The Two Achilles Heels of Complex Systems
Tight coupling and limits to human comprehension
According to anthropologist and historian Joseph A. Tainter, human societies became more complex in response to the problems they faced. Agrarian societies developed writing and math to keep track of debts, trade, laws and contracts; giving rise to a new class of non-productive workers (scribes). At first this and similar increments in social complexity provided a huge net benefit to the community at a relatively modest cost. Then as civilizations grew larger and larger, more and more non-productive roles had to be added to handle the exponentially increasing number of issues — up to the point where the appointment of the next batch of bureaucrats costed more than the benefits they provided (1). After a society had run into the issue of diminishing returns, however, experiencing a large bout of involuntary simplification was only a question of time.
Thank you for reading The Honest Sorcerer, and special thanks to those who already support my work; without you this site could not exist. If you are new here and would like to see more in depth analysis of our predicament, please subscribe for free, or perhaps consider a paid subscription. You can also support my work by virtually inviting me for a coffee, or sharing this article with a friend. Thank you in advance!
Interestingly enough, it’s no different with today’s modern societies and with the systems these societies operate. Perhaps the best example is the electric grid itself. With millions of users, thousands of nodes (substations) and power plants on top of continent wide distribution networks, electric grids are one of the most complex systems a civilization can possibly come up with (strictly after tax laws, of course). The question thus poses itself: is there a limit to the grid’s complexity? What can be done to avoid “a large bout of involuntary simplification” when adding more complexity ceases to provide positive returns? Oh, and what does tight coupling and limits to human comprehension has to do with all this…?
Before we delve into the topic, it must be first understood that these large grids all operate on alternating current, with polarity shifting between positive and negative at regular intervals. And while other parameters such as voltage can be different across various grids, and even within the grid itself — think high voltage lines — one thing must remain stable: the frequency of the alternating current. (Don’t concern yourself if it’s 50 Hz as in Europe, or 60 Hz as in America — all that matters that the selected value must remain constant across the entire network.) Think of this frequency as the heartbeat of the system; ensuring the synchronization of all electrical devices hooked up on the grid. Even small deviations from the norm can lead to device malfunctions, shortened equipment lifespan, increased maintenance costs, or, in extreme cases, power system failures, exposing entire regions to the risk of blackouts.
Back in the heyday of the grid’s expansion (or throughout much of the 20th century), alternating current was supplied exclusively by huge spinning generators (alternators). These multi-ton devices were revolving exactly at grid frequency (50 or 60 Hz) and were rotated by either hydro, steam or gas turbines. Thanks to their massive weight they also acted as a flywheel, smoothing out minor intermittencies in the system. Due to efficiencies of scale these generators had to be deployed as part of large power plants — housing several of them — thus their electric output could be centrally regulated. In essence, the grid used to be one big humming machine.
Operating the electric grid came down to performing a delicate balancing act between consumption and electricity generation. Companies were thus incentivized through pricing to use a constant amount of power throughout the day, perfectly in sync with the capitalist incentive of sweating one’s assets. The electric utility enjoyed stable and plannable operations, with maintenance periods scheduled months ahead and equipment running at full utilization, while manufacturing plants and smelters enjoyed low and stable prices. Investments were thus easy to plan and the economy devouring the planet’s resources could grow in a steady and unabated fashion.
Then came the first oil shock in the early 1970's, when the steady 7% year-over-year growth in world oil output has stopped and reversed within a couple of years (2). Growth has returned very slowly, but never reached the prodigious rates of the past. Power plants previously fed by fuel oil had to be stopped or converted to burn other fuels, and nuclear reactors popped up like mushrooms to fill in the gaps. This first round of adaption to a sudden decrease in world oil production did not effect the grid’s stability, though, as these new power plants all utilized the same principle of generating electricity: spinning huge turbines with massive alternators attached.
The increase in complexity came in the form of power plant design: nuclear reactors required a number of safety measures, including the addition of a large number of mechanical equipment from simple valves to back-up generators. All with their respective tendency to fail at the worst possible time… necessitating redundancies and back-ups upon back-ups. These additional safety measures (compared to say a coal fired power plant) have made nuclear reactors expensive to build and operate, requiring a highly skilled and trained special class of engineers (3). Social complexity has thus risen hand in hand with technical complexity in response to our growing geological predicament: the depletion of high grade, plentiful resources and their eventual replacement with harder-to-get and harder-to-work with ones.
Fast forward a couple of decades and we get to the second major oil crisis. Starting around 2004 oil supply growth has slowed down after decades of anemic, 1.4% year-over-year increases. This, combined with a surge in demand from China, has resulted in skyrocketing oil and commodity prices (as these, too, are mined and moved by oil) and made recovery from the 2008/9 financial crisis all the more harder to achieve. Were it not for shale oil and other energy-intensive-to-get forms of petroleum production (from Canadian tar sands to Venezuelan ultra-heavy oil), the world would already be on the decline side of Hubbert’s peak. At first, panic rose among government circles, countries were bombed and regimes were toppled in oil rich nations, but thanks to these non-conventional forms of oil coming online much faster than anyone previously expected, decline was averted. At least for a while.
As a result of the shale oil boom the discussion has shifted away from fears of peak oil, towards “tackling” the climate crisis. The focus, however, remained the same: reducing the (western) world’s reliance on fossil fuels. From the grid’s perspective this policy shift has marked the end of an era, an era of stability and predictability. Especially in Europe. Before we go on, here we must briefly tie back to the topic of grid frequency and the type of electric supply. As you remember, the grid operates on alternating current (AC), provided by synchronous generators. “Renewables” — made entirely from non-renewable materials with coal oil and gas — on the other hand, produce direct current (DC), which must be converted to alternating current before it could be fed into the grid. And herein lies a major technical challenge.
Converting direct current from solar panels and wind turbines into alternating current on the grid happens in inverter units sitting right next to each wind and solar installation. And how does this inverter remain in sync with the grid? How does it “know” when exactly to switch polarity? Well, simply by mimicking the frequency it “sees” on the network. Applied in a small scale, where the vast majority of electricity is still produced by large generators, this is a brilliant idea. This mimicking behavior — as it was demonstrated in Spain recently — however, presents a massive vulnerability to the entire grid. Beyond a certain percentage of wind and solar on the network (called penetration) sampling the grid for frequency creates a potential for a rapidly propagating self-reinforcing feedback loop.
This is exactly what happened in the Iberian peninsula on the 28th of April, 2025. For reasons yet to be clarified the grid’s frequency fell (most probably due to a momentary lack of supply) and thousands of inverters dropped their frequency in response. This self-amplifying behavior triggered safety shut-downs in power plants to protect sensitive equipment and thus to prevent a much longer than necessary black-out from occurring. With power plants self-disconnecting their AC generators the grid’s frequency dropped further still, initiating shut-downs in all of the remaining power plants (and solar panels) ending in a complete, country-wide blackout. All this in less than a few hundred milliseconds. Think about that for a while.
High penetration of “renewables” has not only introduced weather dependent, intermittent electricity supply to the grid, but has also created a massive vulnerability. Something which can lead to a complete blackout even on perfect sunny days, when no interruption was to be expected. Replacing gas and coal fired power plants with wind and solar has come with the removal of large generator units, which otherwise could’ve been able to smooth out shorter intermittencies and drops in grid frequency. As for proof, look no further than neighboring France. Thanks to its fleet of 57 nuclear reactor units, the French grid had high enough spinning inertia to stabilize the grid and to stop the propagation of a cascading collapse starting in Spain. Still, it was a near miss, and if France fell just as Spain and Portugal did, the whole European continent could’ve turned dark in a matter of minutes.
The case of the recent blackout in the Iberian peninsula has showed us something else, beyond the importance of stability. It exposed the two Achilles heels of every complex system operated by humans: tight coupling and limits to human comprehension. The addition of large number of “renewables” onto an old grid designed to run on AC generators has increased complexity — and thus vulnerability — beyond the point of human comprehension. Engineers tasked with maintaining grid stability had to intervene ever more often by curtailing or rerouting excess wind and solar production and firing up gas power plants in anticipation of bad weather. But weather is notoriously hard to predict, especially in a rapidly shifting climate regime, forcing operators to come up with emergency measures often within minutes, and sometimes in a matter of seconds. All this at the same time when inertia (and thus peace of mind) from AC generators have been removed to allow room for the addition of yet another batch of wind and solar… Which, on the other hand, are prone to cause a cascading collapse within seconds should things go slightly wrong, thanks to their tightly coupled nature. And you thought that the work of air traffic controllers was stressful.
Were it not for such tight coupling and the removal of spinning reserves, the sudden increase in the number of electricity generating devices would still cause a major headache. Adding more individual pieces of equipment (solar panels, turbines, grid scale batteries, transformers, high voltage lines etc.) is not only costly, but also increases complexity to yet another level. Should more than a tiny fraction of the gazillion pieces of equipment fail at the same time, a cascading grid collapse spreading the entire continent would be really hard to avoid. Yet, utilities still keep adding more and more safety and other equipment to the grid to make room for even more wind and solar — even as some of the network components are more than half-a-century old — creating a Frankenstein monster of immense complexity (4).
And this takes us back to the limits of human comprehension. Should we then trust AI to oversee the grid for us…? Well, how is that for a vulnerability in an age of hackers? Not to mention concerns related to the emergence of a general artificial intelligence (cough, Skynet)… Or how about a good old AI hallucination, when the machine “thinks” it’s better to shut down a city to prevent a grid collapse, when in fact it just misinterpreted some irrelevant data points? I guess you start to see my point: trying to resolve issues arising from an exponential increase in complexity by adding yet another layer of complexity (AI) is asking for more trouble. Is a total, continent wide grid collapse resulting in chaos and anarchy inevitable then? Well, yes and no. Our stupid tendency to add more complexity when actually less would be preferred will almost certainly lead to a major blackout in the near future. Maybe not this year, maybe not even in this decade. On the other hand, I do not expect a civilizational crash and burn event either: there are still plenty of highly skilled people working on maintaining a stable grid, and restoring its functionality within a couple of days at worst.
What should make us and our wise betters and elders think, is the question how will we adopt to an ever less reliable and ever less affordable supply of electric power? All this, of course, in parallel with a long decline in world oil production… Although oil is no longer directly burned in power plants, it’s sister — natural gas — still is. And since most of the gas comes to the surface as associated gas (meaning: together with oil), as oil wells get shut off due to depletion (5) and a loss of profitability, we can expect a similar decline in natural gas production. Since coal is in the process of being phased out both in Europe and in America thanks to its ever increasing extraction costs (ultimately due to the depletion of rich easy-to-get reserves), a rise in natural gas prices will most likely translate into higher electricity bills (6). This, on the other hand, would lead to more and more people self-disconnecting (as they could no longer afford electricity), putting an ever increasing burden of maintenance costs on an ever shrinking customer base. A recipe for success? Hardly. A slow but accelerating decline? All the more so.
We are lurching towards a post-electric world, where national grids will gradually become unaffordable, and thus will be broken down into smaller and smaller local grids. Lacking adequate funds some areas will be left without power for months, while rich neighborhoods will enjoy uninterrupted supply for many decades to come. Electricity will be increasingly intermittent for the most of us in the West: first every once in a while then, as our economies continue to deteriorate over the decades ahead, ever more often. Western manufacturing plants will struggle to keep operating and remaining profitable under such conditions, leading to a final round of offshoring into countries where electricity supply is still more or less stable. Instead of a massive blackout turning our cities into scenes of a zombie apocalypse, a long slow decline sprinkled with intermittency looks much more plausible.
Until next time,
B
The Honest Sorcerer is a reader-supported publication. Please consider a subscription or perhaps buying a virtual coffee… Thanks in advance!
Notes:
(1) Increases in social complexity requires energy and resources. The clerk who no longer produces food still had to be fed — just like soldiers in an ever growing army, local administrators, tax collectors and the rest. Such societies could grew in complexity only as far as their food production system allowed. Hence the need for territorial expansion and the need to get more workers producing surplus food and tradable goods. Complexity is thus a function of surplus energy: the portion remaining after energy producers (agriculture, oil wells, mines etc.) get their share. If the energy cost of energy goes up, surplus energy vanes, leaving complex societies with a conundrum: simplify voluntarily or go bust. Needless to say, most choose the second option.
(2) The Arab oil embargo was just the cherry on top. The real reason lied in geology: as petroleum geologist M. King Hubbert accurately forecast, oil production peaked then begin to decline in 1970 in the then biggest oil producer of the world, America.
(3) The need for highly trained specialized engineers and skilled labor necessary to operate nuclear reactors ought to cause concern when it comes to deploying a large number of small modular reactors (if they ever become commercially available).
(4) China, in the meantime, still keeps expanding its coal fleet to compensate for a massive addition of “renewables”. Since their grid is much newer, and expansions were planned with intermittency in mind, stability (as I understand) is not yet of an issue there. This is not to say, that they will never reach incomprehensible levels of complexity, or that their grid doesn’t face the same fate. They are just a couple of decades to half a century behind the West.
(5) Depletion does not mean that we run out of stuff, but that we run short of the affordable portion of it. As the easy-to-get petroleum gives way to harder and harder to get oil, the energy needed to haul and deliver liquid fuels to markets will continue to increase, eventually cannibalizing every other source of energy — including electricity. As oil production struggles to remain flat and begin to decline, so will the demand for power rise, even as the grid providing it gets more and more fragile and expensive to maintain.
(6) Even if we somehow managed to find the trillions of dollars needed to completely overhaul the grid, the price of a kilowatt-hour of power would still continue to rise. The addition of a gazillion pieces of equipment (solar panels, turbines, grid scale batteries, transformers, high voltage lines etc.) would create a massive liability, as each and every one of these gadgets would need to be replaced at regular intervals (ten to twenty years). The grid “overhaul” would thus not be a single event, but a continuous activity, till we run out of affordable copper and other critical materials. When do we realize that there is no infinite growth — nor stable equilibrium — for a civilization hell bent on depleting its resources?
If you think it’s hard finding competent people to work on these problems and operate these systems, just wait a few decades when all the current holders of such expertise start to retire. Many of them are already retired and what is coming after them isn’t pretty. Think about how much standards have been lowered in schools all across the country. The goal seems to be to get the students through high school and out the door whether they’ve learned anything or not. And then on to college where the degrees have been so watered down that a masters is now needed to show you have bachelors level knowledge and so on. How are employers going to know who to hire? Even medical and law schools are lowering their requirements for graduation. Woe to the person that needs a competent doctor (which is everyone sooner or later) or even a competent lawyer.
Fascinating and frightening, another reason why the drive to Net zero is madness. Thank you