The Evolution of Hierarchy



Join 36.9K other subscribers

Two weeks ago I went to a studio in Amherst, MA, to participate in a BBC Forum on Hierarchy. It was broadcasted last week, and you can listen to it here.

I arrived at the studio in plenty of time, but there were inevitable SNAFUs. First, I couldn’t get into the building (not sure why they lock it during the business hours), and for a while nobody was responding to my frantic hammering at the door. Then, I had to drive to a distant parking lot (they did not have a parking space for me near the building) and run back, so I started the program out of breath.

Still, I thought the program went quite well. The format of the show doesn’t force you into reducing the complexity of your ideas to sound bites; instead giving you time to develop your argument in a logical way. The other two panelists were really interesting and articulate people (read about them on the BBC site). It was a pleasure to listen to their perspectives.

In today’s blog I’d like to revisit the main idea I was trying to get across. Nothing can substitute the written word for clarity and conciseness.

For over 90 percent of our evolutionary history humans lived in small-scale groups. Agreeing on a common course of action to achieve collective goals was relatively easy. It is quite possible to come to a consensus in a small group of ten or fewer people simply by having a common discussion, listening to everybody’s concerns, and not allowing anybody to dominate the decision-making process. If such a group finds it impossible to achieve a consensus, it disbands and reforms in a different configuration, or sheds uncooperative individuals.



It is much harder, but still possible to concert a common policy in groups consisting of many tens of individuals; very difficult in groups of hundreds; and simply impossible once you get into thousands. That’s probably one of the fundamental principles of social dynamics (which the Occupiers did not seem to understand).



The only way that large human groups can arrive at a common course of action is by structuring interpersonal connections. Imagine a large human group as a network in which any particular node (a person) has a thick connection to just a few other people, between 4 and 10 of such connections. This is a small enough number so that each person in a pair can know the other quite well, and they can rapidly coordinate their objectives and actions.

So the coordinator at the center of the network speaks to and gets feedback from ten other people. Those ten speak to ten more people each, and so on. It is possible to connect a group of any size using this scheme, by adding extra levels of communication and decision making. For example, a society with one million members would need six levels. We need only 10 levels to connect everybody on Earth!



This is a hierarchical organization in its neutral, informational (or, as my father would say, cybernetic) sense, which should not be confused with ‘hierarchy’ in its negative sense, that is, unequal distribution of power. Biologists are quite used to speaking of hierarchies in the neutral sense (for example, we speak of life being organized in a hierarchical way, from molecules to organelles, to cells, to tissues and organs, to individuals, and to populations or societies).

The problem, of course, is that as soon as you put someone in a central position of a decision-making network, you give them a lot of structural power, and they can, and often will, attempt to convert it into more malignant forms of power: ability to order others around, and acquisition of a disproportionate amount of resources. If there are no effective checks on leaders, then hierarchy breeds inequality. Sociologists even have a name for this, the Iron Law of Oligarchy.

So what happened to humans during the Holocene (which started roughly 12,000 years ago with the end of the Ice Age) is that they spread all over the Earth, and their numbers increased to the point where human societies came into frequent contact and conflict with each other. Warfare became more intense and exerted a very powerful selection on human societies. Societies that were larger and better organized outcompeted smaller and more shambolic ones. Hierarchical organization was one of the cultural traits that was heavily favored by the new selection regime in the Holocene.

Politically centralized societies (chiefdoms) appeared and spread, and then evolved into larger scale societies, complex chiefdoms and archaic states. These societies were extremely unequal. Large segments of populations were enslaved, or became the fodder for bloody rituals, in which multitudes were sacrificed on tops of pyramids. The rulers set themselves up as gods on Earth.



Such extreme forms of inequality alienated the 99 percent of the population, and corroded society-wide cooperation. Ironically, inequality reduced the ability of the despotic archaic states to compete with other societies (which is why they were extremely fragile, and often did not survive their founders). Now cultural evolution began working ceaselessly to make societies less despotic and less unequal. Not because evolution cares about our wishes (and, particularly, that the majority of the population prefers more equitable outcomes), but because more equal, just social arrangements win in competition with the grossly unequal, tyrannical ones.

We’ve come a long way since the days of the early despotisms, but we have an equally long way to go to construct truly just societies. What lessons do Sociology and Cultural Evolution have for us?

First, it’s a pipe dream to imagine that a large-scale society (e.g., a million people or more – a small nation by today’s standards!) can be organized in a non-hierarchical, horizontal way. Hierarchy (in its neutral sense) is the only way to organize large-scale societies. In itself, it’s not a bad thing.

Abuse of power and gross inequality are unquestionably a bad thing. Evolution has been working to eliminate the worst excesses, but we need to help it along. ‘Evolution’ sounds like such a faceless, impersonal force. But it actually is a result of a multitude of actions of real people, from common folks to rulers. Our societies today are much less despotic than the early ones, thanks to such people as the Israelite prophet Amos, who denounced powers that be, and the Chinese sage Confucius, who championed justice, sincerity, and morality.

So what we need is not less hierarchy, but more control over our leaders to ensure that they govern for the collective good, rather than for selfish needs of themselves and their cronies. How we achieve this end is a big question, to which I don’t have a ready answer – and certainly not something that I’d want to deal in this post.

Notify of
Most Voted
Newest Oldest
Inline Feedbacks
View all comments

Anyone who has done any significant editing on Wikipedia can readily understand just how difficult it is for large groups to function without a hierarchy.

Peter Turchin

Any large group cannot function without hierarchy. They either acquire it and become successful, or sit around debating endlessly, or fall apart.

Antonio I read to Victor M. Yakovenko for mathematical models of inequality. I think that It’s very interesting. What do you think about this?

Lee Doran

This is an excellent post Peter and a good basis for further thought and discussion. I have one initial thought and that is we are bedevilled not only by the requirements for hierarchy in a structural sense in order to make it work, but we are also challenged by the values in whose ‘favour’ any given hierarchy is working.

It seems to me that, most basically and generally, specific hierarchies ‘work’ in support of their given (and usually quite well-defined ‘in-group’) — and they often do it (or did traditionally) by demonizing and de-humanizing their out-group(s).

An associated and critical issue, in my view, is the values that drive their specific group. If they are group-specific and local and relatively immediate (food, or shelter or women or beads or…) that’s one thing… if they are larger and more inclusive (the nation, the continent … dare we think: the globe!?) then we are getting close to our ‘modern’ dilemmas… of globalization socially and climate physically…

As more and more people attend to the more universal issues and their enabling values, the closer we get to something like openness and access that of necessity seems to bring equity with it… or so I would hope!

Thanks again!

Best to all,


Peter Turchin

In principle, hierarchies (in the neutral sense) can work for any kinds of values. It’s our job, as group members, to ensure that our group works for worthy goals. More on this here

Sergey Sechiv

a missing foundation for the Jeffersonian “checks & balances” model: “the hundreds”

Ross David H

I found fairly convincing Mancur Olson’s analysis was on to something. To summarize, he thought that it took time for small groups to acquire the ability to manipulate the government to their interests. The longer it had been since a society had been disrupted by a defeat in a major war (or other equally-sized shock), the smaller the groups that were in control of the government.

He wrote this before the Wall fell, but I think that Germany’s unification served as a benevolent shock of this sort, and explains a lot of why Germany has done better since 1989 than Japan; the latter has had longer (since WWII) to acquire entrenched, smaller interest groups in control of things, whereas in Germany the barnacles were shaken loose in 1989 (for example the current Chancellor wasn’t even a West German citizen).

What is somewhat lacking though, as far as I know, is a good, objective metric for how far this process has gone in a given society. I know you have been working on this kind of thing, Dr. Turchin, but I’m curious if there’s a single metric (presumably a weighted composite of several others) that you think can be used to test an hypothesis of this sort. Have you looked at James K. Galbraith’s inequality database, for example? It only includes industrial economies but it might work for more recent societies.

Peter Turchin

I am familiar with this argument of Olson. Looks plausible. However, as far as I know it was never tested empirically.

Doug Jones

String quartets don’t need conductors; symphony orchestras (past a certain size) generally do. In the heady early days of the Soviet Union, the effort was made to do without conductors for symphony orchestras. The main one lasted for ten years before being disbanded. Check out “Conductorless orchestra” and “Persimfans” (=First Conductorless Symphony Ensemble) on Wikipedia.

Doug Jones

Peter Turchin

That’s a great example, Doug!

Yoram Gat

> The problem, of course, is that as soon as you put someone in a central position of a decision-making network, you give them a lot of structural power, and they can, and often will, attempt to convert it into more malignant forms of power: ability to order others around

What is the difference between being “in a central position of a decision-making network” and having the “ability to order others around”? How can the former be legitimate while the other malignant?

> more control over our leaders to ensure that they govern for the collective good

This is the usual failed recipe of “checks and balances” or “accountability”. Supposedly we first give “leaders” power to abuse and then somehow make sure they don’t.

A more promising (and historically validated) approach is setting things up so that decision makers are not pre-disposed to abuse their power to begin with. This can done by having the interests of the decision makers aligned with those of the population. This in turn is done by using sortition to select decision-makers (rather than elections or any other competitive mechanism).

Instead of having rule by an elite, with its own distinct characteristics and interests, sortition-based selection relies on the law of large numbers to guarantee that the decision-makers, who are a statistical sample of the population, are truly representative of the population – their material interests, their ideology, and any other characteristic – are the same as those of the population. The decision-makers can then be expected to promote the interests of the entire population by promoting their own.


On the subject of inequality, one has to be careful that some proposed solution does not produce Yet Another Arrogant Ruling Class. I like George Orwell’s Animal Farm, an animal allegory about Soviet Communism and how it produced a new ruling class.

I’ve found some interesting research that shows how economic and social inequality can emerge without the winners being especially skilled or diligent or crooked, although such attributes certainly help. PLOS ONE: Entrepreneurs, Chance, and the Deterministic Concentration of Wealth It describes simulations of populations of simulated entrepreneurs. Each one invests their wealth in various businesses and the like, though the simulations only include the return rates, which are positive with some scatter. It’s the scatter which makes some of the entrepreneurs come out ahead, because some are luckier than others. The average return being positive then makes a small fraction of the entrepreneurs much more wealthy than the others, to the point that nearly all the wealth becomes owned by only a few of them.

It would be interesting to do this sort of simulation with finite resources, like land, or finite markets, like customers for various goods and services. One may also include bankruptcy for entrepreneurs whose wealth becomes too small, so they drop out of the simulation. It would be interesting to see if one gets integrative-disintegrative cycles out of such simulations.

  1. Home
  2. /
  3. Cliodynamica
  4. /
  5. Regular Posts
  6. /
  7. The Evolution of Hierarchy

© Peter Turchin 2023 All rights reserved

Privacy Policy