What Complexity Science Reveals About Civilizations

BLUF: Recent decades have witnessed the emergence of quantitative approaches to understanding why complex societies fail. Researchers have moved beyond historical narrative to mathematical modeling, with mixed results. While scholars like Joseph Tainter, Peter Turchin, and others have advanced our understanding of civilizational dynamics, their theories remain contested, their predictions probabilistic rather than deterministic, and the gap between elegant models and messy historical reality remains formidable.

Introduction: The Quest for Pattern

In the spring of 2010, Nature magazine published an unusual item in its "Correspondence" section. Peter Turchin, a mathematical biologist at the University of Connecticut, had submitted a brief prediction: the United States would experience significant political instability around 2020, driven not by any specific event or actor, but by mathematical patterns embedded in centuries of historical data.

Ten years later, amid pandemic, riots, and polarization, Turchin's forecast seemed remarkably prescient. Journalists and public intellectuals seized on the prediction as evidence that human societies follow hidden mathematical laws—laws as predictable as physics, laws that might foretell our own civilization's fate.

But the story of how we came to believe this, and what the evidence actually supports, is far more textured and uncertain than the popular narrative suggests.

The Tainter Framework: Complexity and Diminishing Returns

The intellectual foundation rests on a 1988 book that few in the general public have read but many cite: Joseph Tainter's The Collapse of Complex Societies.

Tainter, an archaeologist then at the USDA Forest Service and now a professor at Utah State University, examined why seventeen civilizations across five thousand years experienced rapid, substantial losses of sociopolitical complexity—the Roman Empire, Maya civilization, the Chaco culture, and others. His question was deceptively simple: why did these societies, at the height of their sophistication, suddenly fragment?

His answer centered on economics. As societies grow more complex—adding bureaucratic layers, military infrastructure, and administrative machinery to solve problems—each new increment of complexity yields smaller benefits relative to its costs. Eventually, a society invests enormous resources merely to maintain its existing complexity. When external stresses arrive (invasion, crop failure, climate shift), the society cannot afford the additional complexity required to respond. It collapses.

The mechanism is drawn from economics: the law of diminishing marginal returns.

Tainter illustrated this with the Western Roman Empire. Agricultural output declined as population grew. The empire's response was rational: conquer new territories to access surplus grain, metals, and slaves. But as the empire expanded, the cost of maintaining communications, garrisons, and administration grew proportionally. The empire split in two. The Western half, chronically stressed, fragmented into smaller polities. By the 5th century, that bureaucratic superstructure—which had seemed unshakeable—simply dissolved.

The Graph That Didn't Say What People Think It Said

In his 1988 book, Tainter presented a graph showing the relationship between societal investment in problem-solving (the x-axis) and the marginal return on that investment (the y-axis). The curve descends: more investment, smaller returns.

This image has become iconic in collapse discourse. It is often presented as a universal law: all civilizations follow this curve. All hit the wall at maximum complexity.

But this is a misreading. Tainter himself was careful to note that the curve describes a tendency, not an iron law. Not all societies experience declining returns in the same way or at the same speed. Innovation, new energy sources, or technological breakthroughs can reset the curve. Tainter acknowledged this explicitly, writing: "The fact that such examples can be compiled does not mean that all economic trends follow the same curve, nor that socioeconomic processes in complex societies follow only the law of diminishing returns."

The theory is powerful but not deterministic. It explains why societies might collapse, not that they inevitably will.

Peter Turchin and the Science of History

Where Tainter was primarily an archaeologist using economic reasoning, Peter Turchin represents a newer breed: the mathematical historian. His toolkit includes computational modeling, time-series data analysis, and techniques borrowed from physics and evolutionary biology.

Turchin's central claim is that history is not random. Beneath the chaos of events—the decisions of leaders, the accidents of circumstance, the contingencies that historians emphasize—there are structural patterns that recur across centuries and continents.

In 2010, he published a brief article in Nature titled "Political Instability May Play a Role." Based on a computational model incorporating data from 1780 to 2010, Turchin identified a pattern: the United States experienced waves of instability roughly every 50 years, with peaks around 1870, 1920, and 1970. Mathematically, the next peak should arrive around 2020.

The model synthesized multiple inputs: real wage trends, wealth inequality, competition for elite positions (measured by the ratio of lawyers and PhDs to available positions), and state fiscal capacity. These variables, Turchin argued, capture the structural pressures that precede violence and social upheaval. He called the aggregate measure the Political Stress Index (PSI).

A Prediction Tested

What made Turchin's work notable was not merely that it proposed a theory, but that it made a specific, testable prediction before the fact. In 2020, Turchin and collaborator Andrey Korotayev published a retrospective assessment in PLOS One, examining whether the prediction had held.

The results were nuanced. They found that "measures of socio-political instability such as anti-government demonstrations and riots increased dramatically during the 2010–2020 decade" in the United States, United Kingdom, and several Western European countries. The broad pattern matched. But the analysis also noted that the 2010 prediction had not specified which year within the decade, and the model's precision was lower than initially hoped.

Importantly, Turchin has never claimed this represents a law of nature. Rather, it is a probabilistic model: given these structural conditions, instability becomes more likely, though not certain. Leadership, institutional capacity, and contingent events still matter enormously.

Yet the popular narrative often flattens this nuance. Headlines proclaimed that Turchin had "predicted 2020" with eerie accuracy. The claim that he was "called crazy until every prediction came true" is an overstatement. His 2010 article was published in one of the world's most prestigious journals; his subsequent work has been peer-reviewed and cited by academics across multiple disciplines. What changed was not scientific acceptance, but public awareness.

Yaneer Bar-Yam and the Food Price Signal

In late 2010, as global food prices spiked, a group of researchers at the New England Complex Systems Institute (NECSI) in Cambridge, Massachusetts began tracking the connection between commodity price volatility and social unrest.

The lead figure was Yaneer Bar-Yam, a physicist-turned-complexity-scientist whose work spans networks, epidemiology, and economic systems. In December 2010, Bar-Yam's team analyzed historical data linking food price surges to riots and protests in 2007-2008 and earlier periods. They identified what they described as a threshold: when the UN Food and Agriculture Organization's Food Price Index exceeded 210, social conflict became likely in food-insecure countries.

On December 13, 2010, Bar-Yam sent this analysis to the U.S. government. Four days later, Mohamed Bouazizi, a Tunisian street vendor, set himself on fire in an act of protest that catalyzed the Arab Spring.

The timing was striking. Within weeks, the revolutionary narrative crystallized around food price volatility: skyrocketing grain costs had pushed the poor over the edge. Bar-Yam's model seemed to have caught a genuine causal mechanism.

The Complications

But subsequent scholarship has complicated this picture significantly.

Food policy analysts and economists, notably those at the Stimson Center, examined whether high international food prices actually translated to high local prices in Tunisia, the initial site of regime change. The evidence proved ambiguous. Tunisia's government had implemented price controls and subsidies that, according to FAO reports, kept domestic food prices relatively stable even as international prices surged. One 2012 analysis found that Tunisia and Algeria had the lowest "pass-through coefficients" in the region—international price shocks barely affected what consumers paid.

The controversy was not whether food prices matter—they clearly do, especially in countries dependent on grain imports. Rather, it concerns whether international price spikes were the primary trigger for Tunisia's specific uprising, or whether structural grievances (unemployment, corruption, political repression) were more fundamental, with food prices as one of many contributing factors.

Bar-Yam and colleagues responded to critics, arguing that even modest pass-through effects matter at the margin, and that their analysis pointed to broader regional vulnerability. The debate remains unresolved in the literature, with economists divided on the relative weight of food prices versus other drivers of the Arab Spring.

What is clear: Bar-Yam's prediction was not a simple "4-day forecast" of events. Rather, it was an identification of rising systemic risk based on threshold analysis. Whether that risk materializes depends on political context, state capacity, and grievance levels—variables not fully captured in price data alone.

Mancur Olson and Institutional Sclerosis

A third intellectual strand comes from political economy. Mancur Olson (1932-1998), an economist at the University of Maryland, published The Rise and Decline of Nations in 1982—predating both Tainter's archaeological synthesis and Turchin's computational approaches, but influential in shaping how economists think about long-term growth and stagnation.

Olson's central insight was counterintuitive: societies that enjoy long periods of political stability tend to decline, while those disrupted by war or revolution sometimes surge ahead. Why?

His answer: stability allows interest groups to accumulate. Unions, industry associations, professional cartels, lobbying blocs—these "distributional coalitions" form over decades to protect their members' economic rents. Once entrenched, they resist change. They use political power to fix prices, prevent entry, and extract wealth without creating it. The result is institutional sclerosis: the system moves more slowly, makes fewer new investments, and loses dynamism.

Post-war Germany and Japan, having lost their entrenched interest groups through military defeat, rebounded with vigor. Britain and the United States, with stable institutions intact, accumulated rigidities and stagnated in relative terms (though they remained wealthy).

The Persistent Problem

Olson's theory remains influential in public choice economics and institutional analysis. It provides a parsimonious explanation for why stable, wealthy democracies sometimes underperform relative to expectations, and why crises can reset growth.

But it also has limitations. Olson struggled to explain why the United States, despite centuries of stability and heavy interest-group density, continued to dominate economically through the late 20th century. Later scholars have noted that some stable societies innovate successfully (Switzerland, Singapore) while some war-torn nations remain poor despite institutional disruption.

Olson's theory describes a real mechanism—the political economy of collective action and rent-seeking—but it is neither universal nor deterministic. Like Tainter's framework, it illuminates one important force among many.

Recent Developments: Complexity, Phase Transitions, and Uncertainty

Beginning in the early 2020s, new work has attempted to bridge complexity science and historical collapse studies. Researchers have explored whether civilizational dynamics exhibit the properties of complex systems: tipping points, critical thresholds, and sudden phase transitions.

In 2025, a paper by Khanh and Hoa on the arXiv preprint server argued that entropy collapse in feedback-amplified systems (including AI systems, economic institutions, and biological populations) behaves as a first-order phase transition—a sudden shift rather than a gradual decline. The authors drew on concepts from physics to argue that systems optimizing for specific metrics rather than holistic function will eventually lose adaptability catastrophically.

The work is mathematically rigorous and published in an open forum, but it has not yet undergone traditional peer review. Moreover, the applicability of entropy collapse theory to historical civilizations remains speculative. The paper does not claim to predict civilizational collapse in any specific timeframe; rather, it argues that such collapse, if it occurs, would be sudden rather than gradual.

This is an important distinction. A system exhibiting first-order phase transition behavior does not announce its collapse in advance. The bridge holds until it doesn't. But the absence of early warning signals does not mean the system is in equilibrium; it means that certain types of critical failure leave no visible precursors. This is mathematically interesting but offers limited practical guidance for policy or planning.

What We Actually Know: And What We Don't

Four decades of research from Tainter, Turchin, Bar-Yam, and others has yielded genuine insights:

Complex societies do experience declining marginal returns on investment in complexity. This is not universal law, but it is a real mechanism that has contributed to historical collapses. The Roman Empire's case is well-documented: managing an empire of 70 million people across three continents became increasingly expensive per unit of benefit.

Structural pressures do accumulate in societies over decades. Inequality, elite competition, state fiscal capacity, and public expectations all shift gradually. Turchin's work demonstrates that these variables correlate with violence and instability across multiple centuries and regions. Again, not deterministic, but genuinely predictive in a probabilistic sense.

Food price volatility does correlate with social unrest, especially in vulnerable populations. Bar-Yam and colleagues identified a real relationship. The complication is that correlation is not causation, and many confounding variables affect both prices and unrest.

Interest groups do accumulate and can impede adaptation. Olson identified a genuine phenomenon in political economy. The question is whether this mechanism is strong enough to predict state failure, or merely one factor among many.

But we also must acknowledge the limits:

We cannot predict when collapse will occur. Models can identify periods of elevated risk, but contingency remains decisive. The United States was under high structural stress in both 2020 and 2024, yet did not collapse. Sometimes societies navigate critical periods. Sometimes they don't.

Universal laws remain elusive. Tainter's framework works well for some cases (Rome, Maya) and less well for others (China's recurring dynasticism, Japan's ability to recover from devastation). Turchin's model fits American history better than other societies. No single theory explains all collapses.

The relationship between complexity and collapse is bidirectional and context-dependent. Complexity can generate fragility (by increasing interdependence and feedback loops), but it can also generate resilience (through redundancy, diversity, and adaptive capacity). Which dominates depends on system design.

Models are simplified representations, not descriptions of reality. All the frameworks discussed here—Tainter's economic model, Turchin's structural-demographic index, Bar-Yam's threshold analysis—are useful simplifications. They illuminate certain aspects while necessarily obscuring others. Applying them to real-world policy requires careful judgment about what the model captures and what it ignores.

The Broader Context: Complexity Science and Policy

What has emerged from thirty years of collapse research is neither apocalyptic prophecy nor reassuring predictability. Instead, there is growing recognition that:

Complex adaptive systems can exhibit critical transitions. This is well-established in ecology, climate science, and network theory. Whether human societies exhibit similar dynamics remains an active area of research.
Early warning of critical transitions is difficult. In some systems, critical transitions are preceded by "flickering" or rising variance. In others, they appear sudden. The mathematical theory suggests that the absence of warning signals does not guarantee stability.
Multiple stress factors can interact. When several structural pressures rise simultaneously—inequality, environmental degradation, state capacity decline—the risk of instability compounds. This is intuitive but hard to quantify precisely.
Institutions matter enormously. Societies with flexible institutions, distributed decision-making, and capacity to adapt tend to navigate stress better than rigid, centralized systems. This is perhaps the most actionable insight from collapse research: the quality of governance and institutional design shapes whether structural stress leads to adaptation or catastrophe.

Case Studies: Recent Applications

Peter Turchin's Retrospective Analysis (2020)

When Turchin and Korotayev examined their 2010 prediction in 2020, they did not simply declare victory. They found that anti-government demonstrations and riots had indeed increased dramatically across the US, UK, and Western Europe during the 2010-2020 decade. The broad pattern held.

But they also noted:

The model had not predicted which specific events would occur
It could not distinguish between violent upheaval and non-violent movements
The precision of year-level prediction remained limited
External shocks (the 2008 financial crisis, COVID-19) had accelerated timelines in ways the structural model alone could not have forecast

In their 2020 assessment, they cautiously extended the prediction into the 2020s, suggesting that structural pressures would remain elevated for years or decades. But they avoided claiming certainty about outcomes.

Complexity Science in the COVID-19 Context

The pandemic offered a natural experiment in how complex societies handle stress. Initial responses by governments proved highly variable. Some societies adapted rapidly (South Korea, Taiwan); others experienced governance failure and excess mortality.

Bar-Yam and the NECSI team applied complexity science frameworks to pandemic response, arguing that centralized, command-and-control approaches were fragile, while distributed, adaptive networks of local authorities performed better. This work is ongoing and remains somewhat preliminary, but it represents an attempt to apply complexity science to real-time policy, not merely retrospective analysis.

Critiques and Open Questions

Skepticism about quantitative collapse science comes from multiple directions:

Historians argue that reducing the fall of Rome or the Maya to mathematical models ignores contingency, human agency, and the specific meanings people attributed to their world. A society doesn't "collapse" until people stop maintaining the system; understanding that psychological and social transition requires narrative, not just metrics.

Economists note that many of the variables used in structural-demographic models (elite overproduction, measured by PhD-to-job ratios) are data-dependent and culturally specific. What counts as an "elite position" in 13th-century England differs from 21st-century America. Models trained on one context may not transfer.

Climate scientists point out that many collapse models underweight environmental variables. While Tainter discusses resource depletion, climate models treating water, soil, and biodiversity as dynamic variables are more recent. Some researchers argue that environmental thresholds, not institutional variables, drive collapse timing.

Complexity theorists themselves remain uncertain about the applicability of phase transition theory to historical societies. The mathematical frameworks work well for systems with few variables and high feedback (lasers, magnetic materials, simple economic models). Historical civilizations are far messier. Whether phase transition metaphors illuminate or mislead remains debated.

Conclusion: Living with Uncertainty

What emerges from forty years of collapse research is not a prophecy of doom or a hidden mathematical law governing history. Instead, we have:

Better understanding of mechanisms that have contributed to past collapses
Recognition that multiple stressors compound
Probabilistic models that identify periods of elevated risk, without predicting specific outcomes
Acknowledgment that institutional quality, adaptive capacity, and leadership matter enormously
Awareness that complex systems can exhibit critical transitions, but predicting when they occur remains difficult

The uncomfortable truth is that we can identify conditions that make societies fragile—high inequality, institutional rigidity, environmental degradation, fiscal stress. We can note that several such stressors are present in contemporary wealthy democracies. But we cannot say with precision whether these pressures will precipitate adaptation, gradual decline, or sudden collapse.

Perhaps this is as close as we can come to a "science of history": not prophecy, but informed humility about what we can predict and what remains contingent on human choice.

Sources and Further Reading

Primary Works

Tainter, Joseph A. The Collapse of Complex Societies. Cambridge University Press, 1988.

Turchin, Peter. "Political Instability May Play a Role." Nature, vol. 463, 2010, p. 608.

Turchin, Peter, and Andrey Korotayev. "The 2010 Structural-Demographic Forecast for the 2010–2020 Decade: A Retrospective Assessment." PLOS One, vol. 15, no. 8, 2020, p. e0237458. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0237458

Olson, Mancur. The Rise and Decline of Nations: Economic Growth, Stagflation, and Social Rigidities. Yale University Press, 1982.

Lagi, Marco, et al. "The Food Crises and Political Instability in North Africa and the Middle East." arXiv, 2011, arXiv:1108.2455v1. https://arxiv.org/abs/1108.2455

Bar-Yam, Yaneer. "Food Crisis and the Dynamics of Global Economic Supply Chains." NECSI Report, 2011. https://necsi.edu/food-crisis

Khanh, Truong Xuan, and Truong Quynh Hoa. "Entropy Collapse: A Universal Failure Mode of Intelligent Systems." arXiv, 2025, arXiv:2512.12381. https://arxiv.org/abs/2512.12381

Scholarly Assessments

"Were High International Food Prices An 'Early Warning' Of The Arab Spring? Probably Not." Stimson Center, 2014. https://www.stimson.org/2014/were-high-international-food-prices-an-early-warning-of-the-arab-spring-probably-not/

Turchin, Peter. "The Science Behind My Forecast for 2020." Cliodynamics: The Journal of Quantitative History and Cultural Evolution, 2020. https://peterturchin.com/the-science-behind-my-forecast-for-2020/

Goldman, Gustavo. "Toward a General Theory of Societal Collapse: A Biophysical Examination of Tainter's Model." arXiv, 2018, arXiv:1810.07056. https://arxiv.org/pdf/1810.07056

Ortmans, Benjamin, et al. "Modeling Long-Term Social Pressure, Political Instability, and Conflict Using Cliodynamic Macrohistory." In Conflict, Governance, and State Fragility, edited by Robert Bates and others, Oxford University Press, 2020.

Historical Context

Gibbon, Edward. The Decline and Fall of the Roman Empire. 1776-1789.

Spengler, Oswald. The Decline of the West. 1918.

Toynbee, Arnold. A Study of History. 1934.

Diamond, Jared. Collapse: How Societies Choose to Fail or Succeed. Viking, 2005.

Contemporary Analysis

Turchin, Peter. Age of Discord: A Demographic Interpretation of American History, 1780-2020. Beresta Books, 2020.

Corey, Robin. "Mancur Olson at the End of History." Slow Boring, 2022. https://www.slowboring.com/p/mancur-olson-end-of-history

Yglesias, Matthew. "The Failure to Build." Slow Boring, 2021-2024. [Series examining institutional sclerosis in American governance]

Complexity Science Frameworks

Bar-Yam, Yaneer. "Concepts: Complexity." New England Complex Systems Institute. https://necsi.edu/

Homer-Dixon, Thomas. The Upside of Down: Catastrophe, Creativity, and the Renewal of Civilization. Island Press, 2006.

Scheffer, Marten. Critical Transitions in Nature and Society. Princeton University Press, 2009.

Key Takeaways

Collapse has patterns, but no universal law. Tainter, Turchin, and others have identified recurring mechanisms (diminishing returns, structural pressures, institutional rigidity) but none predicts collapse with certainty.
Models are tools, not prophecies. Turchin's Political Stress Index offers useful guidance about when societies are fragile, but cannot predict specific events or timeframes precisely.
Multiple stressors compound risk. Societies under simultaneous pressure from inequality, environmental degradation, and institutional rigidity are more vulnerable than those experiencing single stressors.
Institutions remain decisive. How societies organize decision-making, distribute power, and adapt to change matters as much as structural variables. Some highly stressed societies adapt; others collapse.
Humility is warranted. We can identify fragile conditions. We cannot predict collapse with the precision that physics predicts eclipses or chemistry predicts reactions. History remains partly contingent on human choice.

Search This Blog

Science, Time, and Spirit

The Mathematics of Collapse: