An Intermediate Retrospective on Ages of Discord

Peter Turchin


Join 36.9K other subscribers

The publisher of the Japanese translation of Ages of Discord (about to be published) requested a foreword, which presented me with a welcome opportunity to review how this book has fared, and to place it within my overall—long-term—research program attempting to understand the causes of recurrent waves of social and political instability that afflict all complex human societies organized as states. What follows is a lightly edited version of this foreword.

The story begins more than 20 years ago.  At that time, my colleagues and I had made substantial strides in understanding past crises—the Age of Revolutions during the long nineteenth century, the General Crisis of the seventeenth century, the late Medieval crisis. Our research also reached back in history to the crises of Antiquity. This work was eventually summarized in Secular Cycles, co-authored with Sergey Nefedov and published by Princeton University Press in 2009.

But even before the publication of Secular Cycles, as I gave research talks in various academic departments on this topic, I was almost invariably asked: so, where are we in the cycle? For awhile I resisted such inquiries, answering that I wished to keep my focus on past societies in order to avoid entering the minefield of contemporary contentious politics. But at some point repeated questions reached a critical mass. I decided to investigate how well the theory developed for pre-industrial states and empires would fare when applied to contemporary societies, which have clearly evolved a lot during the past two centuries. A natural focus of my investigation was the United States, a society which I knew from the inside.

Collecting the data needed for a structural-demographic “diagnosis” is a lot of work, and it took a few years before the general pattern started to emerge. But what I saw, frankly, shocked me. It turned out that general structural-demographic theory was quite applicable to the United States—after making relatively obvious adjustments, mostly taking into account the technological effects of the Industrial Revolution. But what was disturbing was that the data indicated that we were well along the road to political disintegration. The structural factors undermining social stability in past societies—popular immiseration, elite overproduction, and state fragility—were all trending in negative directions in the early twenty first century America.

In January 2010 a top science journal, Nature, asked “a selection of leading researchers and policy-makers where their fields [would] be ten years from now.” I decided to use this opportunity to publish a scientific prediction stemming from this structural analysis. In my “2020 Vision” I wrote, “The next decade is likely to be a period of growing instability in the United States and western Europe” and then briefly explained the empirical basis for this forecast. My goal was not to predict the future—I don’t believe that the future can be predicted with any accuracy. Instead, I aimed to subject the structural-demographic theory to a rigorous empirical test. If this prediction turned out to be incorrect, then subsequent analysis would tell us how the theory needs to be improved.

Unfortunately, the prediction turned to be disastrously correct. Of course, nobody in 2010 could know about the 2016 presidential election, won by Donald Trump, the COVID-19 pandemic, or the death of George Floyd on May 25, 2020. Such “triggers” of instability are inherently unpredictable. But at the more fundamental level, the structural drivers of instability continued to develop in the ways that were predicted by the theory. As I gave talks at academic departments between 2010 and 2020, the final slide of my presentation stated that we were still on track for a 2020 crisis.

In early 2020 my colleague Andrey Korotayev and I revisited the prediction made ten years before, and looked at various quantitative measures of instability in major Western countries. We discovered that incidence of anti-government demonstrations and violent riots had been declining prior to 2010, but this trend was reversed after 2010.

We submitted the article that presented this analysis to a scientific journal in the Spring of 2020. Then, as the article was undergoing review, the surging COVID-19 pandemic and the death of George Floyd at the hands of the Minneapolis police delivered a tremendous shock to the U.S. political system, triggering an explosion of nationwide, months-long series of urban protests. The “Turbulent Twenties” have arrived.

How does Ages of Discord fit in this story? The opinion piece, published in Nature in 2010, was very short and could only sketch out the causal factors pushing America into crisis. In the years after 2010 I also published several academic papers that fleshed out the details. But in order to be convincing, I needed to bring together different strands of this research. In particular, I had to integrate theory (and mathematical models that made it concrete) with data on long-term social, economic, demographic, and political dynamics in America, 1780–2010. This required a book-length treatment and the result was Ages of Discord.

As you read Ages of Discord, keep in mind that it was published in 2016, months before Donald Trump was elected U.S. president and years before the start of the Turbulent Twenties. The book focuses on structural causes—hidden forces that have been pushing America to the brink of civil war (and now maybe beyond). While no scientific theory is 100 percent correct, the disastrous success of the prediction made in 2010 tells us that the structural-demographic model is a useful way to look beyond surface events and to gain a deep understanding of what is happening in America. Keep in mind, also, that the theory is general. All complex societies organized as states are vulnerable to the three forces of instability—immiseration, elite overproduction, and the decline of state power. This may sound like a pessimistic conclusion, but my take is different. As solid scientific understanding of why ages of discord happen becomes better, we should be able to “engineer” our societies to avoid the extreme, most negative outcomes.

Notify of
Most Voted
Newest Oldest
Inline Feedbacks
View all comments
Raymond Reichelt

Good post Peter! I am reminded of a statement in Tolkien’s Lord of the Rings where the wizard Gandalf tells Frodo that while we cannot choose the times we live in, what matters is how we use the time that we are given

Sanford Zane Meschkow

Being a life-long science fiction fan I can’t help being reminded of Isaac Asimov’s FOUNDATION series and the science of psychohistory developed by Hari Seldon. He predicted that the galactic empire would crash and started a secret project to shorten the dark age that would follow the breakup of the
empire. Hope you have gotten your version of the Foundation started already. Good luck!

Steffen T. Laursen

Thanks for this blog post. Is a followup volume in the plans?

Best, Steffen

Giovanni Venturi

Very nice! Reminds me of the Foundation series of isaac Asimov. The whole approach can be studied through mathematical models.. nonlinear chaotic approaches..

Tove Karlsson

I most of all miss a micro level explanation to the results of Ages of Discord. When structural-demographic theory was only applied to pre-modern, agricultural societies it looked like rather simple psychology lay behind it: When people risk seeing their children starving to death, they get agitated and prone to violence. Also, when upper class people who were only taught how to fight wars become redundant, they invent new causes to fight wars. That is rather straightforward.

When the same theory is applied to modern times, the underlying psychological assumptions become much less clear. For some reason, it seems, people react a bit similarly to crowding in modern times as in pre-modern times. But why do they? It is not exactly like the ghost of starvation is grinning in their faces.

Have you, or anybody else, made any attempt to explain the results of Ages of Discord on a micro level? I mean, has anyone made any attempt to develop a theory of social psychology that matches the observations and predictions of Ages of Discord?


Well, maybe it has something to do with Maslow’s hierarchy of needs? If starvation is removed as a potential destabilising factor, you just climb a step upwards on the pyramid until you find a limiting factor. It could be lack access to adequate housing, or maybe lack of access to suitable romantic partners or status or whatever. Society gets too crowded and then cannot sufficiently meet people’s needs at whatever the current level of Maslow’s pyramid they are at.

Maybe all that the Industrial Revolution really did was to (temporarily?) allow humanity to climb a step or two on Maslow’s pyramid?

Chris Goble

Tove, didn’t Graeber and Wengrow, in Dawn of Everything, propose a generalized social psychology to societal cycles? They postulated three social landscape wells/factors: 1) sovereignty (state power), 2) knowledge keeping, 3) heroic competition. They didn’t go into any deep reasons why these three social-psych wells might exist, but they did do some “back of the envelope” thinking on it. But, it definitely wasn’t anything close to the quantifiable, falsifiable work that makes Cliodynamics so attractive. If I recall, their only real engagement with elite over production was 1) it could lead to ever growing state power until people just “walked away”, or 2) it could lead to heroic competition and depopulating war cycles.

But, there was no attempt to cross the micro-macro social-psych divide.

Looking at multi-level selection theory, I wonder if phase changes between the “Left” and “Right” don’t portend something deeper. Jordan Peterson’s view of the “Left’s” susceptibility to toxic caring and the “Right’s” susceptibility to toxic ?production? may cross over during elite over-production, giving rise to villainous resource extraction from commoners. From the left crossing right, “I care so much about you people, I as an individual deserve to have no bounds”. From the right crossing left “I am so individually successful, I deserve to have no bounds so I can care for you all via trickle down benefits”.

I think D.S. Wilson’s real early work had a 2×2 matrix similar to that idea. It was in terms of stable self-interest vs group-interest points, based on Hutterite group-interest and Ayn Rand self-interest
(yes the novelist .. it was a humorous test of his theory)

Andrew Wilson

Fascinating, as always, and thank you for the insights. Elite overproduction, to be sure; immiseration, absolutely (especially by Covid-induced lockdowns, shortages and fears); weakening of state power –now, this point is interesting. Here in Portland, Oregon, the state has lost control and is struggling to provide even basic public services. Food, clothing, shelter, education, health care, personal safety: the state of Oregon is failing in all dimensions. In totalitarian surveillance states, on the other hand, digital technology allows them to tighten their grip on the people 24/7/365. The brutal Anschluss of Hong Kong by the PRC does not exactly look like a State which is losing its grasp. Rather the reverse. Comments??? A momentary divergence between free and non-free societies, or, a defining split?

Tim Richardson

Your definition of “crisis“ is uncomfortably vague.

Trump was interesting and Brexit was alarming, but neither one of these events made me miss a meal.

Arguably, the Fed’s decision to raise interest rates has a greater short-term and long-term impact on the day-to-day lives of many ordinary Americans.

Much more than any of these so-called “crises“.

Paolo Ghirri

Sorry for my english. I read well, but write not so well. You have to think at a chain of “Crisis”, step by step A lead to B, B to C C….Z there Is not direct link between A and Z but without A you don’t have Z. And every step the instability of the sistema Is higher. Trump Is one step (if i have to choice the overall First step i keep Reagan) Fed raise just another.

Z Cademy

Trump is a data point.
So was Ronald Reagan, hanging chads, “de-fund the police”, SBF, and a whole army of other things we could probably both name. [Over a couple of good drinks, in a cozy pub.]

A few data points don’t make a crisis. A long list of them, coming faster and faster and affecting more and more of the general population, and you have a crisis.

Aidan Barrett

Just as you predicted and Strauss and Howe in the early 1990s!

Tim Richardson

I did, I really enjoyed UltraSociety. Peter even says in the beginning that he wrote it just for me; untrained, but strongly opinionated laypersons 🙂

I just hear everybody from Peter Turchin to Ray Dalio to Peter Ziehan predicting the end of the world.

The end for who? Certainly not the average American. We are currently fighting proxy wars in various parts of the world and fighting a real World War in the Ukraine.

Lots of people are in crisis, right now. That’s why I feel like we need a better definition of crisis.

Are we talking 2008 recessionary-style crisis? 1930s depression-style crisis? 1200 BC Iron Age collapse style crisis?

The Fed wants to raise interest rates until 3 million more Americans lose their jobs. For every 1% increase in the unemployment rate, 40,000 people die (Brad Pitt, The Big Short). These people are having a really bad day, too. But, not on the scale of the Ukrainians.

The future is already here, it’s just not evenly distributed.

As long as the American dollar remains the world reserve currency, the American government can inflate our way out of a monetary crisis. The consequences of American inflation will not be felt in America. They will be felt in other countries whose currency does not hold reserve status.

Peter included a link to the Crisis DB which I clicked on. It’s a technical paper and I’m eagerly awaiting his book for the layperson.

Z Cademy

I don’t think Prof Peter is predicting the end of the world; rather that we are at the nadir of the current cycle. As the cycle turns, things will get better.

As I see it, the key question is, what length of cycle are we in? If this is a basic 100-ish year cycle, things should be getting better soon, with the next peak coming around 2070. But if I remember, there are longer cycles too, where things get *slightly* better on the 100 year cycle, but not truly good for several centuries.

Neither is going to be the end of the world, unless you have a very short time horizon, or if we actually do mess up so badly as to make ourselves extinct.

FYI, I’m another “untrained, but strongly opinionated laypersons”. B.A. Anthropology, SF book fanatic, and general deep thinker, but not trained in any of this.

Loren Petrich

Elsewhere in this site, our host has Books > Ages of Discord with supporting data.

The US had a peak in the 1820’s, making its Era of Good Feelings very aptly named. It then had a trough in the 1890’s, toward the end of the Gilded Age, gilded meaning only outwardly looking golden. It had another peak in the 1950’s, the Eisenhower Era, and it’s now headed to another trough. So with a half-period of 60 – 70 years, we are in another trough.

Loren Petrich – lists various other cycles.

Schlesinger liberal/conservative cycle: min/median/max lengths 8-12-18, 12-16-32 years. The Turchin peaks and troughs are in conservative phases.

Samuel Huntington’s creedal-passion periods: roughly 15 years each 60 to 70 years, in Schlesinger liberal periods: Revolution, Jackson, Progressive, and Sixties. So the US is due for another one.

Party systems: the US has been dominated by two parties despite its founders not wanting them. In each one, each party has a characteristic platform and constituencies. The first one started in a Schlesinger conservative period (1796), the second to fifth ones in conservative-to-liberal transitions (1826-28, 1854-56, 1894-96, 1930-32), and the sixth and current one in a liberal-to-conservative transition (1974-80). So he US is due for the emergence of a new party system.

Loren Petrich

There are other proposed cycles of US history, and some of them line up well with Peter Turchin’s long cycle.

Schlesinger liberal-conservative cycle, an alternation between reform and stagnation, public purpose and private interest. They have lengths (min-median-max) 8-12-18 and 12-16-32, and the US is currently in a sort of Gilded Age II, one that is overdue for ending. Each Turchin peak and trough was in a conservative period and was soon followed by a liberal period.

Huntington creedal-passion bursts, about 15 years out of 60 – 70 years. Revolution, Jackson, Progressive, Sixties, and the US is now due for another one. Each one just after a Turchin peak or trough.

Party systems. The US has gone through six of them so far, with the first one emerging in a Schlesinger conservative period, the second to fifth ones emerging in conservative-to-liberal transitions, and the sixth and present one emerging in a liberal-to-conservative one. Each system lasts for about 40 years, so the US is due for the emergence of another one. The second and fourth ones began just after a peak and a trough, and the sixth one somewhat in between.

Scott Taddiken

To Tim Richardson, Peter is actually quite precise with his definition of crisis. Or I should say “definitions” because like all good statisticians or econometricians he reruns the analysis with multiple plausible “targets” “y-values” whatever you call it.

I don’t think you’ve read any of his books or more rigorous articles?

Dennis Smithson

My recollection from Ages of Discord is that you mention a lack of clarity as to why the elites downsized during the mid-20th century (in the US). Am I recalling that correctly? Any new insights on that?


Do you think there is every going to be enough political will to engineer a society out of crises if the theory or receipt is 100% correct?

Frances Atherton

This is an interesting article! I’m always interested in studying the causes of social and political instability and this article gives a good overview of how our research has been progressing over the years.

Loren Petrich

Thank you, Peter Turchin and your colleagues, for making your CrisisDB data available. I decided to look for correlations between different types of calamities, and I did principal components analysis on the data.

There is only one component that stood out from the rest: one that is approximately the total number of calamities in a crisis. So I subtracted that one out by using the fraction of crises that have each type of calamity.

After doing so, I found two components that weakly stand out from the others, as I did before that subtraction.

The first one has a negative correlation between two sets of calamities. The first set is fragmentation, downward mobility, epidemic, decline and collapse, while the second one is uprising, civil war, and deposing and assassinating the leaders. The first one is more or less social weakening, and the second one is rebellion against existing leaders.

The second one also has a negative correlation between two sets. The first set is decline, epidemic, uprising, and civil war, and the second set fragmentation, conquest, and deposing of leaders. So it’s weakening of hold on territory vs. other types of calamities.

I plotted the position of each crisis by those two components, and I found a roughly disk-shape scatter without very evident patterns.


” All complex societies organized as states are vulnerable to the three forces of instability—immiseration, elite overproduction, and the decline of state power.”

I noticed you’re consistently downplaying POPULATION GROWTH as a major source of instability, even though it’s a fully one half of the structural-demographic theory. It is also a half that is very solidly rooted in natural sciences (such as biology, ecology, etc.), while “elite overproduction” is pretty handwavy.

Then, of course, in the polity in question population growth is driven entirely by immigration, and anti-immigration sentiments are persecuted by the elites (who presumably want cheaper labor, among other things). That, and your personal background, might explain your reluctance to point it out. I HOPE your own “globalist” elite status doesn’t have anything to do with it, though.

  1. Home
  2. /
  3. Cliodynamica
  4. /
  5. Regular Posts
  6. /
  7. An Intermediate Retrospective on...

© Peter Turchin 2023 All rights reserved

Privacy Policy