Cassandra has moved. Ugo Bardi publishes now on a new site called "The Seneca Effect."

Saturday, March 17, 2018

The View From Les Houches: What is the origin of Collapse?

At the physics school of Les Houches, in March 2018, Gregoire Chambaz of the University of Lausanne gave a talk on the phenomenon of "collapse caused by diminishing returns of complexity." (The image above is not from Les Houches but from a meeting in Lausanne last year).

In itself, it is already interesting that a meeting of physicists gives space to the idea of societal collapse, but the school of Les Houches was one of the rare cases of a truly interdisciplinary meeting. The result was a wide variety of approaches, including the talk by Gregoire Chambaz who approached the problem examining the concept of "diminishing returns of complexity" proposed by Joseph Tainter already in 1988. You can find a summary (in French) of Chambaz's work at this link.

If you are a reader of this blog, you probably know Tainter's graphic to explain his concept. Here it is.

The idea is that, as societies become larger, they must develop more and more complex control systems in order to manage the whole system. These control systems may be in the form of bureaucracy, an imperial court, the army, the church, the legal system, and more. And, as these systems become larger, they become unwieldy, rigid, and unmanageable. The effort needed to increase their size is not matched by the benefit they provide. According to Tainter, this is the ultimate reason for the collapse of large societies.

As a model, Tainter's one has proved to be hugely popular and surely it is a "mind sized model," easy to understand and providing an immediate grasp of the evolution of the system. The problem is that Tainter's model has no evident basis in physics. There is no precise explanation of what would cause the behavior that Tainter proposes, not it is possible to measure concepts such as "the benefits of complexity." It is only a qualitative model.

Can we model this kind of collapse using physics? Perhaps. In principle, there could be two reasons why the system stops improving its performance as it grows in size. One could be an effect of entropy. If you work in a large organization, you understand how, over time, it becomes a tangle of contradictory rules and of people and offices which seem to exist only to prevent any work being done (OK, I have in mind the University of Florence, but I am sure it is not the only case in the world). But how to quantify this effect?

Then, the reason for this behavior could be another one. Maybe it is not an intrinsic property of a large system to lose efficiency as it grows, but an effect of the slow decline of the net energy that it uses. That would explain many things and I put together a tentative dynamic model a few years ago which seemed to work. We are working on improving it taking into account the dynamics of the Seneca Effect. It is a work we are doing together with my coworkers Sara and Ilaria, but it will take a little time before we publish it.

Overall, the impression I have is that we are starting to develop an extremely rich field of studies, that of critical phenomena in complex networks. Tainter gave us a first indication of the way to go, but there is much, much more to do before we can say we have a solid theory explaining the periodical collapse of civilizations we observe in history.  But we keep going.


  1. It's interesting to note that with machines, dimensions improve efficiency. Bigger steam engines are more efficient than smaller, as Isambard Kingdom Brunel proved in 19th century. With human societies, it's vice versa. Bigger group is less efficient either because of diminishing returns due to complexity, or due to decrease of net energy.

  2. Hi Ugo. One way to look at it is to see how complex systems, for example legal and political systems, grow until they reach a resource limit. This limit may be well known: the grain harvest, or taxes, or in our case in Australia, the availability of diesel fuel.

    Any system to the right of C2 on Tainter's curve is likely to be very vulnerable to a sudden change in resource availability, much more so than when the system is between C1 and C2 where it it is in its growth phase and has much more flexibility, as its growth is being governed by its utility rather than its consumption of resources. The reason for this vulnerability is that the utility of the sytem has become fixed or is diminishing in value and every one in the system is working harder and harder to try and stay in the same place.

    It should be possible to build a model of the growth phase C1 to C2 showing how growth is governed by external needs for the "good" the system produces (eg growth of legal complexity and resources thus consumed, arising from laws governing automobile safety and pollution abatement) and another describing the curve to the right of C2 as arising from internal constraints of the system now in existence (careers invested in it, sunk costs leading to over-investment in established policy directions which suffer from diminishing returns, maintenance costs associated with training new staff, everyone working harder to maintain output and thus vulnerable to accident or disease etc).

    However the complexity of something like Industrial Civilisation itself ultimately defies any attempt to describe it accurately. This can be seen in a crude form in the endless news items where 'Futurists' simply project from a snapshot of the world they've known for twenty or so years and sketch out scenarios of distant futures which depend on the already failing institutions to produce more of the same: ie. endless growth along what are very fragile current paths.

    Guarding against this 'Normalcy Bias' is very difficult. Perhaps it's easier to be a story teller than an analyst of complex human systems!

    1. Hi Lloyd. I couldn't agree more with your comment. I wonder if you are in Sydney, because I would surely love a friend in town who thinks like you. It is really lonely in the various circles I inhabit because of this normalcy bias.

  3. Hmmm ... I remember reading a long time ago of two theories of ageing. One was that it was 'programmed', the other that it was an accumulation of copying error in a self repairing complex system.

  4. Deux livres que je conseille: "Mémoires trouvées dans une baignoire" de Stanislas LEM et "The Fate of Empires and Search for Survival" de Sir John Bagot Glubb. Tout ça est arbitraire mais peut constituer des pistes de réflexion.
    Jean-Luc Jourdain

  5. My take on this has been that the main cause of diminishing return is due to the increasing "transaction costs" between all the parties in the network. If we look at complexity mainly as an increase in specialisation in all levels, I believe it is quite easy to see that to have one guy specialise in being a blacksmith while the other ten are farmers has quite apparent advantages and very low transaction costs, assuming that he is fully occupied or farm as a part time occupation when there is little to do. But with increasing number of specialist occuptations the coordination/transaction costs keep increasing but not necessarily the benefits. In addition each person either is an idler or fills the job with admin, which normally takes other peoples energy. Well something like that. Must admit I have't read Tainter, so perhaps this was a redundant comment. Keep in writing Ugo.

    1. I think I see what you are saying, Gunnar, and I certainly see it here in the metropolis where I live. I have another theory of a contributing factor--the more frequent occurrence of the butterfly effect and the larger magnitude of said effect because of the ever-increasing size of the system. I see this in the Sydney mass transit system as a bit of a microcosm. Cheers to you, Ugo, and all the readers. What a wonderful corner of the internet.

  6. My I suggest an alternative way to think about Seneca Cliffs?

    Assume that behavioral sciences are progressing, and are becoming increasingly effective at steering human behavior. Also assume that knowledge of the behavioral sciences is not equally distributed, but that there are a few experts and very many people who do not realize that they are being steered. Also assume that the people being steered are susceptible to addiction. Now assume that a handful of corporations and wealthy humans have control of the monetary assets.

    Then it follows that the corporations will hire the behavioral scientists to create addiction in the unwashed masses, and that once society comes to regard the unhealthy behavior as 'normal', then it will be extraordinarily difficult for people to break the addictions.

    Now, so far as GDP is concerned, selling sugar to sugar addicts, and opioids to opioid addicts, and alcohol to alcoholics, and faux status to almost everyone, is good and economically productive. It can even be taxed, which supports the central bureaucracy...which is, by government definition, a very good thing indeed.

    The problem of course, can be understood if we are familiar with Antonio Damasio's book The Strange Order of Things: Life, Feeling, and the Making of Cultures. Damasio has written 'a pathbreaking investigation into homeostasis, the condition that regulates human physiology within the range that makes possible not only the survival but also the flourishing of life.' (from Jacket blurb).

    The corporations and their kept scientists are definitely NOT keeping humans in the homeostasis condition...quite the contrary.

    And thus, eventually, the society will collapse.

    Don Stewart

  7. Hi Ugo, I notice that everyone keeps copying the original graphic from Tainter's work, but no one ever notices that "benefits" is spelled incorrectly on the y axis. Maybe you could make a new graphic with this spelled correctly and maybe people on the web would then start copying your corrected graphic instead.

    1. Hadn't noticed that! Even the great Tainter makes mistakes!

  8. Ugo I see no reason why any complex system could not continue to grow theoretically into infinity were it not for physical boundaries (limits). Tainter's approach like a lot assumes, a priori, that the parts are perfect and inputs and outputs are likewise stable flows. Neither humans nor any other physical biological being is perfect but the best current physical recombination of organic chemicals that can or has been capable of being produced using the physical available resources, DNA and numerous physical inputs (water, chemicals, energy, environment, and space/time). Its form defines its function. Rigidity is but a form of stasis. Complexity is but a form in space/time that suits the available resources need for the organism and structure to survive. To my mind collapse begins when the form is altered by external or internal change that adaption or the structures design cannot sustain. In a very simple form, you cannot adapt to less food or water as organism beyond a certain point because the organism begins to fail or can no longer function as it did. The more complex and interrelated the parts are the greater the risk of failure of a part leading to failure of the whole. The rest is but a narrative. To my simple mind this is why big animals go first and the smallest last. We are the last of the big animals and we are consuming and changing the factors we need at such a prodigous rate we can only collapse.

  9. Ugo the 'Limits to Growth'I posit charted (within the constraints of the available data) the physical limits that we as a complex organised structure can sustain. I am not so sure that the discipline of Physics can add much more but to further refine the quantification. History provides the narrative or chart of how we humans have changed and adapted the structures which have been able to sustain growth, in its essence it is provides a cogent explanation and model of predation which is the means by which what we call society, human nature and our supporting social structures or economic organisations have survived or not survived. We are absorbed by our own hubris, delusions and success at what we thought were overcoming physical limits. We have one mental characteristic that distinguishes from any other organism - the recognition of our mortality and we cope with delusion. But we can be predated upon by something as simple as microbe which could prove fatal. By santifying reproducing ourselves we have staved off mortality, for a time. The chemical pot pourri that sustains us is however diminished by our increasing mass, by its pollution and substances poisonous to our survival. We understand that raising the earth's temperature is a recipe for instability in all the earth's biosystems, we understand that there is only so many resources we can obtain, yet we collectively are bound to a current form of systems organisation that means we do nothing about these critical failure points or stave off collapse, the predatory gene design wins every time and we no longer have time to nourish and support beings who are not so or adapt, quite the opposite we are very adept at killing such human adaptions amongst our own group of bio-organisms, guaranteeing our collapse.

  10. More thoughts on Damasio and Seneca Cliffs

    From page 20:
    (Bacteria) appear to lead a simple life, operating according to the rules of homeostasis, but there is nothing simple about the flexible chemistries that they operate and that allow them to breathe the unbreathable and eat the uneatable. In the complex, albeit un-minded, social dynamic they create, bacteria can cooperate with other bacteria, genomically related or not. And in their un-minded existence, it turns out that they even assume what can only be called a sort of ‘moral attitude’. The closest members of their social group, their family so to speak, are mutually identifiable by the surface molecules they produce or chemicals they secrete, which are in turn related to their individual genomes. But groups of bacteria have to cope with the adversity of their environments and often have to compete with other groups in order to gain territory and resources. For a group to be successful, its members need to cooperate. What can happen during the group effot is fascinating. When bacteria detect ‘defectors’ in their group, which really means that certain members fail to help with the defense effort, they shun them even if they are genomically related and therefore part of their family. Bacteria will not cooperate with kin bacteria that do not pull their weight and help with the group endeavor; in other words, they snub noncooperative turncoat bacteria….The variety of possible bacterial ‘conduct’ is remarkable. In one telling experiment designed buy the microbiologist Steven Finkel, several populations of bacteria were to fend for resources inside flasks equipped with different proportions of the necessary nutrients. In one particular condition, over multiple generations, the experiment revealed three distinct successful bacteria groups: two that fought each other to the death and suffered major losses in the process, and one that had sailed discreetly over time, without any frontal engagement. All three groups made it into the future, a future as long as twelve thousand generations.

    …In their unmanned orientation to survival, they join with others working toward the same goal. Following the same undeliiberated rule, the group response to overall attacks consists of automatically seeking strength in numbers following the equivalent of the principle of least actions. (

    And on page 23:
    Their (social insect) genetically set and inflexible routines enable the survival of the group. They divide labor intelligently within the group to deal with the problems of finding energy sources, transform them into products useful for their lives, and manage the flow of these products. They do so to the point of changing the number of workers assigned to specific jobs depending on the energy sources available. They act in a seemingly altruistic manner whenever sacrifice is needed. In their colonies, they build nests that constitute remarkable urban architectural projects and provide efficient shelter, traffic patterns, and even systems of ventilation and waste removal, not to mention a security guard for the queen. To be continued

  11. continued
    Now a few interpolated comments from me. Humans have built on the achievements of ‘lower’ forms of life, notably the microbes which generate soil fertility and are symbiotic with the human proper, notably in our guts. The microbes are essential players in human digestion of food. Humans could not exist without the soil food web and the equivalent cultures in the oceans and lakes and streams.

    Humans have also used both ancient and recently evolved mechanisms to create peculiarly human cultural artifacts, from tribes to language to the internet and smart phones. We have also been guilty of failure to understand just how vulnerable we are, and how buying a shelf full of ‘antimicrobial’ products is not a very smart thing to do. In addition, as in my fable of a society which turns life and death decisions over to those with money and power, we have developed suicidal cultures.

    Consequently, a human culture is a networked system of both ancient non-human cultural inventions along with our own very recent additions. As Damasio observes, the history of human cultures is characterized, thus far, by increasing sophistication followed by resets toward the basics of homeostasis, which relate to the individual. Thus, we might speculate that the survivors of a nuclear holocaust (if any) would be dependent on surviving non-human cultures (from the simple to the sophisticated) plus a very local human culture. If the survivors lived long enough, and could find appropriate sources of energy, they might reconstruct some of the complexity we can observe in the history of humans. Whether we would create cultures which fight to the death (as two of Finkel’s colonies did), or cooperated for the communal good (as one of Finkel’s colonies did) would depend of lots of different factors. History might not repeat, but would probably rhyme.

    The post-collapse arc would likely be either the biological equivalent of Feynman’s principle of least action or extinction.

    In other words, we would probably see a Seneca Cliff, followed by either a slow re-creation of more complexity, or else extinction.

    Don Stewart

  12. I don't think that physical models can explain all humain behaviors. Models can be created, but they are limited by the creativity of the modeling person.
    Complexity requires competences. Increased complexity increases the needed competences. There is a big temptation to replace competences by controls. The companies that have the best survival chances seem to be the ones that have been able to limit or reduce complexity instead of increasing controls. Some do it the wrong way, like outsourcing to companies that never are non profit organisations. If you loose the knowledge, you might loose the market.
    Measuring work quality with KPI is something impossible because the work is the most difficult and critical when everything goes wrong. KPI is more a measure of normal activity flow than of quality work.



Ugo Bardi is a member of the Club of Rome, faculty member of the University of Florence, and the author of "Extracted" (Chelsea Green 2014), "The Seneca Effect" (Springer 2017), and Before the Collapse (Springer 2019)