The key point of this contribution, its piece of the puzzle that, together with the others, forms the overall picture before our eyes, is this:
You can't push things too far.
We cannot go on indefinitely making life increasingly unbearable with impunity.
Enough is enough.
(It should go without saying that enough has two sides: quality and quantity.
Enough in terms of quality is the excessive degree of pervertion.
Enough in terms of quantity is the excessive number of pervertions.)
And when the rope snaps, it snaps suddenly, and it explodes in our faces.
And plenty of people get hurt – not just those who deserve it.
Thus, common sense and simplicity, perpetually snubbed by idiocy and overcomplication, are perpetually proven right, not wrong.
Those on the side of common sense and simplicity are not a silly ball and chain.
On the contrary, they are the ones trying to save us from the idiots and their umpteenth disaster.
In 1988, an anthropologist named Joseph Tainter published a graph that should have changed everything. The graph showed a single curve that described what happens inside every civilization right before it collapses. Not some of them. All of them. Seventeen civilizations across five thousand years, and they all hit the same wall.
In this deep-dive documentary, we trace the most unsettling discovery in complexity science: every civilization that has ever collapsed was, at the moment of its disintegration, more complex than it had ever been. From the Roman Empire's legal code exploding from 140 provisions to millions of lines, to the Ottoman bureaucracy expanding 40x in 250 years, to the US Federal Register hitting 106,000 pages in 2024, the pattern is identical. Civilizations do not collapse from external shocks. They collapse from the weight of the systems they built to save themselves.
We follow four independent researchers who arrived at the same conclusion from completely different fields. Tainter from archaeology. Yaneer Bar-Yam from physics at MIT, whose food price model predicted the Arab Spring four days before it began. Peter Turchin, the Soviet dissident's son who predicted 2020 instability from 2010 data and was called crazy until every prediction came true. And Mancur Olson, the North Dakota farm boy who proved that political stability itself breeds the conditions for collapse.
Then we confront the mathematics. A 2025 proof demonstrated that civilizational collapse behaves as a first-order phase transition. Not gradual decline. Sudden failure. No warning signals. The system appears stable until the instant it is not. The bridge holds until it doesn't. And the math says you cannot tell the difference between stability and the moment before collapse.
Every claim is grounded in published research, specific data, and the words of the scientists who produced them. The numbers are real. The curve is real. And you are on it right now.
There Is a Complexity Threshold That No Civilization Has Ever Survived
by:
Prior Signal
Faultlines Studio
Transcript:
In 1988, an anthropologist in Utah published a book with a graph in it that should have ended careers. The graph showed a curve, a single curve that described what happens inside every civilization right before it collapses. Not some of them, every single one he could find. 17 civilizations on six continents across 5,000 years of recorded history. and they all traced the same ark. They all hit the same wall.
His name was Joseph Tainter. He was not famous. He was not trying to be. He had spent most of the previous decade on his hands and knees in the red dirt outside a dormant volcano in New Mexico, cataloging the ruins of people who had walked away from their own cities. And what he found in those ruins, the pattern that kept showing up no matter which dead civilization he examined, is something that I have genuinely not been able to stop thinking about since I first encountered it. I have spent an embarrassing number of late nights with this material. And the deeper I went, the worse it got. Because here is what Tainter discovered. And here is what nobody wanted to hear.
Civilizations do not collapse because of barbarians. They do not collapse because of drought or plague or earthquakes or resource depletion or any of the things you were taught in school. He tested 11 of those conventional explanations across all 17 cases. Every single one failed. Some of them failed embarrassingly. The barbarian theory, the one that gets trotted out in every documentary about Rome you have ever watched, cannot explain why Rome was beating those exact same barbarians for centuries before the collapse. The barbarians did not suddenly get stronger. Rome got weaker. And the reason it got weaker is the reason every civilization that has ever collapsed got weaker.
They built too much to save themselves and the thing they built is what killed them. What nobody knew at the time Tainter published, what would take another three decades and researchers from completely unrelated fields to confirm is that this pattern goes far deeper than metaphor or historical tendency. It is a mathematical law and it has a threshold, a specific calculable point on the curve where the system flips from functional to doomed. And the most recent proof of that threshold published just last year says something about its behavior that I personally think is the most disturbing finding in any field I have ever covered on this channel. It says there is no warning. The system looks fine right up until the moment it is not fine and then it is over. But I am getting ahead of myself. We need to see the curve first. We need to understand what Tainter actually found because the setup matters. You need to feel the weight of the evidence before we get to what it implies about the room you are sitting in right now.
Taint's argument starts with something so obvious it sounds almost stupid when you say it out loud. Societies are problem-solving organizations. That is what they do. That is all they do. A group of humans encounters a problem. Drought, invasion, famine, trade disputes, crime, and they solve it by adding complexity. They create a new role, a new institution, a new law, a new tax, a new branch of government. Early on, this works beautifully. The first irrigation system transforms a village. The first legal code prevents blood feuds. The first standing army stops the raids. Each addition of complexity produces enormous returns. Your life gets tangibly better. The math is easy. Invest in complexity, get back more than you put in. But here is where the curve starts to bend. And I need you to pay attention to this next part because it explains everything that comes after. You solve your easiest problems first. That is just human nature. The low-hanging fruit gets picked early. Each subsequent problem is a little harder, a little more expensive, a little more resistant to simple fixes. The 10th law you write does not produce the same benefit as the first. The hundredth bureaucratic office does not deliver the same efficiency gain as the tenth. The returns on each additional unit of complexity start declining. Not immediately, not dramatically, but steadily, predictably, mathematically.
Tainter graphed this. He called it the «marginal return on complexity». And the curve he drew has a shape that once you see it, you cannot unsee. It rises steeply on the left, flattens in the middle, and then at a certain point, it crosses zero and goes negative. Past that point, every new regulation, every new agency, every new layer of administration costs more than it produces. The civilization is now spending more on the machinery of problem solving than the machinery is returning in solved problems.
And the cruel part, the part that I think makes this fundamentally different from every other theory of collapse I have encountered is that the people inside the system cannot stop because the problems have not gone away. The barbarians are still at the border. The crops still need water. The trade routes still need protecting. So the society keeps adding complexity, keeps climbing the curve past the point of negative returns, keeps spending more to get less until the whole structure becomes so expensive to maintain that collapse, actual collapse becomes the more economical option.
That is Taint's thesis. Collapse is not a catastrophe. It is an economizing process. A civilization shedding complexity it can no longer afford. And he did not just theorize this. He showed it happening across 17 independent cases. I want to walk through one of them because the specific numbers matter here and the one that has the best data is also the one you think you already understand.
Rome 451 BC. The Roman Republic governs itself with the 12 tables. 140 provisions short enough that school boys memorized them. The entire legal apparatus of what would become the most powerful civilization in the western world fits on a few bronze tablets in the forum.
That is where the curve starts. Now watch it climb. By the time Emperor Justinian ordered a consolidation of Roman law in 534 AD, his legal commission had to compress 3 million lines of accumulated legal text into something usable. 3 million lines. They managed to get it down to 150,000 spread across the digest alone which pulled from the writings of 38 separate jurists. The Codex Justinianis, the portion dealing specifically with imperial constitutions, contained 4,562 individual statutes from 140 provisions to 45,562 statutes. That is a 33 times multiplication of the legal code across a single millennium. And I am not even counting the Theodosian code that came before it, which already held over 2500 constitutions by 438 AD. And no, I am not making that up. These documents survive. You can go read them.
But the legal code was just one dimension of the complexity explosion. The administrative structure tells the same story from a different angle. Under Augustus around 27 BC, Rome administered roughly 25 to 28 provinces, clean, manageable. By the time Diocletian took power in 284 AD and decided the empire needed restructuring, he doubled the number of provinces to around 100, organized them into 12 dasis and four prefectures and separated civil and military authority into entirely parallel command chains. The civil service roughly doubled from an estimated 15,000 to 30,000. A contemporary named Lactantius, admittedly hostile to Diocletian, wrote that there were now more men using tax money than there were paying it.
Diocletian also issued the edict on maximum prices in 301 AD, which set price ceilings on over 200 goods, services, and wages across the entire empire. The penalty for violating it was death. It was universally ignored within 4 years. Think about that for a second. The most powerful man in the Western world issues a death penalty-backed economic regulation covering 1,200 categories and the complexity of the economy simply routes around it like water around a rock. The system had become too complex for even absolute authority to control.
And then there is the coinage. This is Tainter's favourite metric, and honestly it might be the most elegant proof of his entire thesis. Under Augustus, the Roman Denarius was roughly 95% silver around 3.9 grams of actual precious metal. By Marcus Aurelius about 75%. Under Septimius Severus below 60. By the time of Gallienus around 269 AD, the dinarius contained between 2 and 5% silver. The coin still looked like a coin. It still had an emperor's face on it, but it had been hollowed out from the inside. Each debasement was a complexity response. The army needed paying. The bureaucracy needed funding. The frontiers needed defending. Minting more coins from less silver was the simplest available solution.
The first debasement genuinely financed real capability. The last ones financed nothing. They were the empire feeding on its own skeleton. And here is the part that should disturb you. If you plot the silver content of Roman coinage over time, you get a curve. And that curve matches the shape of Tainter's marginal returns graph almost perfectly. Steep value on the left, flattening in the middle, collapse on the right.
But one civilization is not a law. One data point, even one as well doumented as Rome, is an anecdote. And Tainter knew that. So he kept looking. The Ottoman Empire followed the same trajectory with no knowledge of Tainter's theory and no structural connection to Rome. In 1365, Murad I administered two provinces. Two. The imperial council had three viziers. By the late 1500s, the empire governed 42 provinces with 11 viziers and a central bureaucracy that had grown from 38 salaried scribes in 1549 to roughly 1,500 by the 1790s. That is a 40 times increase in administrative staff in 250 years. The Janissary Corps is maybe the most vivid example of complexity eating its own host. Originally about 2,000 élite soldiers recruited through the Devshirme system, trained for up to 14 years in the seven tiered Enderun Palace Academy. By 1808, the call had swelled to 140,000, of whom only about 10% were actually trained soldiers. The rest were civilians who had purchased enrollment for the tax exemptions and privileges.
They had stopped being a military force and started being a protection racket. In 1810 alone, Janissaries set over 2,000 arson fires in Constantinople and then demanded bribes to extinguish them. I actually find this detail almost darkly funny in a way that makes you wince. The institution built to defend the empire was literally burning down the capital for pocket money. Sultan Mahmud II solved the problem in 1826 by surrounding their barracks with artillery and killing over 6,000 of them in a single day. The event is officially called the auspicious incident which is one of history's more savage euphemisms.
The Ottoman legal system reached a complexity that borders on the absurd. At its peak, four to five parallel legal jurisdictions operated simultaneously. Sharia courts, secular canon courts, Millet courts where non-Muslim communities adjudicated under their own religious laws, capitulary courts for foreign nationals, and after the Tanzamat reforms, a new layer of secular state courts on top of everything else. The Messel, the Empire's attempt at legal codification between 1869 and 1876, produced 1851 articles across 16 volumes and it covered only civil law. Family law was not even touched.
The Maya Chako Canyon, the Western Zhou dynasty, which collapsed in 771 BC. Tainter tracked monumental construction among the Maya as a proxy for complexity investment, and the curve peaks and collapses in exactly the shape his model predicts. The Chakoans in New Mexico built an extraordinary trade network linking different altitude zones to hedge against climate variability.
And it worked until the coordination costs exceeded the benefits. At which point they did something remarkable. They just left, walked away from their great houses and went back to simpler ways of living. Voluntarily shed complexity. Tainter would say they made the rational choice. The curve told them it was time to go. 17 civilizations, every continent, 5,000 years, the same curve.
There is one more voice I need you to hear before we leave the historical record because he comes at this problem from a direction nobody expected. And honestly, his story is one of the saddest in this entire saga. Manka Olsen grew up on a farm near Buckton, North Dakota. Son of Norwegian immigrants, brilliant kid, won a road scholarship to Oxford, got his PhD at Harvard, and in 1965 published a book called The Logic of Collective Action that, in the words of Jonathan Roush, blew a hole in the hull of American political science. He proved something uncomfortable: rational individuals will not act to achieve common group interests unless the group is small enough or offers selective incentives.
The free-rider problem. It sounds abstract until you realize it explains why small organized lobbying groups consistently defeat the diffuse interests of millions of citizens. But it was his 1982 book, The Rise and Decline of Nations, that connects directly to Tainter's Curve. Olsen showed that in politically stable democracies, small interest groups accumulate over time like plaque in an artery. Each one lobbies for protectionist policies whose benefits are concentrated among members while costs are spread invisibly across everyone else. One lobbying group is harmless. A thousand of them accumulated over decades of uninterrupted stability makes the entire economy rigid, slow, and incapable of adapting. He called it institutional sclerosis.
And here is the punchline that I keep coming back to. His core prediction was testable and it tested well: the longer a society enjoys uninterrupted political stability, the slower it grows. Postwar Germany and Japan, whose old interest group networks had been completely destroyed, experienced economic miracles. Stable, undisrupted Britain stagnated under layers of accumulated coalitions. Stability itself is the disease. The thing that feels like health is the thing that is killing you.
Olsen collapsed outside his office at the University of Maryland on February 19th, 1998. Heart attack, dead at 66. He was buried at the Grew Norwegian Lutheran Cemetery near the farm where he grew up. I find something almost unbearable about that. A man who spent his life studying how institutions die, dying suddenly without warning, in the middle of his work. His colleagues said he was in the middle of his most ambitious project yet. It was never finished.
But this was not just about the past anymore. A physicist at MIT was about to prove that this pattern has the force of a natural law behind it. Yani Bayam grew up in Boston, got his PhD from MIT studying semiconductor defects, and then did something unusual for a physicist. He pointed his mathematical tools at human civilization and asked a simple question. Is there a maximum complexity that a hierarchical organization can handle? The answer is yes, and the limit is hard. Bayam formalized something called the multiscale law of requisite variety which is a mouthful but the core idea is elegant enough to explain in a sentence.
The complexity of your organization must match the complexity of the problems it faces at every scale simultaneously. Not just overall: at every level. A government that has enough total bureaucratic capacity but concentrates it all at the top cannot solve problems that require local adaptation. A decentralized system that handles local issues beautifully cannot coordinate a national response.
And here is the part that really matters. Bayam proved a conservation law. When your total resources are fixed, increasing complexity at one scale necessarily decreases it at another. You cannot have it both ways. There is no free lunch in complexity space. Bayam published the proof. It is a theorem not a suggestion. And Bayam did not just theorize, he made predictions. His food price model identified a critical threshold specifically when the FAO food price index crosses 210 above which social unrest becomes highly probable.
He submitted this analysis to the United States government on December 13th, 2010. 4 days later, a fruit vendor in Tunisia set himself on fire and the Arab Spring began. I spent a long time sitting with that one: 4 days; the math was right. The government had the paper on their desk and nobody acted. His ethnic violence model published in science in September 2007 found that when ethnic groups are distributed in patches between 20 and 60 km wide, violence erupts with 90% geographic correlation. Below 20 km, people are well mixed and peaceful. Above 60, well separated and peaceful. In between, the math says blood. He tested it against actual violence in India and the former Yugoslavia. 90% match.
But I personally think the most underrated part of Bayam's work, the piece that ties back to Tainter in a way that should keep you up tonight is the implication for modern institutional complexity. Because what his math says is that technology does not save you. Technology increases the complexity of the environment faster than it increases the complexity of the organizational response. The world gets harder to govern faster than governance gets better at governing it. The gap between environmental complexity and institutional complexity widens. Which means every smartphone, every algorithm, every interconnected global supply chain functions as an accelerant for the very complexity it was designed to manage.
Now, I need to show you the modern numbers. And I want to be honest, this is the section I found most difficult to research because the numbers are so large they start to feel fake. They are not fake. Every figure I am about to give you comes from a published government source or a peer-reviewed study. The United States Federal Register, the document where all new federal regulations are published, started in 1936 with 2,620 pages. In 2024, it hit an all-time record of 106,109 pages. That is a 40 times increase. That single year contained 3,248 final rules, which works out to 19 new regulations for every one law that Congress actually passed. If you wanted to read the 2024 Federal Register at a normal pace of 250 words per minute, it would take you 44 work weeks. That is almost an entire working year of doing nothing but reading regulations. And then the next year's batch would arrive. The Code of Federal Regulations, the accumulated body of all current rules, grew from 22,877 pages in 1960 to approximately 190,000 pages today. The Maka Center at George Mason University counts the binding restrictions in that code. The words like shall and must and may not. And the number has crossed 1.1 million. If you add state regulatory codes, you get 416 million words and over 6 million restrictions. Reading them all would take 11 and a half years.
The US tax code started as 27 pages of law in 1913. A one-page form rates of 1 to 7%. Today, the comprehensive tax reference a professional actually needs spans approximately 70,000 pages containing 4 million words. That is five times the Bible, seven times war and peace. Americans spend somewhere between 6 and a half, and 7.9 billion hours per year on tax compliance, equivalent to about 3.8 million full-time workers at a cost estimated between 413 and 546 billion dollars annually.
That is not a typing error. That is roughly the GDP of Norway spent every year on the act of figuring out what you owe. And the financial regulation comparison is I think the single most damning data point in this entire story. The Glass-Steagall Act of 1933 separated commercial and investment banking in 37 pages. 37. It included a personal threat of 5 years imprisonment for violators and it worked for 66 years. The Dodd-Frank Act of 2010 attempted to address a similar category of systemic financial risk. It runs 848 pages, 23 times longer, and required agencies to write nearly 400 new implementing regulations. By mid 2013, those implementing rules had generated 13,789 additional pages of regulatory text. 42 words of regulation for every one word of the original law. Over 60% of the implementation deadlines were missed.
That is the curve. You are looking at the curve right now in your own country, in your own tax return, in your own health care system where the number of billing codes jumped from 14,000 under ICD 9 to over 155,000 under ICD 10, which includes a specific code for being pecked by a chicken and a separate code for a subsequent encounter with a flaming pair of water skis. I wish I were joking. The code is V901.07 XD. Look it up. And the healthcare numbers just keep getting worse the deeper you dig. Between 1975 and 2010, the number of physicians in the United States grew by about 150%. The number of healthcare administrators grew by an estimated 3,200%. The current ratio is approximately 10 administrators for every one doctor. Administrative costs in US healthcare consume roughly 1 trillion annually, somewhere between 15 and 30% of total national health spending. Hospital administrative expenditures hit $687 billion in 2023 versus 346 billion for direct patient care. That is a 2:1 ratio. Hospitals in America now spend twice as much on administration as they do on actually treating patients. The United States spends $1,500 per person per year on healthcare administration. Germany, the next highest among developed nations, spends $306.
Let that difference sit for a moment. And doctors, the people this system supposedly exists to support, complete an average of 39 pre-authorizations per week, spending 13 hours of their time on paperwork for insurance companies. 93% report it delays care. 29% report the system has directly caused a serious adverse event for a patient. The complexity built to manage health care is now actively preventing healthcare.
I want to give you one more example because I think it shows the curve operating at the corporate scale and it involves a company you know: General Electric. At its peak under Jack Welch, GE reached a market capitalization of roughly $600 billion, the most valuable company in America. Welch grew it 30fold through acquisitions across wildly diverse sectors: jet engines, insurance, NBC, plastics, credit cards, synthetic derivatives. GE Capital became in effect an unregulated bank with a light bulb logo. The stock eventually lost $540 billion in value, a sum larger than the GDP of Belgium. In 2018, GE was removed from the Dow Jones Industrial Average after continuous membership since 1896. By 2021, it was broken into three separate companies. A Yale analysis called it a massive management mistake.
An activist investor called it the biggest failure of corporate governance in corporate history. And I would argue it is neither of those things. It is Tainter's curve playing out inside a single organization across a single human lifetime. GE did not fail because of bad management. It failed because the complexity of managing that many unrelated businesses in that many unrelated markets exceeded the capacity of any hierarchy to coordinate them. Bayam would have predicted it from the math alone.
And then there is software which might be the purest expression of the complexity curve operating in real time. Windows 1.0 in 1985 contained approximately 5,000 lines of code. Windows XP in 2001 had around 40 million. Windows 11 is estimated at 60 to 100 million lines. That is a potential 20,000fold increase in 40 years. And the consequences track the curve with mathematical precision. The Standish Group's Chaos Reports found that in 1994 only 16.2% of software projects were delivered on time and on budget. 31% were cancelled outright. Budget overruns averaged 189%. By 2020, the success rate had clawed its way up to 31%. Which is still a catastrophic failure rate for a mature industry. Projects exceeding $10 million are 10 times more likely to be cancelled than those under a million. The healthcare.gov launch in October 2013 is the case study that haunts everyone in the field. Original budget $93.7 million. Final cost over $1.7 billion. 47 private contractors with no single individual in charge. Stress tests the day before launch showed the side failed with only 1,100 simultaneous users against an expected 50 to 60,000. A McKenzie review delivered months earlier had flagged more than a dozen critical risks. The steering committee did not act. They launched anyway.
Over the prior decade, 94% of large federal IT projects had been unsuccessful. And nobody stopped to ask whether maybe, just maybe, the systems we are building have crossed the complexity threshold beyond which human institutions can manage them.
And here is where I need to introduce the man who saw what was coming and was called crazy for it. Peter Turchin. Born in the Soviet Union in 1957, his father Valentine Turchin was a physicist and dissident who was expelled from the USSR in 1977. Peter studied biology, earned a PhD in zoology from Duke, and spent years modeling the population dynamics of beetles and lemmings before asking a question that changed his life. Do human societies follow the same mathematical cycles as animal populations? In February 2010, he published a letter in Nature, one of the most prestigious scientific journals on Earth, predicting that the coming decade would bring a period of growing instability in the United States. He identified the specific drivers. Stagnating real wages, an expanding wealth gap, an overproduction of people with advanced degrees competing for a fixed number of élite positions, exploding public debt. He gave a date. He said instability would peak around 2020. Nobody believed him. He told Time magazine in 2020, quote, "They had no reason to believe I was not crazy." End quote. Then came Occupy Wall Street in 2011. The rise of political populism across the West, Brexit, the George Floyd protests, January 6th, 2021.
Every single event fell within his predicted window. His computational model published in the book Ages of Discord in 2016 had identified roughly 50-year cycles of American instability with peaks around 1870, 1920, 1970 and projecting forward to 2020. The model hit its target.
Turchin's concept of élite overproduction is I think one of the most important ideas nobody talks about. When a society produces more aspiring élites than it can absorb into the existing power structure, the surplus becomes what he calls counter élites. People with the education and ambition to lead but no path to power. They mobilize popular discontent. They fragment political consensus. They turn the machinery of governance into a battlefield. American households worth 10 million or more grew from 66,000 in 1983 to 350,000 in 2010.
Francis Bacon warned about this in the 17th century. Sedition follows, he wrote, when more are bred scholars than preferment can take off. 400 years ago, and we still have not listened. Turchin built something called the Seshat Global History Data Bank. Nearly 300,000 records linked to over 400 historical policies from the Neolithic to the Industrial Revolution, coded across 1,500 variables. This is not speculation dressed up as science. This is the largest quantitative history data set ever assembled. And what it shows across every civilization in the database is the same pattern Tainter found in his 17 case studies. A 2017 paper in the proceedings of the National Academy of Sciences used Seshat data to identify a single dimension of complexity that structures global variation in human social organization.
Civilizations do not develop in random directions. They climb the same axis and they fall off the same edge. Turchin warns that the instability period is not over. Historical data across multiple civilizations show that phases of enhanced instability typically last 10 to 15 years. If his 2010 prediction was correct and every marker suggests it was, then we are somewhere around year 14 or 15, which could mean the worst is behind us. Or it could mean we are in the final escalation before the peak. The data cannot distinguish between those two scenarios from inside the window. You would need to wait until it is over to know which one you were in. And by then, of course, the answer would be academic.
I want to pause here and say something directly because I think it matters. I am not a doomed prophet. I did not go into this research looking for evidence that civilization is ending. I went into it because I was curious about a graph in a 37-year-old anthropology book. But the convergence of evidence across these four completely independent research programs, Tainter's Archaeology, Bayam's physics, Turchin's ecology and history, and Olsen's economics is something I cannot explain away. They did not collaborate. They did not side each other, at least not initially. They started from different continents, different centuries, different academic disciplines and they arrived at the same place.
But here is what made me lose sleep. And I am not being dramatic. I genuinely had trouble sleeping after I found this paper. In 2025, two researchers named Kan Ho and Hoa published a mathematical proof on Arksive that civilizational type collapse behaves as a first order phase transition. I need you to understand what that means because it is the difference between a slow decline and a light switch. A second order phase transition is gradual. Think of a magnet being slowly heated. It loses its magnetism progressively. There are warning signs. The system slows down. Fluctuations increase. You can see it coming. A first order phase transition is not that. A first order phase transition is water freezing. Liquid one instant, solid the next. A bridge holding and then not holding. No gradual weakening. No creak before the snap. The system performs its function right up until the threshold and then it does not.
Kan and Hoa derived an exact threshold formula: alpha subscript c equals 1 / (1 minus beta) where alpha is the feedback amplification rate and beta is the adaptive coupling of the system. Below the threshold, the system self-corrects. Above it, sudden irreversible collapse. And for systems with power law feedback dynamics, which include institutional dynamics, market competition, and natural selection, the standard early warning framework, the one scientists use to detect approaching tipping points, fails completely. There are no warning signals. The standard indicators that work for second order transitions produce no useful information whatsoever for first order ones. The system looks stable at every point on the curve including the last one.
I want to be careful here because I know how this sounds and I do not want to be irresponsible with this. There are real criticisms of every researcher I have mentioned. Tainter's curve is conceptual, not precisely calibrated. Turchin's model has been accused of being too linear. Bayam's work on civilizational complexity is harder to test than his food price or violence models. And Kan and Ho's paper is a theoretical proof about a class of systems, not a direct measurement of any specific civilization. But here is what I cannot get past. Four independent research traditions, archaeology, physics, ecology, and economics. Starting from completely different premises, using completely different methods, studying completely different data sets, all converge on the same conclusion: there is a threshold. The threshold is generated internally by the problem-solving apparatus itself. And the dynamics of the threshold suggests that crossing it is sudden, not gradual.
Tainter made one more observation that I have not been able to shake. He noted that in the ancient world, when a civilization collapsed, its neighbors survived. Rome fell but Persia continued. The Maya lowlands emptied but the highlands adapted. Collapse was local because civilizations were local. Modern civilization is not local. The supply chains are global. The financial systems are interconnected. The communication networks are instantaneous. No single nation can collapse independently. Tainter wrote, "Because it would simply be absorbed by neighboring competitors. Therefore, when the threshold is crossed, world civilization will collapse as a whole."
A 2023 study in PNAS by Martin Scheffer found that pre-modern states had a median lifespan of roughly 200 years. They become more vulnerable to collapse as they age. Not because of any single cause, but because accumulated complexity creates brittleness. The United States is 250 years old. The European Union's core institutions are 70 years into their experiment. Both are in the zone where Sheffer's data says vulnerability peaks. And Turchin says the instability phase is not over. Historical data shows these periods last 10 to 15 years. If his model is right, and it has been right about everything else so far, we are somewhere in the middle.
The real question is not whether the pattern is real. The evidence for that spans 5,000 years and 17 civilizations and the largest quantitative history database ever built. The real question is the one the math itself closes the door on. Can you tell where you are on the curve before you cross the threshold? And the answer according to the most recent proof is no. Not because we have not looked hard enough, not because we need better data or smarter models, but because the mathematics of first order phase transitions do not produce warning signals. The bridge feels solid at every point, including the last point. The absence of a warning signal is built into the physics itself.
You interact with the curve every day. Every time you file a tax return. Every time you wait for an insurance company to authorize a procedure your doctor already approved. Every time a software update breaks something that was working. Every time you sit in a meeting that exists to plan another meeting. That is not bureaucratic inconvenience. That is the texture of a civilization on the right side of Tainter's curve, where each additional unit of complexity costs more than it returns. And if the math is right, and four independent research programs say it is, then the feeling of solidity you experience right now, the sense that the systems around you are functioning, that tomorrow will resemble today, tells you nothing. The math predicts you would feel exactly this way at every single point on the curve, including the last one.