalist plans for political control of AI or for broader social change.

The Effective Altruists used to have a similar belief that they were inherently fit to rule, albeit with a different basis. While the Rationalists once believed that they wielded, or someday would wield, superior mental practices, the Effective Altruists once believed that they were uniquely willing to work in the wider interests of humanity rather than venal self-interest. They would frequently speak of a person’s “alignment,” a term originally from artificial intelligence theory used as shorthand for benevolent and impartial concern for all humanity, but soon repurposed to mean adherence to Effective Altruist professional backscratching and doctrine. Many Effective Altruists conflated the concepts, not only the terms. The term has continued spreading, and in recent months I’ve heard “alignment” used this way by unrelated social movements and elite university groups.

The Effective Altruist self-image was abruptly shattered by the 2022 arrest of Sam Bankman-Fried, Effective Altruism’s second-largest donor, for running one of the largest financial frauds in world history. This provoked a crisis of faith among Effective Altruists, and a great deal of rethinking. Their self-image as morally superior is now much more limited and healthy, reduced to about the level of any other political faction. The assumption that any Effective Altruist is ipso facto more benevolent and trustworthy than anyone else is no longer tenable, and no longer spoken in public.

Perhaps the changes in these movements are for the better. In any case, the less-ambitious self-appraisal has led them both to quietly withdraw these justifications for believing their members would govern AI better simply because of who they were.

Finally, there is also the position that all AI development should be halted indefinitely, to resume only after the creation and deployment of extremely powerful human intelligence augmentation technology, which would create researchers far smarter than any existing human being. This is advocated by Eliezer Yudkowsky, and as far as I can tell by no one else, although even for him the plan is sensibly “not being emphasized very hard in current messaging.” He hopes that once this project has engineered “[h]umans augmented far enough … that they never put faith in a step that will not hold,” the resulting specimens “can build ASI [artificial superintelligence] without making any hidden fatal mistakes”.

Of course, there is no indication that massive intelligence augmentation will be developed any time soon, only very weak reasons to suspect that it’s obtainable at all without multiple revolutionary breakthroughs in our understanding both of genetics and of the mind, and no reason at all to imagine that augmenting human intelligence would by itself instill the psychological changes towards humility and caution that Yudkowsky desires. But leaving aside the conflation of intelligence and epistemic humility, if we are overoptimistic to the point of pollyannaism, then we can imagine that it might take 25 years to create human intelligence augmentation technology which meets Yudkowsky’s standards for the creation of uplifted superhumans, and then 25 years for the newly-created supermen to grow up and embark upon their careers in AI research, for a total of a fifty-year ban on all AI research.

Of course, 25 years to superhuman intelligence augmentation would be shockingly, shockingly fast. While there is no firm ground for any prediction as to how long it will take before any technological breakthrough, if ever, it seems more likely that such a regime would have to last worldwide for a century or several centuries before such technology were created. Yudkowsky, presumably, is well aware that maintaining a worldwide ban on AI research for that long is nearly impossible, and would be historically unprecedented in several different ways, going far beyond, for example, existing restrictions on nuclear weapons in reliability, extent, duration, and most especially enforcement.

In Search of a Strategy That Might Work

The proposal’s impracticality notwithstanding, while the philosophical arguments for the speculative danger of AGI seem more than strong enough to justify technical research into the apparent problem, they are nowhere near a level of certainty strong enough to justify hundreds of years of panopticon global surveillance, banning “[a]nything you don’t need for computer game graphics,” and deigning to allow “small amounts of high compute [only] in datacenters under international supervision.” And of course, as algorithmic progress continues on whatever hardware is available, the bar for what someone can “safely” keep on a private computer would drop and drop and drop, even without considering how the datacenter bureaucrats would inevitably push for and receive more and more and more surveillance powers, requiring the recall and seizure of once-allowed computers.

The plan is a bad one, but it does have one very important virtue. The argument for the plan is at least locally valid, if you grant all of its absurd premises. In other words: if every single one of Yudkowsky’s many suppositions about hypothetical breakthrough technologies in fields as diverse as computing, genetics, and human psychology all turn out to be completely correct, and if the ridiculous impossibility of implementing the plan is overcome, and if we set aside the horrific costs to human liberty and to technological development, then the plan does at least help somewhat with the problem it claims to address. This makes it unique among political proposals for preventing the development of unaligned AGI.

The usual AI Doomer plan does not even make it this far. Even if all of their technological assumptions are correct, the plan of establishing progressively more onerous regulations so that bureaucrats can hobble AI research in hopes that this would “buy time” would not result in a better chance of creating an aligned AGI. Instead it would just shift the same AI research from privately-owned companies to the U.S. government, most likely via the military, most likely employing the same individual researchers as today.

The AI Doomers’ plans are based on an urgency which is widely assumed but never justified. For many of them, the urgency leads to a rush to do something—anything—even if their strategy is unsound and has historically done exactly the opposite of what they profess to want. In the absence of a specific policy proposal which is likely to work, those who accept the premises of AI Doom would do better to work on improving institutional functionality across the board, rather than attempting to “buy time” in the hopes that our declining institutions will unaccountably become more able to handle their concerns, rather than less able. Otherwise, their current program is on track to once again accelerate the research they hope to prevent.

Claims of resource depletion and collapse have accompanied industrial society on every step from steamboats to container ships, from hot air balloons to Mars rovers. Of the many such arguments that have been put forward in our long history, the best was made in 1865, when William Stanley Jevons published The Coal Question. This book argued that industrial society was eventually doomed because of the limited supply of coal for power. Because consumption grows exponentially and supply is finite, we will eventually run out of coal and be unable to power our machines.

It’s true that the supply of coal is finite. What Jevons missed, understandably in his own time, is that coal is not the only possible source of industrial power. When he published, effectively all industrial power came from coal. In 2023, this was down to 26%. John D. Rockefeller founded Standard Oil, the first oil-producing megacorporation, five years after Jevons finished his book. A century after Jevons’s argument, in the 1960s, oil finally surpassed coal as the world’s top power source. Today coal, oil, and methane gas each provide about a quarter of the world’s power, with coal providing a bit more power than methane gas and a bit less than oil.

Peak Energy is Far in the Future

After World War II, as oil power approached and overtook coal power, the world saw similar predictions for the imminent arrival of “peak oil,” the forecasted point when oil production reaches its maximum and begins to decline. These prognosticators were more reckless and aggressive than Malthus or Jevons, in part because their motives were more ideological, and predicted that peak oil would arrive very soon, in some cases as early as the 1960s. This was bound up with the anti-industrial turn of the environmentalist movement, and rising activism for the dismantling of industrial society, which has since been euphemized as “degrowth.” Among the profusion of warnings from academics and activists, the most influential was 1972’s The Limits To Growth, which argued that exponential growth in resource demands would exceed the world’s possible supply while exponentially increasing pollution would destroy the biosphere, together causing “a rather sudden and uncontrollable decline in both population and industrial capacity.”

Since then, oil and coal production have both increased over time, sometimes with decade-long plateaus or minor dips, in defiance of these “peak” forecasts. There are two main reasons that the 20th century models and computer simulations underestimated production, even when they were done in good faith. The first is the ongoing discovery of new deposits. The Earth is incredibly, incredibly huge, and we are nowhere close to finding all of the stuff we can use. In May 2024, a Russian survey reported the discovery of Antarctic oil fields with estimated reserves of over 500 billion barrels. If these early estimates survive deeper investigation, it would be by far the largest oil discovery ever, three to five times bigger than Saudi Arabia’s world-record Ghawar field. “There are more things in … earth, Horatio, than are dreamt of in your philosophy.”

The other factor is that new technologies periodically turn useless rocks into economically valuable resources. The development of hydraulic fracturing—fracking—made it possible to extract oil from previously worthless shale, turning the U.S. from a net importer of oil to a net exporter and undercutting the economic base of rival exporters like Russia, Venezuela, and Iran. As of 2022, fracking accounted for 66% of U.S. oil production and 80% of U.S. methane gas. Earlier improvements like the invention of offshore drilling platforms had similar effects, and future technologies will no doubt continue the trend. From the 1865 vantage point of William Stanley Jevons, the 19th century technology to refine and exploit oil, or the 20th century technology to cryogenically liquify methane gas for transport and storage, also counts as turning useless rocks into new energy. It remains to be seen what advances will come in the 21st century, or the 22nd, or the 23rd.

Of course, none of this changes the fundamental logic of Jevons’s argument, it only pushes the timelines around. However gigantic the deposits of fossil fuels may be, they are still finite. If future technologies increase accessible reserves by a factor of ten, or a thousand, or a million, the underlying point remains the same: if exponential growth continues, it will use these deposits, not necessarily on the timescale of an individual person’s life, but certainly on the timescale of civilizations.

If fossil fuels were the only source of energy, this would be a civilization-shaking concern. Fortunately, they are not. In principle fission power, fusion power, geothermal power, and solar power are all capable, individually, of powering an industrial civilization far larger than our own for far longer than recorded history has lasted—billions of years, in the case of solar power. Decade by decade, the share of power from non-fossil power sources is growing at a rate roughly comparable to oil’s slow growth to eventually overtake coal. Of these sources, only fission power is yet developed enough that it could actually bear the weight of industrial civilization with our present technology. That alone is sufficient to knock down Jevons’s argument.

Geothermal or solar power might plausibly reach this point within our lifetimes, depending on advances in drilling and battery technology, respectively. Such swift success is far from certain, but if exponential growth continues and there is no interruption from civilizational collapse, then it seems very likely within a century or two, at most. Fusion power is more speculative and I’m hesitant to rely on breakthroughs in a technology where success has supposedly been right around the corner since before I was born, but who knows. We also shouldn’t discount the possibility that entirely new forms of power will someday be invented using scientific breakthroughs inconceivable to us today—after all, Jevons published not just before the creation of nuclear power plants, but before the discovery of radioactivity. None of these hypothetical advances are strictly necessary, however, since fission power with present technology is more than enough to fuel industrial civilization’s growth far past the point that anyone can forecast in any but the broadest strokes.

We Won’t Run Out of Food or Metals

More rarely, we see similar depletionist arguments for limits on durable metals like iron or aluminum. These are somewhat less dire, because if we did run out, then the possibility of recycling means this would merely prevent civilization from growing further rather than necessitate a collapse. The Limits To Growth used exponential models of resource consumption to illustrate scenarios where aluminum supplies ran out in 2003 and copper supplies ran out in 1993. Obviously this did not happen. Production of both metals has continued to increase.

Prospectors have not searched anywhere close to the entire globe and estimates of the world’s total reserves are revised upwards decade by decade. Prospecting for ores like bauxite or chalcocite which make people’s eyes glaze over is far less thorough even than prospecting for oil deposits which make corporations salivate and politicians sweat. Technological improvements regularly make more deposits accessible, whether directly by improving the economics of digging, or indirectly by allowing ores with lower and lower concentrations of metals like copper to be processed. This is not a new phenomenon. As T. A. Rickard recounts in the first volume of his 1932 magnum opus Man and Metals, p. 447, in the early 1900s slag heaps from ancient Roman lead mines were reprocessed “on a large scale” to extract metal the original miners had been unable to separate from the rock.

In principle, the limits are set by our ability to dig. The deepest mine ever dug is a gold mine 4 kilometers deep. If we could exploit all the matter within 4 kilometers of the surface—we can’t, yet—that would only amount to about one five-hundredth of the Earth’s volume. If mining technology continues to improve over the next centuries at about the same rate as the previous centuries, much much more of the Earth’s matter will come within our reach. Mining technology has long been a great stimulant to technology and growth. The world’s first useful steam engines were developed to pump water out of 0.05-kilometer-deep mines for coal and copper, making them possible to operate without flooding.

Improvements in materials technology also allow greater and greater substitutability of different resources. Within Elon Musk’s industrial empire alone, there are major efforts underway to use steel for rockets that once used only titanium, and aluminum for cars that once used only steel. Plastics and ceramics now serve many purposes which could once be fulfilled only by metals. Progress on this has been steady for centuries, and we are nowhere close to the limits imposed by the laws of physics and the structure of atoms.

The last major limit to growth is food production. Famously, this was the subject of Malthus’s warning in 1798. In 1968, Paul Ehrlich’s The Population Bomb predicted overpopulation would outpace food production and cause hundreds of millions of deaths from famine as early as the 1970s. This did not happen, although the fear it engendered prompted many governments and many individual families to reduce the number of children born. With decline in global fertility, this has become less of a concern among intellectual elites, with some early voices even worrying about population decline. The world’s population has fallen and risen many times in history. On the scale of centuries it will presumably fall again, and rise again, and so on, and the global population’s long-term upward trend will continue. Food will remain a necessity regardless.

The Limits To Growth makes much of the finite quantity of arable land as the reason that Malthusian mass starvation is inevitable. Technically this “limit” has been broken many times, from Bronze Age irrigation channels on the Nile River, to Iron Age dikes on the Yellow River, to medieval land reclamation works in the Netherlands, to today’s desalination plants in Israel. However, on a global scale, increases in arable acreage from projects like these have been relatively small. Much, much more significant have been increases in yield per acre, especially since the “Agricultural Revolution” began in 17th century Britain.

This exponential trend has continued steadily for centuries, thanks to improvements as diverse as the tractor, the Haber-Bosch process for mass-producing nitrogen fertilizer, the breeding of superior crops and livestock, pesticides, advances in soil science, and countless others. The improvements in efficiency show no signs of stopping. Since The Limits To Growth was published in 1972, the average yield per acre of most staple crops has roughly doubled, and the world’s population has increased by about the same fraction. As a result, the amount of land needed to feed the world’s growing population has stayed nearly the same. There is no particular reason to expect this trend will abruptly stop any time in the next few centuries. In 2200, the total amount of land needed to feed the world’s presumably higher population is as likely to be smaller than today as it is greater.

Reaching Towards the Limits of the Solar System

There must exist some maximum point of development that the Earth can sustain. We can imagine a civilization so advanced that they can mine every atom of uranium down to the Earth’s core, and capture every photon that strikes the planet, and convert it all to useful power with the unattainably perfect efficiency of a Carnot engine. This would be a civilization so inconceivably vast that, in comparison, we might as well be a pack of chimpanzees using grass to fish for termites. But logically, they would still be finite.

Equally logically, a civilization anywhere near that advanced would not be limited to the Earth. With a fraction of that knowledge and matter and energy, they would have no trouble settling on other planets, or mining asteroids, or building orbital solar-power arrays. Even our own civilization can already make fumbling, uneconomical attempts at these projects for scientific purposes, and the usual course of technological growth is that fumbling, uneconomical prototypes eventually become load-bearing industrial foundations, even if it sometimes takes centuries of development. The basic proofs of concept were mostly deployed in the 20th century, and have slowly improved since then. The exact course cannot be plotted, but the general direction seems clear.

When we look at spaceflight, we are in roughly the same position as Ben Franklin when he observed the early tests of hot air balloons in the 1780s, and understood vaguely but accurately that developments in flight would revolutionize warfare and “give a new turn to human affairs.” Eventually, perhaps the speculative spaceborne megaprojects like O’Neill cylinders will prove practical, or perhaps it will take a very different form. The Earth accounts for only about 0.2% of the matter orbiting the Sun and receives less than a billionth of the Sun’s energy. A sufficiently advanced civilization would end up in space somehow, because that’s where all the good stuff is.

The dream, of course, is to settle not just our own solar system, but to reach other stars. This seems much more doubtful. Sending ships to other stars is possible, in the sense that it is permitted by the laws of physics. Whether such a complex machine and its payload can be maintained on a journey of lightyears, with nothing except the mass and energy it brings along, is a tougher question. Unlike interplanetary travel, where our probes have already begun blazing the trail and men have already walked on the Moon, interstellar travel is nowhere close to anything that humans have ever done. Science fiction authors have dreamed up possibilities, but nothing that even vaguely resembles their precursors has ever been built. Perhaps future technology will solve these problems, presumably with methods that no one has yet thought of. Perhaps the challenges will prove impractical even to an ultra-advanced interplanetary civilization, as their prototypes encounter new difficulties unknown to us. If we someday figure it out and cross the chasms between stars, then we may eventually be able to cross the chasms between galaxies—a million times further—or we may not. It is far too early to say anything with confidence about such speculative technology.

Even if we can be confident that an interplanetary civilization, at least, is permitted by the resources available to us, none of this means that humanity inevitably will reach these heights. If the limits of Nature do not stop us, we might do it ourselves. Improvements in economic technology go hand-in-hand with improvements in weapons technology. Current nuclear arsenals are probably not enough to destroy industrial civilization, especially since they have shrunk fivefold with the end of the Cold War, and it is even less likely that they could end humanity itself. Fears of nuclear winter are generally overblown. Still, I would rather not try the experiment.

More recently, fears have risen of more exotic technologies like bioweapons or recursively self-improving artificial intelligence. Personally I’m more worried about the presently-inconceivable superweapons that will be developed after another century or two of technological advancement, far beyond our own imagination as the nuclear bomb was beyond Jevons. Mankind has worried about our capacity to destroy ourselves ever since the widespread use of the machine gun in World War I, and regardless of the specifics of any particular technology, we remain no closer to finding a solution while our power continues to grow.

Separately, it is entirely possible that humanity will simply choose not to leave the nest, at least for a time. Civilization after civilization has achieved several centuries of growth and scientific advance, and then petered out. At its height, classical Greco-Roman civilization made advanced prototypes like the Antikythera mechanism or the aeolipile, but by the time barbarians were pouring across the borders and carving out their kingdoms, these had vanished entirely—and more prosaically, yet more importantly, large-scale mining and water-powered mills vanished from Europe for centuries.

The stunning technological and industrial growth of China’s Tang and Song dynasties was eventually succeeded by the legendary stagnation of the Ming and Qing dynasties, famously illustrated by the decommissioning of Zheng He’s fleet in the 1430s. Our own civilization’s rate of advance remains substantial, but it has slowed down somewhat since its most frenetic pace in the days of Henry Bessemer, Thomas Edison, Henry Ford, Fritz Haber, and the Wright Brothers. Sooner or later, we will go the way of the Romans, whether that takes seventy years or two hundred and fifty. When it happens, it could delay mankind’s advance by a few centuries, unless and until a new civilization arises and picks up the torch.

Mankind’s trajectory depends on us, here, today. Every generation must build the society they want out of the parts left to them by their forebears. If we dismantle our engine of progress out of innumerate fears of resource shortages, or destroy it as we fight out internal resentments, or overburden it with a host of corrupt idlers and apparatchiks, or let it lapse out of mere sloth, then the advances of recent centuries will slow to a crawl. If we build a robust and dynamic society, if we repair and rejuvenate our institutions, then we will take the next step on the journey to set our grandchildren walking on other worlds.

That year, neuroscientist Matthew Schrag discovered doctored images in this and many of Lesné’s other papers, including others purporting to provide evidence for the amyloid hypothesis. These images had been manually edited and cropped together to falsely show support for the papers’ hypotheses. Notably, these frauds all made it through the formalized “peer review” processes of Nature and six other academic journals undetected, before eventually being uncovered by unrelated channels.

Schrag’s investigation that uncovered the fraudulent papers began as a tangent from his work uncovering doctored images used in studies supporting simufilam, an experimental drug for Alzheimer’s disease. The suspicion would prove vindicated when in June 2024 Hoau-Yan Wang, a paid adviser to simufilam’s developer, was indicted by a federal grand jury for fabricating data and images in simufilam studies for which he obtained $16 million in National Institutes of Health (NIH) grants, following a 2021 petition to the Food and Drug Administration, a method of reporting research fraud which is highly unusual if not unique.

Follow-up to evidence of Lesné’s fraud was slow. Schrag’s discovery kicked off two years of wrangling, eventually leading all of Lesné’s coauthors—but not Lesné himself—to agree to retract the 2006 Nature paper. As Science reported in 2022, “The Nature paper has been cited in about 2300 scholarly articles—more than all but four other Alzheimer’s basic research reports published since 2006, according to the Web of Science database. Since then, annual NIH support for studies labeled ‘amyloid, oligomer, and Alzheimer’s’ has risen from near zero to $287 million in 2021. Lesné and [his coauthor] Ashe helped spark that explosion, experts say.”

Scientists must now untangle the strands of fraud woven through decades of arguments stretching across a billion dollars worth of research. The paper’s contribution to the allocation of this billion dollars might also be a reason why such a widely-cited paper, presumably read by thousands of experts where some must have spotted the fraud, wasn’t reported earlier. Whether the amyloid hypothesis survives or not, this fraud has likely delayed the arrival of life-saving medication for tens of millions of people, perhaps by many years. If so, it is a humanitarian disaster larger than most wars.

No Consequences for Fraud 

All the coauthors of the Nature study who have spoken publicly deny any knowledge or involvement in the fraud, but acknowledge that the images were clearly tampered with. So far, the perpetrator has been left safely in the passive voice. Who actually manipulated the images? Was it Lesné himself? Was it a coauthor acting with his knowledge? Was it a harried underling, acting alone and in secret to fulfill their supervisor’s harsh expectations? Circumstantial evidence suggests it was probably Lesné, but there is no hard proof yet.

In any case, Lesné’s employer, the University of Minnesota, is not interested in these questions. Lesné apparently remains a professor and continues to receive National Institutes of Health funding. The university has been investigating his work since June 2022. A spokesperson says the university recently told Nature it had reviewed two images in question, and “has closed this review with no findings of research misconduct pertaining to these figures.” There is no ongoing effort from the university to find the culprit.

Lesné’s apparent fraud is not an isolated incident. In 2023, Marc Tessier-Lavigne, then the President of Stanford University, was forced to resign after the revelation of falsified data in his earlier research at drug developer Genentech, including a now-retracted paper on the amyloid hypothesis which has been cited over 1,000 times. In a case of successful dynastic skill transfer, public reporting was spearheaded by Theo Baker, scion of an influential family of journalists, during his freshman year as a Stanford student. In 2024, Tessier-Lavigne bounced back, becoming the CEO of the newly-founded drug discovery company Xaira Therapeutics, with over $1 billion committed by its venture capitalist backers.

Once again, Stanford’s official statements do not finger the culprit, but blame Tessier-Lavigne’s “management and oversight of his scientific laboratories,” where “Multiple members of Dr. Tessier-Lavigne’s labs over the years appear to have manipulated research data,” according to the Stanford Board of Trustees’s official report. The report contains incredible positions like “the process through which the science of the paper was developed in Dr. Tessier-Lavigne’s lab, culminating in its publication in February 2009, lacked the rigor expected for a paper of such potential consequence, although the Panel did not find, based on the evidence available to it, that Dr. Tessier-Lavigne was aware of this lack of rigor.” The panel maintains that “it would not be reasonable to expect Dr. Tessier-Lavigne to have identified these instances of research data manipulation prior to or at the time of the respective papers’ publications,” even though many of the fabrications were later identified by visual inspection of the manipulated images. Stanford’s public position is that its researchers cannot reasonably be expected to know what happens in their own labs and gets published in their own names.

Once again, there is no effort to find the supposed true culprits. If the trustees’ report is taken at face value, then there must be several unidentified fraudsters in multiple laboratories, but they seem unconcerned with the obvious implication that these supposed fraudsters are continuing to make up fake data for critical medical research. The trustees maintain that the fraud arose from “lab culture and management,” and therefore implicitly that there is no single individual who bears moral or procedural responsibility for actually falsifying the data. As we saw with the University of Minnesota, this attitude is typical of academic overseers.

While Tessier-Lavigne was at least forced out, rather than enjoying total impunity like Lesné, this should not be mistaken for justice. Medical fraud at this scale is mass manslaughter. If you ripped medicine out of the hands of every man, woman, and child in Spain, that would be about as many people as suffer from Alzheimer’s disease worldwide, whose treatments—and perhaps someday a cure—were all delayed for the benefit of one man’s career. In a just world, the perpetrators would not simply bounce from one prestigious executive position to another. In a just world, the trials of Sylvain Lesné and Marc Tessier-Lavigne would be at least as widely publicized as that of Sam Bankman-Fried. If found guilty, they would be imprisoned for decades.

The Replication Crisis Replicates Across Fields 

Medical fraud may be the most evil form of academic fraud, but it is not the only one, or even the most common. More famous is the “replication crisis” in academic psychology, where since around 2010 it has become widely known that most studies in psychology fail to replicate—which is a polite way of saying that the research is mostly fraud, statistical massaging such as “p-hacking,” or simple bullshit. This has had surprisingly little effect, and the field has proceeded with barely even any cosmetic changes, even though all informed observers now know that it’s mostly fake. 

Intellectual discourse long assumed that psychology was especially bad compared to other fields, whether due to difficult-to-study subject matter in the “soft sciences,” political pressures, or whatever other cause. Yet even while the ivory tower grappled with the problem, academic psychology remained politically influential. Works like Nudge, the influential 2008 book of psychology and behavioral economics, provided tremendous support for the role of “experts” managing public opinion during the COVID-19 pandemic. In recent years, it has become clear that psychology was relatively good. The indifference to truth is endemic to all or most fields, and psychologists like the Nobel-winning Daniel Kahneman, the author of Thinking, Fast and Slow, a book which popularized many of these results, were able to publicly acknowledge and discuss these problems more than a decade ahead of their peers. This came years after Kahneman’s work, based in large part on studies which failed to replicate, had played a substantial role in inspiring the influential “Rationalist” movement.

Perhaps the most egregious example in psychology is superstar professor and bestselling author Dan Ariely, likely the most famous psychologist alive. He has had several issues with papers failing to replicate or clashes with research ethics bureaucracies, but the most blatant incident came to light in 2021. The blog Data Colada, which focuses on analysis of research misconduct in the social sciences, published an analysis by “a few anonymous (and fed up and frightened) whistleblowers” which showed that the data in a 2012 paper on dishonesty was clumsily fabricated to provide a positive result. All five coauthors agreed that the data is obviously fabricated and that Ariely, one of the paper’s coauthors, had been the one to pass on the fabricated data to his colleagues. 

Ariely implausibly claims the data was already fabricated when he received it from the insurance company the authors partnered with, and the insurance company claims the manipulation occurred after they sent it to Ariely. This “dog ate my homework”-tier excuse was apparently enough for Ariely’s employer, Duke University, to sweep it under the rug. In 2024, Ariely said that a Duke University investigation determined he had not knowingly manipulated the data, although Ariely himself remains the only source for this claim and Duke told reporters “we are not in a position to confirm or fact-check anything on this.” While most reporting has focused on Ariely as the near-certain perpetrator of the fraud, it is also significant that of the paper’s four other coauthors, at best none of them bothered to look at their own data closely enough to spot that it was impossible nonsense. One wonders what they do with their time instead. Many of Ariely’s other papers do not pass the smell test, although no one has yet looked closely enough—and the underlying data may or may not be available—to identify the specific fraud.

This is not just a matter of picking up a few outliers that will inevitably occur in any massive endeavor. This is a systematic corruption of the culture of academia’s researchers, with the active complicity and encouragement of the administration. In the time it took to write this article, I passively stumbled across more examples of consequential research fraud by high-profile field-leading academics than I have space to describe here. If you want more examples, you can find them all day long. How many similar frauds remain undiscovered? One study found that, between 2000 and 2021, the fraction of European biomedical papers which were later retracted quadrupled.

And why not? If bureaucratized peer reviewers or even a paper’s own coauthors aren’t expected to pay enough attention to notice blatant fraud, if fraud is only revealed by third-party investigators when they choose to make a years-long personal crusade in the face of institutional headwinds, if those frauds which are uncovered don’t come to light until decades after the fact, then we can be confident that almost all of the frauds have gotten away with it. Why wouldn’t you report the splashy result that you promised the grantmakers? Why wouldn’t you accept the prestige and promotions and money? Would you be deterred by the fear that, in twenty years, there will maybe be some bad articles online but you keep your job while your colleagues rally around you?

Even in the case of Tessier-Lavigne being forced to leave Stanford to lead a billion-dollar startup, it seems that he is the exception that proves the rule, facing any consequences at all not because his fraud was especially bad, but because Theo Baker had the family connections to take a prominent scalp—for which he should be applauded, in this case. Hoau-Yan Wang may end up paying a steeper price, if he is convicted of fabricating simufilam research, but it is revealing that this is only possible because his accusers gave up on pursuing truth and justice through the system of scientific journals. 

Instead of calling for a retraction, they appealed directly to federal regulators at the Food and Drug Administration, prompting an investigation by the Securities and Exchange Commission (SEC). In all the clearly-identified frauds I’ve looked at, I have never found one who was seriously punished by their scientific peers or their university employer, unless the university first came under extraordinary pressure from other leading power centers like journalists or federal prosecutors.

The Epistemic Bankruptcy of Academia

Academics should perhaps take a lesson from one of America’s other privileged guilds. Financiers have a sophisticated apparatus for catching those who break its rules and codes, however lax those codes may sometimes be. Federal bureaucracies like the SEC or the Treasury are staffed largely by former and future financiers and bankers, who move freely through the “revolving door” between private companies and government bureaucracies. As a result, they share much the same norms, the same sense of what’s fair, and the same sense of what behavior is illegitimate. In cases of outright fraud, the fraudster’s former peers in the federal agencies will come down like a ton of bricks, and their employer will swiftly cut them loose. 

The SEC maintains a bounty program for whistleblowers, who are awarded between 10% and 30% of the fines the commission imposes as a result of their information, and regularly announces bounties in the tens of millions. Those caught are often sent to prison. Sam Bankman-Fried’s sentence of 25 years may arguably be too lenient for billions of dollars of theft and fraud, but that he is in jail at all gives the financiers a much firmer moral grounding than the Alzheimer’s disease researchers. This approach, of a professional society granted extraordinary privileges while also required to submit to Congressional oversight and to assist in self-policing, provides a much more functional model for oligarchic guilds which have been delegated core social functions.

The U.S. financial system is hardly the greatest edifice of justice in the world. Yet, it demonstrates a basic level of self-policing, effort to uphold professional standards, and accountability to the rest of society. Academic institutions fall far short of these minimal standards. Known serial frauds are sheltered by their bosses and feted by their peers. The culture encourages this at every step of the way, starting with PhD candidates ordered to produce a positive result by any means necessary, continuing with coauthors and grantmakers who can’t be bothered to look at the data and check whether it makes any sense, all the way to department heads and famous bestsellers being widely cited even after they’ve been caught. Those who do not commit fraud themselves usually tolerate it in their peers. The minority who will not tolerate frauds usually weed themselves out quietly. I have lost count of how many friends of friends entered a PhD program, had an adviser who tacitly or explicitly demanded they commit fraud to get publishable results, and quit in disgust without raising a public stink. What that says about those who remain is not encouraging.

Fraud is certainly not the only problem in academic science. Lesser crimes such as massaging data to create false positives and “HARKing” (hypothesizing after the results are known) are even more common. Plagiarism has increasingly become a scandal, most famously forcing the resignation of Claudine Gay from the presidency of Harvard University, the highest post in America’s entire academic system. And many fields are overrun by the publication of bullshit and word games, meaningless enough that classifying it as “true” or “false” is not even useful. Fraudulent data is the worst of these sins, responsible for the biggest pile of dead bodies, since the damage it does is hardest for other scholars to undo. Yet these are all symptoms of the same disease.

With some honorable exceptions, most academics don’t care very much about the capital-T Truth. Oh, they’d prefer Truth to lies, if Truth only cost two dollars. But if the cost is at all serious, they won’t pay it. It is a fault in the institutions that academics must choose between wealth and prestige and career success on the one hand, and pursuit of Truth on the other. But it is a fault in the professor’s character that he would turn his back on Truth, for this. Our civilization created the universities and carved out their special privileges because we recognize that the unencumbered pursuit of Truth is among the noblest callings, and also among the most useful. These people have been given a sacred trust, and they have abandoned it in order to secure a comfortable position. You cannot advance human knowledge by throwing venal careerists and apparatchiks at the problem, no matter how many tons of paper you print. Knowledge comes, and always has come, from insane people on a spiritual quest for Truth, who will neglect worldly concerns in pursuit of the insatiable drive to know. Today’s academics are living in the ruins of a temple built by better men.

Hope Rests with Truthseekers Outside Academia

I have little hope that academic science can be reformed from the inside. Frankly, it seems too far gone. If they want to prove me wrong, it will take a massive purge of the fraudsters and bullshitters, and probably a sharp reduction in the institutions’ total staff. There is some extremely rare precedent for imprisoning medical researchers for defrauding NIH grantmakers by fabricating data: in 2006 Eric Poehlman was sentenced to one year, in 2015 Dong-Pyou Han was sentenced to four years and nine months, and Hoau-Yan Wang might soon join their number. 

One conviction per decade is not nearly enough to solve the problem, but it shows that the legal tools already exist. Perhaps the doctors can take a page from the financiers, and create an NIH department dedicated solely to catching and imprisoning fraudsters. Perhaps they can offer a bounty of 10% of the fraudster’s NIH grant for those who turn them in. Whatever the details, any hope of solving the problem from the inside will require imprisoning hundreds of perpetrators, and permanently barring thousands from all scientific institutions. I might also suggest a public cashiering, in which the university’s leaders ceremonially strip the offender of their academic gown and hood, and tear up their diploma.

Unfortunately I doubt that our scientific institutions have the stomach for interventions like these, or for anything else which could possibly work. These are ideas for what actions might actually fix the problem, not ideas for what actions might be within bounds for the institutions as they exist today. More likely, fraud will grow more and more common as young scientists realize that lies are the best way to advance their careers and that serious punishment is about as likely as being struck by lightning. A fringe of credentialed scientists with unusual integrity and courage will continue their efforts on their private blogs and newsletters to document and publicize the many, many known frauds. Universities will not support them, and the legal and social pressure to shut down these embarrassing windows into the sausage factory will slowly grow more intense.

More likely, reform will come through circumvention from outside the academic system. There is no shortage of people who pursue the sacred quest for Truth. Increasingly, they are not pushing forward the frontiers of knowledge in peer-reviewed journals and university campuses, but in fringe niches of internet discourse. Because the internet commentariat’s intellectual elite is more attentive to an argument’s substance than whether it observes the bureaucratic forms, these circles are much less vulnerable to the problems which afflict academia.

The most influential philosopher of recent times is Eliezer Yudkowsky, an uncredentialed blogger and fanfiction author who lacks even a high school diploma. The adoption of his thinking contributed to the formation of major AI labs such as OpenAI and later Anthropic. These same labs have made some of the biggest technological advances of recent years, primarily within industry and with only light involvement from the academic system, while poaching talent from it. A growing fraction of the intellectual elite are leaving behind the institutions for internet self-publishing. Even archaeology is seeing some efforts to work outside of the academy, with well-funded projects like the “Vesuvius Challenge” to decipher the Herculaneum Papyri. Unless the academy can somehow reverse course, this trend will spread to more and more fields.

This situation has a precedent. From the 1500s through the early 1700s, most of the world’s best scientific work was done by individual natural philosophers in their private capacity. They were in dialogue not through formal academic institutions but through loose informal networks of peers, like the “Invisible College” and the “Republic of Letters.” It was in these distributed conversations that modern science was developed and codified, and made its first major discoveries, establishing the intellectual authority of its practitioners. Over time these networks were gradually folded into the university system and to institutions like Britain’s Royal Society, which enshrined their intellectual legitimacy and preserved their epistemic virtues for centuries.

With a bit of luck and a great deal of effort, we might follow a similar path. The fringe internet intellectuals are not yet ready to bear the full weight borne by the academic system of old. As the academic system abdicates more and more of its responsibilities, the work will get easier, but there is a tremendous amount of work to do. Our output is too scattershot, and not nearly comprehensive enough. We mostly have not built the channels which can make our work legible to other elites or to the public. Perhaps most importantly, most of us don’t intend for our work to become widely accepted and contribute to the onward march of human knowledge, rather than just to know for ourselves. Our only advantage is that we hold the epistemic high ground. But in the long run, there is nothing that matters more.

This isn’t just a problem for individual researchers—entire societies can forget history. The march of entropy destroys witnesses, artifacts, scholarly records, and all other evidence of the past. Entire civilizations are lost. For thousands of years, scholars assumed the Hittites mentioned in the Old Testament were a minor power at best, until nineteenth-century archaeology revealed them as one of the great Bronze Age empires, who were nevertheless forgotten by Late Antiquity. Sometimes knowledge is rescued from oblivion by lucky finds, but usually it is not. We’re far from finding everything worth finding. We never will. Evidence doesn’t wait for us patiently—it is slowly destroyed by human activity and the forces of nature.

Many excellent historians, however, hold the opposite view. It’s very common to see historians implicitly or explicitly assert that knowledge in their field increases over time. For example, in his 1962 masterpiece Medieval Technology and Social Change, Lynn White Jr. assumes greater clarity from archaeological discoveries are yet to come: “Despite prodigious labours by Hungarian archaeologists, the stratification of Avar materials is not yet clear…[Avars] may well have been the first people of Europe to use the stirrup, but the time of its arrival is still uncertain.” Meanwhile, in a more recent article, nonprofit founder Jason Crawford writes, “I note at the outset that this is an old book, published 1925 and revised 1940. Probably a lot has been learned in the last 80 years and the following has already undergone revision, which I’ll uncover when I read more modern sources.”

The historian’s optimism rests on three promises. The first, expressed by White above, is that there are lost artifacts that can be recovered. Secret government records can be declassified, new construction will dig up an ancient tomb, a statesman’s grandchildren will find old letters in the attic and give them to a university, or archaeologists will find the ruins of an ancient temple complex. Such finds improve our understanding of the past, sometimes dramatically.

The second reason for optimism is that historians make better analyses of existing data as time goes on, as Crawford mentions above. After they make inferences from the available material, subsequent historians can take their best arguments and build on them while discarding flawed ideas which do not stand up to scrutiny. By standing on the shoulders of giants, the field will climb higher and higher, like in hard sciences such as physics or biology.

The third reason for optimism is the continued unfolding of history. After all, it is harder to see how an event fits into ongoing trends before those trends have had a chance to play out—time gives us perspective, and hindsight is 20/20. However, while the passage of time may give us a better understanding of a historical event’s effects on the future, it does not improve our knowledge of the event itself. Despite this limitation, knowing what happens next can make it easier to understand which events were important and why.

Archaeologists and Historians Can’t Defeat Entropy

If these three promises are met, then our knowledge of history is steadily increasing. How, then, could past events be so hazy today? Shouldn’t centuries of new finds, ongoing analysis, and knowledge of subsequent history mean that scholars of Henry VIII’s reign know what happened during that period far, far better than scholars of more recent events like the 2008 financial crash or the two world wars? Of course, we usually see the opposite.

These optimistic historians are writing epistemic checks that cannot be cashed. What the three promises leave out is that information is often lost. Firsthand witnesses and expert historians die after passing down only a fraction of their knowledge. If you investigate the 2008 financial crash today, chances are you can still interview someone who worked in finance or government who will give you information that has never been recorded. The information stored in people’s minds is still fundamentally accessible—for now. In a century, much of this information will be irretrievably lost. 

In addition to people, books and artifacts are also lost to entropy in a hundred different ways. The cumulative effect of this destruction is immense, as illustrated by the records of classical civilization. “[T]oday we possess written fragments from only 13% of the ~2,000 ancient Greek authors known to us by name. This does not account for the authors we do not know, and only a small portion of the 13% figure consists of complete works.”

Preserving the ever-growing mass of historical material is too expensive to be practical, so when budgets run thin, even major libraries and archives will discard books and records by the hundreds of thousands. For example, the Manchester Central Library’s recent culling destroyed 210,000 to 500,000 “literary, commercial, educational and political records going back 150 years” with “no subject specialists involved in the process.” This is a standard library practice.

Artifacts are also lost in accidents like the 2018 fire that destroyed 92.5 percent of the 20 million items stored in the National Museum of Brazil, including the only recordings of now-extinct languages. Another example is the 1986 Los Angeles Central Library fire that destroyed 20 percent of the collection and damaged much of the remainder.

In recent decades, digital information has fared no better than paper. Between link rot and changes in software standards, tremendous amounts of digital information become inaccessible over the course of a single decade. The long-term preservation of digital archives remains a hope rather than a guaranteed fact. Even in optimistic scenarios, it would require ongoing effort and maintenance on par with the curation of printed information. As the development of the printing press illustrates, much better ways of recording information can often have only modest effects on how much information gets preserved centuries later.

Preservation is not only a struggle against entropy. Political actors will often deliberately destroy knowledge of their foes’ history and culture, such as the shelling and destruction of the National and University Library of Bosnia and Herzegovina during the siege of Sarajevo, the Chinese Emperor’s “burning of books and burying of scholars” in 212 BC, or the Taliban’s 2001 demolition of the ancient Buddhas of Bamiyan monuments.

Most worrisome of all, knowledge and artifacts are often quietly forgotten. Naturally this abandonment usually leaves no trace, but one dramatic demonstration was uncovered a century ago. Around 500 BC, the Neo-Babylonian princess Ennigaldi-Nanna’ created a museum of then-ancient artifacts dating back as far as 2000 BC. When the city of Ur was abandoned, the museum and its contents were also forsaken. It was not until 1925 that modern scholars rediscovered the museum’s remains. Archeologists shouldn’t just delight in such finds, but reflect on the temporary nature of their own work.

Most losses, of course, are never recovered. Buildings and cities are razed, rebuilt, and abandoned. Rome wasn’t destroyed just by invading barbarians such as the Visigoths in 410 AD, but by the Romans themselves. Emperors would melt down bronze statues and strip buildings of adornments to pay soldiers’ wages. The later Romans built churches by quarrying the material from abandoned pagan temples, amphitheaters, and civic buildings of their fathers.

Indeed, archaeologists’ “new” discoveries are more properly re-discoveries. Lost letters were once just letters, and lost cities were once just cities. Perhaps historians of the year 3500 will excavate the sunken ruins of New Orleans, and this will surely teach them a great deal about the city—but far less than can be learned by any Mardi Gras tourist today. 

The Limits Of Scholarly Progress

It is not just the primary sources that fade over time, but also historical analyses. A glance at the bibliography of any history book will show that scholars overwhelmingly prefer to rely on sources published within their own adult lifetime—most analyses will never be read or cited after the deaths of those who read its initial publication. An especially influential work might be cited by one or two further generations of scholars before fading from the discourse. 

For example, consider Edward Gibbon’s The History of the Decline and Fall of the Roman Empire, published in 1776. It is probably the most influential work of history written in English so far, and throughout most of the nineteenth century it was part of the intellectual bedrock of historical analysis. Any educated historian was expected to be familiar with its arguments about how civilizations fall. Today, Gibbon’s masterpiece is of only niche interest. A scholar of Roman history will certainly have heard of the work, and a large minority have even read parts of it, but its arguments have mostly passed out of the historical discourse. By the year 2200, Gibbon’s footprint will be smaller still.

Works that last longer are scarce indeed. Nearly the only way for a work to gain prominence more than 30 years after its author’s death is if nearly all of their contemporaries’ work is lost, leaving them as our only window into a particular period, as with e.g. Polybius’ history of the Punic Wars, or The Secret Histories Of The Mongols—good for individual reputations, perhaps, but bad for society’s knowledge of history.

Might it be that old works going unread does not represent a loss of knowledge, because more recent works incorporate all the important information and arguments of older works, with improvements besides? This is a decent approximation of what happens in mathematics and the hard sciences, so one might expect to see the same pattern in historical analysis.

Unfortunately, comparing historical analyses of the same topics or events published in different centuries will show that this is not the case. Older works often contain important insights and key facts which are not repeated in newer works, not because they have been debunked or superseded, but because no historian has the time to read and integrate everything which has gone before. 

In our experience, there is no improvement in quality from working with newer sources, either.  If anything, older works that survive tend to be especially good, since mediocre retreads are less likely to endure for a century or three. There are also a few times and places where a field is so strong that selection effects alone are insufficient to explain the high quality of surviving materials—for example, scholarship of the Middle Ages written in the early 20th century is usually far better than anything published in the century before or after.

Historians of each generation prefer their own—but remarkably, the historians of each generation do not prefer the historians of a century prior over those of two centuries ago. We have heard people argue that Gibbon’s eighteenth-century work is inferior to contemporary historians because of the advancing march of knowledge, but never that Gibbon’s work is inferior to the historians of the nineteenth century. Odd, if there really is a discernible increase in quality from generation to generation.

Perhaps this is because history lacks the definitive logical proofs of mathematics and the experimental confirmations of physics, or perhaps this is because history is more frequently weaponized for propaganda. It is only a small exaggeration to say that the last fifty years of scholarship on the American Civil War has been entirely about fighting contemporary narrative battles, and has done nothing to advance our understanding of the period.

Whatever the reason, the “shoulders of giants” effect is mostly absent. New arguments are made, and old arguments are lost. There are local fluctuations in understanding, sometimes lasting for several generations, but no consistent upward trend is visible across timescales of, say, half a thousand years.

The exceptions, of course, are the truly significant archaeological finds. The rediscovery of the Rosetta Stone in 1799 and the excavation of Gobekli Tepe in 1995, for example, led to the recovery of crucial information and massive leaps in our understanding of ancient history. However, it is important to remember that the loss of such crucial information is invisible by its nature. The Rosetta Stone’s rediscovery led us to regain the ability to read hieroglyphics, which is still trumpeted centuries later, but the language’s loss in the early Dark Ages was likely a much quieter affair. 

The Future’s Understanding Of The Past

What is the most important piece of historical knowledge lost in the past century? It is impossible to say because if we knew the answer, the knowledge would probably not be lost.

What does all this mean for the overall trajectory of our knowledge of history? Unfortunately, on average our understanding of any given period will slowly get worse over time. Our direct evidence will gradually decay, with some temporary reversals as lost material is rediscovered. But over the long term, it is impossible for rediscoveries to outpace losses. Meanwhile, the quality of historians’ arguments and interpretations will vary somewhat, sometimes getting better and sometimes getting worse. To take a grossly oversimplified mathematical analogy, imagine a sine wave of analytical quality fluctuating around an exponential decay curve of surviving evidence.

How, then, does the work of historians fit into this? If the field is not a triumphant march towards ever more perfect understanding, is it a futile attempt to hold back an inevitable tide of ignorance? Not at all. Historians slow the decay tremendously, extending the lifespan of useful knowledge by decades at worst and millennia at best. Surely the ancient princess Ennigaldi-Nanna’ would be thrilled to hear that her lost museum was rediscovered 2400 years after her death.

More practically, while two centuries of analysis of the Industrial Revolution still leaves us with an understanding of the period’s economics far worse than that of any industrialist who actually lived through it, we would know much, much less than we do without the work of archivists who preserved the materials, as well as historians who described the important trends in ways that their contemporaries can understand. 

Regardless of how much of this knowledge might evaporate in a thousand years, nations like South Korea and China have already used it to guide their own industrialization, spreading prosperity to billions of people. Historians can carry lessons forward for much longer than this, such as when America’s founders based much of their new nation’s design on the successes and failures of Republican Rome. Historians fight a rearguard battle against the forces of entropy, buying time for their society to learn the lessons of our ancestors’ experience.

Boulton got wind of the burglars’ plot. Like any sensible person, he decided the responsible thing to do was to assemble a posse of his friends and employees, give them guns, and lie in wait. As one biographer told the story:

The robbers…were allowed to break in and seize their booty, and were making off with 150 guineas and a load of silver, when Boulton gave the word to seize them. A quantity of tow soaked with turpentine was instantly set fire to, numerous lights were turned on, and the robbers found themselves surrounded on all sides by armed men. Four of them were taken after a desperate struggle; but the fifth, though severely wounded, contrived to escape over the tops of the houses in Brook-row.

Boulton enjoyed the adventure immensely, as he wrote in letters to family and friends. Boulton had other escapades as well. He organized vigilantes to chase counterfeiters through secret passages, and turned his workforce into a makeshift militia when rioters threatened to sack his factory.

However, we don’t remember Boulton as an eccentric adventurer. Rather, he’s remembered for his role in kick-starting the Industrial Revolution. After James Watt invented his steam engine and spent years laboring in a fruitless partnership with John Roebuck, Boulton befriended Watt and bought out Roebuck’s stake. Boulton quickly expanded his factories, which had previously made knick-knacks like vases and buttons, and began large-scale manufacturing of the steam engines which so far had been only prototypes. These steam engines were soon powering industry across Britain, and within half a century led to the development of railroads and steamships.

So why was a leading businessman like Boulton so prone to chasing burglars? Antics like his might be chalked up to youthful enthusiasm, except that he was 72 when he ambushed the burglars. You might expect such a long-standing captain of industry to be sober and conservative, rationally maximizing his gains with a minimum of noise and fuss. Perhaps his business life was more moderate?

Hardly. Even before he met Watt, Boulton was heavily overleveraged, staying ahead of creditors by expanding into new businesses with borrowed money and charm. After diving into the steam engine business, Boulton found that his earliest customers—copper mines—were on the verge of bankruptcy. To help keep them afloat, he invested heavily in the mines and bought large loads of copper. To make use of his new stockpiles, Boulton then personally designed a steam-powered machine to mint copper coins. When counterfeiters began copying these coins, Boulton grew dissatisfied with the police, organized his vigilantes, and hunted them down. 

This is not unusual behavior for a man in Boulton’s position. Any survey of groundbreaking industrialists will reveal that they’re almost always a bit nuts. There are a great many examples cut from the same outlandish cloth as Steve Jobs, Howard Hughes, and Henry Ford. To restrict ourselves to people from our own time: Jeff Bezos declared in high school that he meant to move all of humanity into space and preserve the Earth as a huge international park; Peter Thiel is some kind of shadowy philosopher-vizier; Larry Page has poured a fortune into futuristic prototypes ranging from augmented reality glasses to self-driving cars to high-altitude balloons that broadcast internet service wherever they go; Elon Musk is Elon Musk. Mark Zuckerberg is almost unique in being a groundbreaking technologist who is a normie at heart—though it’s possible his obsession with Augustus Caesar, including the haircut, is more than just curiosity.

Where does this leave our archetype of the sober captain of industry? Are these people just a myth? Not at all. Pick five major corporations out of a hat, and in four of them, the leadership will be risk-averse managers whose vision of progress is one of incremental improvement to the status quo. There are a lot more Arvind Krishnas than Larry Ellisons.

The difference is that the sober captains of industry are rarely responsible for building new industries. The best of these people can guide an existing enterprise to ongoing success, as with Tim Cook, who developed Apple into the world’s most highly-valued company; or they can build new enterprises using technologies and business models that have been proven elsewhere, as with Leland Stanford, the railroad baron who linked California and the American Southwest to the rest of the country. But large-scale economic innovations require different skills and a different mindset.

To illustrate this difference, let’s compare two men who led the same company. Both embodied the American dream, rising from rural poverty to the forefront of business and public affairs. But the resemblance ends there.

Thomas Edison received no formal education. He was a self-taught tinkerer who supported himself as an itinerant telegraph operator and poured his meager salary into parts for makeshift inventions he would cobble together in rented rooms. He eventually achieved success and modest fortune with two advances in telegraphy: the quadruplex telegraph and an early stock ticker. At this point, an ambitious man in the “sober captain of industry” mold might have sought a high position in an existing company, or perhaps even started a new telegraph company of his own. Instead, Edison founded his famous Menlo Park laboratory, where he invented the incandescent light bulb, and even more importantly, the public power plant to distribute electricity to new light bulb owners. He could have had an illustrious career at the head of the Edison Electric Light Company. But instead, he turned to developing yet more industries. Edison would go on to found a dizzying array of companies. He mass-produced new inventions, including the phonograph and the motion picture, as well as expensive boondoggles in mining and cement manufacturing. Meanwhile, he reluctantly handed off operational control of his power company, which eventually became the General Electric Company (GE) after several mergers.

A later chairman of GE was Owen Young. He was born in a farmhouse in upstate New York in 1874, the same year Edison sold his quadruplex telegraph. Young scraped together an education, and by 1900 his drive and intellect had made him a hotshot young lawyer. Young was deeply involved in several cases and negotiations involving electric companies. This brought him to GE’s attention, and they recruited him as Chief Counsel in 1912. There he finessed negotiations with suppliers, labor unions, politicians, and the press. During World War I, he helped throw GE’s weight behind the American war effort.

Soon after the peace treaty, the United States Navy asked Young to organize a buyout of the British-owned Marconi Company, in order to put crucial radio infrastructure under American control and prevent the British from eavesdropping on secret American communications. Young led the buyout, and from 1919 he headed the resulting Radio Corporation of America, manufacturing radios in close collaboration with GE, where he became the chairman in 1922. A decade later, GE would sell its ownership in RCA to avoid antitrust concerns. Under Young’s leadership, GE and RCA became enormous manufacturers of consumer goods and appliances, cementing their positions as household names. Young was also selected by statesmen for key roles on two commissions renegotiating the payment of German war reparations: the Dawes Plan and the eponymous Young Plan.

Both these men are representative examples of their types. Edison was independent-minded and insatiable. Even after transforming the world around him by introducing lighting, power, and the day’s most popular entertainments—even when he was widely hailed as the age’s most brilliant champion of progress—he risked his fortune on new industrial ventures which could and eventually did fail spectacularly. If you were looking for someone to conservatively manage a piece of necessary infrastructure so that it would still be intact in 40 years, Edison would make a terrible choice. He did not fit into organizations or social roles established by others. He has a popular reputation as, frankly, an asshole, and this reputation was well-earned. Edison had the same pathological need to build that led Boulton to invent a new type of mint just to dispose of his excess copper. It’s difficult to say whether Edison would have approved of Boulton ambushing burglars, but he certainly would have understood.

On the other hand, it’s difficult to imagine Young reacting to Boulton’s reckless antics with anything other than befuddlement. Young was a man of the system, for whom everything should be done by the proper authorities acting through the proper channels, and he was extremely skilled at navigating those channels. He was ambitious enough to rise to the highest positions in industry. But while a man like Edison determined his own ambitions, Young’s ambitions were always received from figures in polite society—initially from wandering Unitarian professors, and later, directly from men like Woodrow Wilson and Herbert Hoover. Young was more than able to oversee the expansion of GE and RCA from manufacturers of heavy industrial capital goods into world-class manufacturers of consumer goods as well, but he would have no idea how to approach novel tasks like scaling up production of Watt’s unproven steam engines or building the world’s first public power plant.

Men like Young are necessary for any industrial society. As stewards of the major businesses and institutions, they keep the lights on and drive incremental progress. It is fortunate, then, that nearly every society encourages or tolerates the ambitions of such men. Countries like China and South Korea, which have reached the forefront of modern industry and continue to push forward the state of the art by small steps, have done extremely well by cultivating this type of leader, even as they limit those who would go further.

However, if we wish to achieve industrial breakthroughs and large-scale economic innovations, then we must also tolerate men like Boulton and Edison. Most societies will not do this. The natural reaction is to shut them down as dangerous threats to social order and economic life, and this is not completely wrong. Even their defenders use terms like “disruption” or “creative destruction” to describe the effects of their projects. Meanwhile, those who buck the status quo face media attacks and lawsuits, and more importantly, inescapable social pressure in their everyday interactions. In some places, the forces of conformity also wield unfavorable legislation, trumped-up corruption charges, or popular riots. Social technologies like American individualism, or the tolerance England once had for a gentleman’s eccentricities, are necessary if a society is to permit these madmen to run their experiments on the social fabric.

An illuminating case is the spat between Elon Musk, surely a mad founder if ever there was one, and the U.S. Securities and Exchange Commission (SEC). In 2018, Musk lied about Tesla’s financing on Twitter, likely to manipulate the stock price as part of his personal vendetta against short-sellers. The SEC sued, alleging securities fraud. If the SEC had used this as a pretext to remove Musk from power entirely, they might well have succeeded outright, and could at least have launched a major legal battle that would have dragged Tesla down for years. If this had happened in a conservative country like Germany or France, where regulators are hostile to economic disruption, such an attack would have been very likely.

Instead, Musk and the SEC reached a settlement two days after the suit was filed—a pace unheard of in America’s sclerotic legal system. Musk and Tesla were fined a combined $40 million, and some largely cosmetic changes were made to Tesla’s organizational chart which left Musk decisively in charge. Apparently, the SEC wants to make it clear that such behavior will be punished, but has no desire to destroy Tesla in general or Musk in particular. For American regulators, the idea of a major industry being run by a nonconformist oddball doesn’t raise red flags the way it would in most other countries.

Or for an older example, take the case of Guglielmo Marconi, who developed the first practical radios. His work was ignored and mocked within his native Italy. One of Marconi’s requests for patronage was later found buried in a government minister’s papers, with a scrawled note reading “to the Longara”, an insane asylum. In search of better prospects, Marconi emigrated to Britain in 1896, which at the time was a powerhouse of industry and technological innovation. Marconi was not even admitted into the country before a customs agent found the strange apparatus in his luggage and alerted the government. With the state’s backing, he soon founded a wildly successful company, dominating markets and making the world’s first transatlantic radio transmissions. Twenty years later, the United States Navy would mastermind Owen Young’s buyout of the American branch of Marconi’s radio company, which had become critical to national security.

While Italian society’s de facto top concern had been to keep Marconi in his place in order to avoid rocking the boat, in Britain the culture and institutions threw their weight behind his rise, even at the risk of rendering obsolete British interests in the trans-Atlantic telegraph cable, which had been the most advanced method of communication for 40 years before Marconi’s trans-Atlantic radio transmissions. Once Marconi had achieved fame and fortune in Britain, Italy was happy to welcome him back as a hero. For Britain before the World Wars, such support for disruptive innovation was not unusual. More than a hundred years before Marconi arrived, Boulton’s business succeeded in part because he persuaded Parliament to pass a special act extending the patent on his fledgling steam engines from 14 years to 25 years.

Tolerant institutions and culture have been a necessary component of industrial supremacy since the dawn of the Industrial Revolution. America was not always the only country that permitted breakthrough industrialists to build unproven new industries and upend the economic order, but it is today. If any other society wishes to match or outpace America in this regard, it will have to do the same.

Almost immediately after his rise, South Korea began implementing his industrial policy. While some of South Korea’s new industry was managed directly by the state, much of it was run by the chaebol, large conglomerates of private industry. The chaebol, however, were far from independent, and the state exerted substantial control over their actions. As Hyung-A Kim writes in her 2003 book, Korea’s Development Under Park Chung Hee:

Nothing would illustrate the effectiveness of the “administrative controls” on the business community more than the SCNR’s arrest of fifty-one prominent businessmen on the charge of “illicit profiteering,” and thus the confiscation of their property…These leading businessmen were released on 30 June, only after they had signed an agreement stating: “I will donate all my property when the government requires it for national construction.”

After establishing who was in charge, the Park regime was not shy about telling the businessmen how, exactly, to use their property for “national construction:”

Ku Inhoe, the founder of Lucky-Goldstar and one of the “illicit profiteers,” was ordered to build a cable factory. He would have preferred to build a textile factory…But in April 1962, exactly four years before the completion of Lucky-Goldstar’s Han’guk Cable Company, Ku was summoned by Colonel Yu Wonsik…Although details of this episode are not documented, which is not surprising, Ku, like many industrial proprietors at that time, seems to have received much more than “verbal” lessons. It took around just ten days for Ku’s company, Lucky-Goldstar, to complete a loan contract of US$2.95 million with Fuhrmeister, a West German company, which had clearly taken extraordinary steps to help address Ku’s needs…After this incident, no business leaders, including Ku, dared to resist directly the directives that emanated from the SCNR’s industrial planning committees.

Once Ku shaped up and came on board with Chung-hee’s program, he was treated very well. His company, Lucky-Goldstar, is known today as LG. Instead of cable or textiles, LG has a sterling reputation for manufacturing advanced electronics such as cell phones. The company is still run by Ku’s descendants.

Under the direction of the SCNR and its successors, chaebol such as LG built up South Korea’s basic industries like agriculture and concrete, then progressed through steadily more advanced industries in record time. This required the construction of factories, and by the mid-70s, South Korean industry had grown large enough that they needed to import machine tools for these factories in large quantities.

Machine tools are tools for cutting and shaping metal more precisely and efficiently than hand tools ever could, originally controlled by manual levers and wheels or rigid metal pattern jigs, but nowadays mostly via computer control.

https://www.youtube.com/watch?v=81UjjSH2iFw

A machine tool in action

These “tools that make the tools” are a crucial piece of modern supply chains, and most advanced manufacturing would be impossible without them. The ability to produce machine tools domestically is one of the foundations of a deep industrial base. Recognizing this, the South Korean state encouraged Hyundai, a chaebol which built the first all-Korean automobile in 1975, to begin making machine tools for the growing domestic market. Since then, the South Korean machine tool industry has grown steadily alongside the Korean economy as a whole. Production today is slightly larger than the machine tool industry of the United States, a country more than six times as populous.

This was just one element of South Korea’s “Miracle on the Han River,” which catapulted the country into the ranks of the world’s most developed nations, even as the government transitioned from an authoritarian dictatorship to a liberal democracy. By now, the South Korean industrial base has not only raised tens of millions out of poverty, but also lifted the nation from a backwater of little geopolitical significance into an important regional player and a crucial part of global supply chains.

Korea’s story may be more extreme than most, but this type of state-led economic development is how every wealthy country on Earth has industrialized. The sole exception is Britain during the original Industrial Revolution. In spite of the Crown’s active economic policy in many other areas, the earliest industry was built with very little government support, and that only tangential and undirected. Examples of state involvement in industrialization abound. America’s government-funded railroad boom of the late 19th century made possible the burgeoning Pacific cities and the agricultural boomtowns of the Wild West, while the demand for rails provided a market to fuel the explosion of steel mills using the groundbreaking Bessemer process. The Five-Year Plans of the Soviet Union developed the heavy industry that would launch Sputnik and kick off the Space Race. In China today, hundreds of millions have been lifted out of poverty by state-directed economic development.

Such transformations require high-level guidance from a competent state coordinating the key industries and establishing crucial infrastructure for the new economic system. Building a single industry in isolation does not transform a society, and frequently cannot be done at all without the other industries that support it. The right industries must be built in the right order, and often in parallel, as in the case of Korea’s automobile and machine tool industries. In the presence of existing industry available in international trade, the coordinated investment in localizing industrial capability is beyond the ability of undirected markets, no matter how good they are at hill-climbing. Radical change has enough different steps, moving parts that must be coordinated, and short-term investment for long-term gain, that it must be planned by a human mind. The great strength of capitalist societies is that industrialists like Henry Ford or Jeff Bezos can transform single industries within an existing economic framework, but the economic framework itself is too large to be remade with the resources of any individual or corporation. Only the state can coordinate many different industries to produce a transformation at the scale of industrialization. Some attempts at this have been more competent than others, of course, but whether the state accomplishes its industrial policy via purchases and subsidies on a relatively free market, or by more directly coercive methods, the economic effects are similar.

The industrial capacity of a nation can often be predicted by its ability to produce machine tools at high quality and quantity, as seen with South Korea as its industrial capacity developed. Machine tools are not only vital to industrial supply chains and an unfakeable indicator of real industrial capacity, but they also provide a lens to examine broader industrial and development policy. Machine tools, like the industrial economy as a whole, require a well-educated workforce, substantial and well-coordinated capital investment, and a robust consumer base—none of which can be produced by a single machine tool manufacturer. Creating the context in which machine tool production can develop and flourish requires coordination between government, business, and finance elites; much of this is captured by the term “industrial policy,” but factors ranging from the nature of the financial and legal systems to culture and positioning within global trade systems are also critical. Substantial state capacity is required to plan and execute the process of industrialization. In a recent report published by Bismarck Analysis, we described the involvement of the state and other elite institutions in the machine tool industry of every major producer, as well as some failed attempts to build this industry. What follows is based on that research.

Britain Sparks the Industrial Revolution

The first machine tools were invented in Britain to support the metalworking needs of early industrialists. Watt’s steam engine, which began production in 1776, was made with a boring machine that had been invented only two years earlier to make cannons. The Crown paid little attention to this nascent industry, although it was quick to buy machine tools to manufacture guns and ships, and many historians believe Britain’s relatively new patent system was a necessary precondition of the period’s rapid technological progress. Nevertheless, there was no state effort to support machine tool production as such, since there was no way of knowing how important such tools would become.

Once the British realized what they had, Crown and Parliament stepped in to support advanced manufacturing. British technology was kept from foreign competitors by a succession of bans on export, starting in the late 18th century. However, as demand rose in the early 19th century, makers of products such as machine tools would routinely circumvent these bans and risk the heavy fines in order to sell to markets in the U.S., Prussia, and beyond. By 1841, under pressure from machine tool makers, these export bans were lifted in favor of a free trade policy. This increased profits but allowed foreign powers to accelerate their coordinated attempts to replicate British technology.

However, foreign powers did not respect British patents, and there was minimal support to protect domestic manufacturers from foreign competition backed by industrial policy. British machine tool makers at first blazed trails beyond what anyone even imagined, but the extreme cost efficiency and technological superiority of British manufacturing relative to foreign competitors, which had propelled the massive boom of the early Victorian era, waned over time. British manufacturing supremacy gradually eroded and ended up facing inwards to the captive markets of the empire. In the years after the Second World War, as Britain’s ability to protect what remained of its once-formidable industrial output slipped away, so too did Britain’s ability to maintain its overseas empire and stand on its own against the new superpowers. Today, the United Kingdom produces a mere 1% of the world’s machine tools.

The American Juggernaut

The first countries to surpass Britain industrially were Germany and the United States, whose policymakers were the first and best at aggressively supporting machine tool production in their own home markets. In America, an early two-prong strategy protected American companies with the series of trade barriers kicked off with the 1828 tariff act, while avoiding recognition of patents protecting English inventions. The federal government also purchased mass-produced rifles, creating a government-supported market for machine tools, much like that which supported Hyundai in Korea. American firms were actually exporting back to England by the 1850s. This was followed by a century of meteoric rise in industrial output. By the end of the Second World War, in the afterglow of extraordinary investment in arms manufacturing, America was fully dominant in the global production of machine tools and almost everything else.

However, in the post-war years, U.S. policymakers were more interested in rebuilding and supporting their Cold War allies than protecting domestic manufacturing. The domestic market remained colossal, but first German and then Japanese firms made massive inroads, and today, the majority of American machine tools are foreign-made. A common thread can be drawn between postwar trade policy and the late British Empire, where the growth of foreign competitors, in conjunction with lack of political support for domestic machine tool production, also led to a substantial fall in output. In a sense, both stopped engaging in deliberate industrial policy for manufacturing, while operating in the global market where other countries were still very much pursuing deliberate policy.

Bismarck’s German Industry

In contrast to the great English-speaking countries, Germany is a perfect example of what continued support for manufacturing can look like. With the Unification of Germany, orchestrated by Otto von Bismarck in 1871, high-level industrial policy became possible. Bismarck met with leading industrialists and coordinated a complex economic policy including targeted tariffs, while German industrialists sought to acquire British technology. By the eve of the First World War, German machine tool production was second only to the United States.

The situation in Germany from the dawn of the First World War until the end of the Second was turbulent. By 1945, the occupied and partitioned German state was devastated. However, after the war, West Germany engaged in massive reconstruction. West Germany supplemented American aid with massive domestic investment and planning focusing on supporting key industries and rebuilding a stable export-oriented manufacturing economy. In contrast, France used American aid for non-infrastructure projects such as the war in Vietnam and continued a pattern of importing machine tools from elsewhere. German programs such as the 1952 “Investment Aid Law” forced hundreds of thousands of firms to contribute funds for investment in bottleneck sectors of the quickly growing post-war economy, while substantial tax breaks, loans, and grants supported machine tools and other heavy industries. By the 1960s, West German industry had largely recovered to pre-war levels and after the rise of newer technologies West German manufacturers became competitive in the U.S. home market, and rose to being one of the top producers in the world.

State intervention in the German economy has long been understated by the German government. When direct U.S. aid played a substantial role, Cold War ideology made it politically untenable to speak of central economic planning and interference with private enterprise. In the modern era, with the proliferation of free trade agreements and zones, the best way to ensure access to foreign markets is to sing the praises of the free market. The modern German state, however, is more subtle rather than less active in promoting specific modes of economic activity: from structuring financial institutions to encouraging domestic investment and supply chains, to vast subsidized research institutes for manufacturing technologies, to the German role in shaping the European Union, the German state has supported high-value manufacturing since Bismarck.

Japan’s Rise

Japan’s path to a domestic machine tool industry and robust domestic manufacturing base has a lot in common with Germany’s, but started from a less developed point. Japan was a largely agrarian and feudal society, but beginning in 1868 under Emperor Meiji, Japan catapulted into vastly more modern forms of government and economy. Careful Japanese observers traveled around the Western world, especially Germany, to determine what worked best. In 1889, Emperor Meiji put in place the new Meiji constitution, based on that of the newly unified German Empire, and Japanese policymakers brought in foreign machinery and advisors to reproduce the capabilities of foreign industry. The largest focus was on bringing military technology and heavy industry to a level that allowed Japan to compete with Western powers, culminating with the Japanese annexation of much of East Asia. By the Second World War, Japan was advanced enough to produce a vast navy and support the invasion of mainland China, but their infrastructure was still behind that of Western Europe and the United States. It was in the post-war years under U.S. occupation and then as a U.S. protectorate that Japanese policymakers worked with their large business conglomerates—the keiretsu—that Japanese manufacturing finally caught up. With a strong focus on exports, Japan overtook first the U.S. and then West Germany in machine tool production, and since 1990, has remained at a similar level of production to now-unified Germany. Japan maintains relatively closed markets and deeply interlinked companies and financial institutions, which combines with substantial ongoing state investment to maintain leading-edge manufacturing for export. South Korea and Taiwan’s later industrialization relied upon a style of industrial policy heavily inspired by Japan, and on the same preferential access to U.S. markets.

In fact, the ability to buy from and sell to American companies and consumers enormously helped all of the great success stories within market economies: Germany, Japan, Korea, Taiwan, and several other Western European countries. These countries dominated production capacity through the end of the Cold War, while many trying to operate outside the U.S.-led global order failed to fully develop a domestic manufacturing base.

Failures in the Periphery

Much of the world has never developed substantial industrial production. In countries such as Brazil and India, industrial policy has had mixed results. Efforts in Brazil began in the 1930s, but were largely decentralized and contradictory. Import restrictions subsidized low-end domestic production, even as imports of higher quality machinery continued; as a result, domestic firms had no natural path to more sophisticated manufacturing, and have mostly stagnated. India has also had largely contradictory and poorly coordinated industrial policy, and despite substantial demand for machine tools in low-end manufacturing, India has limited domestic production. Many countries in the position of India or Brazil rely on global trade systems both for raw materials and for end markets, but do not have the leverage to gain access without taking down their own tariffs and protection of domestic industry. Accumulation of the capital needed to invest in advanced manufacturing is also a challenge; foreign capital investment is difficult to direct to strategic industries, but without foreign assistance or very strong domestic coordination, domestic capital accumulation is almost never achieved.

Russia’s Collapse and China’s Success

Russia and China, however, have a different story to tell.

Soviet policymakers well understood the necessity of machine tools, but did not operate within the market framework that led to American or German success. Instead they centrally planned production via the famous Five-Year Plans. By the time of the Second World War, the Soviets were able, through massive proliferation of low-end machine tools and vast factories across the country, to produce overwhelming numbers of tanks and planes to support the war effort. They still relied on imports from the even more formidable American military machine, but the contrast from 1890 to 1940 is staggering.

Almost as staggering is the fall that came later. Soviet machine tool production continued to rise after the war, and by the 1970s overtook the United States, which by then was sourcing many machine tools from other NATO countries. Soviet industry was often thought to be an unstoppable juggernaut. However, in the 1970s and especially the 80s, the Soviet system began to fray at the edges, and after the collapse of the Soviet Union in 1991, machine tool production dropped precipitously. Soviet industry had always been inefficient, but with the protection of the Iron Curtain, better and cheaper machine tools from the West could not drive Soviet producers out of business. The 1990s brought sudden withdrawal of state subsidies, introduction of free trade, and elimination of price and exchange controls, collectively termed “shock therapy.” This, along with the breaking up of supply chains distributed across many different regions of the now fractured Eastern Bloc, destroyed much of the region’s productive capacity. To this day, Russia tries to rebuild machine tool production capacity to support its relatively modest military needs but fails even at that. Many attempted policies fall flat due to corruption, lack of state capacity, limitations of global trade rules, and the sheer weakness of both consumers and producers of machine tools in Russia. The implications for sovereignty are made starkly evident when American sanctions can restrict Russian access to high-end German or Japanese machine tools vital to maintaining their dilapidated navy.

China comes out of a sister tradition to that of the Soviet Union. After the communist forces won the civil war in 1949, Soviet support for the revolutionary cause became direct aid to the fledgling state to build a bulwark against American influence in East Asia. The early years of Maoist China had uncountable failures—more so even than the early years of communism in the Soviet Union—but one of the more successful programs was a massive transfer of industrial technology spanning 1949 to 1966. While Sino-Soviet relations broke down by the end of this period, early on there were enormous strides as Soviet advisors came in and built entire factories with copied over workflows across dozens of Chinese cities. By the time of the Sino-Soviet split, Chinese machine tool production was self-sufficient, although quality and efficiency were poor.

Liberalization reforms under Deng Xiaoping allowed the Chinese machine tool sector to become far more efficient, along with the rest of Chinese industry. Later efforts to bring in foreign technology through control of lucrative direct investment deals and state-financed acquisition of foreign companies catapulted some leading Chinese machine tool producers to a technological level far beyond what the Soviets had achieved. Chinese machine tool production expanded exponentially from 1990 to 2015 to outfit massive factories built across the country, serving export-focused manufacturing directives driven by Chinese planners. Today, China produces more machine tools than any other country but still relies heavily on imports from Germany, Japan, Taiwan, and Korea for more high-end needs. China has not yet solved the problem of self-sufficiency; its supply chains still rely on essential partners within the U.S. alliance. However, through implementing a market system yet keeping in place massive state-owned enterprises, state-controlled investment banks, import and export restrictions, strong requirements for any foreign investors, currency manipulation, and many other policies, China has to this point managed to build an extraordinary industrial base.

Industrial Policy

Chinese policymakers understand industrial policy. Li Keqiang, for example, spearheaded the Made in China 2025 initiative to develop Chinese high-tech manufacturing. This includes subsidies and intellectual property transfer requirements intended to promote domestic production of equipment such as machine tools. China has used such policies to substantial effect for decades. For a modern economy, machine tools are just as essential as roads, and the state has a necessary role in building both types of infrastructure.

Savvy states often support such foundational industries via manipulating the structure of the market. Targeted tariffs, for example, raise the price on foreign imports within certain categories, thus helping domestic “infant industries” compete where they otherwise couldn’t; in theory, the tariffs are then phased out as the industry matures. This can be part of a broader import substitution strategy wherein trade barriers are put in place while the currency value is kept high, helping producers acquire capital goods necessary to build factories, and controls are placed on certain types of financial transactions. The goal is to allow a product that has been typically imported to instead be made by domestic producers, thus reducing dependence on foreign nations.

Another type of industrial policy is export discipline, wherein government policy supports industries on the basis of their ability to export to global markets (often the U.S. market). This can take the form of direct subsidies, access to loans, or other incentives tied explicitly or implicitly to export volume. This forces domestic producers to make products that are up to world standards, but allows them to do so less efficiently while the industry is being established.

A more direct approach is for the government to simply purchase a favored industry’s products, as the U.S. did for military hardware during and after the Second World War. These massive investments in American industry helped propel the country to staggering heights of production, and launched industries including space travel, the internet, and solar energy.

Industrial policy can also include broader things like education and public research. Germany invests a great deal in education and research for manufacturing. Since the Second World War, German policymakers have tailored their workforce to support advanced manufacturing.

Intellectual property rights are a piece of industrial policy so ubiquitous they are often taken for granted. The United States has long maintained a patent system inspired by the English, but during critical parts of the American industrial revolution English patents were not respected, allowing American manufacturers (with the aid of tariffs) to dominate foreign competitors. Now, American intellectual property law and strong enforcement is an essential part of ensuring investment in research and development, and American negotiators seek to entrench respect for U.S. patents in any country seeking free trade with the U.S. Britain’s history suggests American negotiators will eventually fail in this goal.

Financial systems are another way that states can support favored industries. In countries like Germany and Japan, heavily regulated banks act in part as arms of the state, ensuring easy investment and access to capital for key industries. In contrast, the American system is much more laissez-faire, and relatively low-return investments like machine tool production simply do not do well.

A reality of industrial policy is that every economic situation is unique. The toolbox extends far beyond the few policies discussed here. The reconstruction of post-war West Germany, where an industrial titan was reborn from the ashes, requires different policies from the 1950s reforms that jump started Taiwan’s growth from a nation of poor farmers to a major global exporter of manufactured goods. These are both very different from the problem of reversing industrial decline which now confronts much of the English-speaking world. Countries from the United States to Canada to the United Kingdom have at least begun to de-industrialize, importing a much larger fraction of their manufactured goods.

Decline and Revival

The United Kingdom’s decline prefigured that in the United States. Early signs included the German supersession of British production output before the First World War and the reliance of the British state on imports to support the war effort. As per The Collapse of British Power Vol. 1 by Correlli Barnett:

The shortcomings of the British iron and steel industry alone were proof enough that the Motherland, no less than the empire, was not the great independent power it seemed, for as The History of the Ministry of Munitions states: ‘It was only the ability of the Allies to import shell and shell steel from neutral America…that averted the decisive victory of the enemy’

In the years after the Second World War, the collapse became acute. As the United Kingdom transformed from a hub of manufacturing for its empire and the world to a mere part of the European and Atlantic economic ecosystems, occupying niches such as finance and high-tech, the British state became unable to act independently of nations such as the United States, as demonstrated during the Suez Crisis in 1956. Like the Russians or the Chinese, who cannot operate their essential industries without relying on advanced machine tool imports, the United Kingdom is also dependent in a multitude of ways that it was not before the decline of domestic manufacturing.

The decline in American machine tool production has not progressed nearly as far. The relative power of America in the world may not be at its apex, but there has been no great, obvious collapse—no loss of overseas possessions, no supplicant position with regards to foreign economies like the European Union—and overall American manufacturing output is still strong for now. It was confidence in American power that allowed domestic industries to be replaced by imports from Cold War allies. Policymakers in the 1990s, at the end of history, believed that the victorious American global alliance would flourish into an eternal system of free countries spanning the entire world. If U.S. supremacy would never end, then there was no need for domestic manufacturing capacity to support sovereignty, with the distortions to efficient markets that that entails.

However, even a global superpower with no real peers like the U.S. cannot import critical technologies from competitors without consequences. Reliance on China for cheap steel is one thing—5G infrastructure is quite another. Even stalwart allies like Germany and Japan, which supply the overwhelming majority of advanced machine tools to the remaining American heavy manufacturers, cannot be seen as just protectorates without their own interests. Looking at the history of machine tools, and what happens to countries that have to rely on foreign powers to get them, makes clear that what remains of industry in the former powerhouse of global manufacturing is worth reviving.

The state has a role to play in running an advanced economy. In the United States, many institutions of government are too dysfunctional to do the part required of them, and finance is too used to free rein to take kindly to the reassertion of their responsibilities. The exact path forward is unclear, but coordination among the economic power centers of American society must be reestablished to rebuild the engine of creation.