We live in a world that is noisy, nonlinear, and composed of networks intertwined with other networks. We rely on manmade networks—the internet, supply chains, telecommunication systems—that are both ubiquitous and invisible. On top of that, these systems interact with natural systems, like ecologies and climate.
In fact, it is easy to live life without thinking about how such complex systems work. But the world forces us to take notice when these networks are disrupted. In recent years, we have witnessed the dizzying interactions that led to the propagation of COVID-19 via air transportation routes, different decisions made by different countries and states, and the dueling narratives between statistics and viewpoints. At the same time, less harried but tremendously important battles are taking place in decisions about energy transitions, climate, and the deployment of AI.
Everywhere you look, people have formed opinions about these issues, based on the false, implicit belief that they understand these complex networks.
But it is manifestly impossible for nearly anybody to understand all the intricacies of many of the massively complex systems facing us today. Even experts can only understand sub-systems.
In reality, systems that are nonlinear and complex baffle us. Our intuition, formed in simpler times, is of little use here. And often we can be simplistic, not recognizing complexity. We imagine networks being far simpler than they are. Or perhaps, when we take positions at extremes, it might be less about ignorance and more about the easy comfort of black and white positions rather than the recognition of shades of gray.
But understanding is important. It is hard to make sound decisions about critical issues when systems are poorly understood. Could we be less blind? Or could we at least display some degree of humility when dealing with complex issues, since they rarely accept quick solutions?
Could we, in fact, acquire some level of intuition about understanding complex systems?
What Are Complex Systems?
First, let’s define what we mean by complex systems. Many systems that appear wildly different—sand dunes, ant colonies, traffic, the internet, power grids, travel networks, ecologies, the propagation of epidemics, the brain, financial networks, and the metabolic pathways within a cell—are complex systems. Research results obtained over the last several years allow us to qualitatively and quantitatively understand the common principles that govern the collective behavior of these seemingly completely different physical, biological, ecological, technological, and socio-technological systems. This is the first lesson emerging from complex systems research: there are commonalities that transcend the differences.
The commonalties become clearer when we define the terms “complicated” and “complex.” They sound similar, but when applied to systems, they work completely differently. The components in a complicated system work in unison to accomplish a function or series of functions. There is a master design with instructions for assembly, and every component is there to fulfil a function. For example, a nuclear submarine is complicated, as are jetliners and clocks. Complicated systems are designed to follow prescribed plans—the clock keeps time. Nothing magical or unforeseen happens when all the pieces are put together. When something unexpected does happen—a single defect, for example—it can bring the entire system to a halt, so redundancy needs to be built into the design in the form of multiple backups. Complicated systems do not adapt.
Complex systems are different. Complex systems can adapt, evolve, “learn” to solve new problems, and tolerate imperfect components. They may have elements that are contextual—elements that can take on many functions. Stem cells in biological systems, for example.
A special class of complex systems is networks. All the examples listed above can be thought of as networks. A network is a system of discrete objects (nodes) with connections (links). In food webs, species (nodes) are connected if one preys on another (links); in air traffic, airports are connected by traffic routes.
There now exists a large body of work, produced over the last two and a half decades, that studies the distinguishing categories of networks. It breaks them into classes and uncovers generic laws that govern their formation and robustness. Once we recognize how to see the world as networks, we find them everywhere.
Characteristics of Complex Systems
Complex systems are nonlinear (that is, they are not sequential or straightforward), and they have tipping points as well as multiple hidden positive and negative feedback loops. That means complex systems can have unexpected linkages, large consequences stemming from small actions, and cascade failures. They also can display emergence: the central distinguishing characteristic of complex systems.
Emergence is what happens when elements—“agents” in the terminology of complex systems theory—interact with each other to produce outcomes that could not have been predicted by examining the agents in isolation. Studying one termite does not give us a clue as to how they form mounds; no amount of knowledge about a neuron can explain consciousness. Something seemingly magical happens when many elements interact together. A special case is the emergence of synchronization; fish organizing in schools, birds in flocks, and 10,000 cells in our natural cardiac pacemakers firing in unison to govern our hearts.
Let’s talk about unexpected linkages and failures. Ecological networks provide rich examples of how our intuition may not be prepared to see faraway connections. A “failure” in the network could be triggered by many reasons, such as the disappearance of a key top species or the introduction of an invasive species in the ecosystem. Loss of a top predator, for example, may cause main prey to overexploit its own food resources, leading to its own extinction. In the last few years, we have had reports that bee colonies have been disappearing, with some areas seeing reductions of up to 85 percent of their bee population. That is a cascade failure. Several possible culprits have been named, including pesticides and electromagnetic radiation coming from wireless technology and telecommunication. If bees continue to disappear, worldwide food chain supplies will be affected. Bees pollinate 70 percent of the more than 100 crop species that feed 90 percent of the world’s population.
But now bees have reappeared! “Wait, does America suddenly has a record number of bees?” was the headline in the Washington Post this March. Again, we can rationalize their reappearance, but it was hardly a scenario being contemplated during the time of the gloomy disappearance reports.
Cascade failures occur also in computer networks, finance networks, power transmission networks, water supply networks, and transportation systems. In a power station, when one part of the system fails, the other parts must then compensate for the failed component. But failures do not have to result in catastrophic results. Failures happen all the time, and networks continue to function. Flights get canceled, and air traffic does not come to a halt. Servers constantly fail, and the internet continues working without anybody noticing. Accidents may slow down a supplier, but the entire supply chain adapts.
(Sometimes, however, the system cannot adapt. A volcanic eruption in Iceland in 2010 caused a catastrophic disruption to air travel across western and northern Europe over an initial period of six days. About 20 countries closed their airspaces to commercial jet traffic, affecting approximately 10 million travelers.)
And sometimes far away effects impact distant areas with pinpoint accuracy. The 2011 Fukushima earthquake and tsunami affected a chemical plant, the only one in the world which produced a shiny pigment used by many major auto makers to give their metallic paints a sparkling appearance. The chain reaction? Major delays in delivering certain car models. Ford, for example, had to tell car dealers that it could no longer take orders for F-150 trucks and other models using “tuxedo black” and three shades of red.
Complex Systems in Front of Us
Can one familiar example capture the enormous complexity of the world around us? Human life is composed of about 25 percent of the 118 elements on the periodic table. It is remarkable that DNA, encoding virtually all of life’s information, involves only five of them: oxygen, carbon, hydrogen, nitrogen, and phosphorus. Calcium appears in our bones, and other trace amounts of elements, like iron, appear in hemoglobin. It took millions of years of evolution before this magic mix emerged as the fine-tuned machines we are.
Our smart phones evolved much more quickly. Current iPhones involve 64 percent of the elements in the periodic table. Many are the rarest of elements, so-called “rare earth elements,” such as tantalum, neodymium, praseodymium, dysprosium, and europium. The complexity involved in putting together a product such as an iPhone is enormous, and this translates today to everything imbued with electronics.
Cars, too, require many elements. This wasn’t always the case. A typical car from the 1940s would have involved around 15-20 elements. Iron for body and engine block; carbon for in steel alloys, rubber tires, grease/lubricants; aluminum for some body panels and engine components; silicon in glass windows; oxygen and hydrogen in rubber tires and lubricants; nitrogen and sulphur in rubber tires; lead in batteries; chromium in chrome plating. Additionally, some paints and finishes may have contained elements like titanium, cadmium, and zinc, but in relatively small quantities. Today, cars are essentially computers with wheels, using as many elements as an iPhone and more.
Elements are very unevenly distributed geographically, so gathering them to create products is a complex system, as well. China dominates almost three-quarters of the market of rare earth elements. Australia is top producer of lithium, followed by Chile and China. All the elements that make an iPhone—circuit boards, sensors of various kinds such as accelerometer and gyroscopic sensor, display outputs, battery, camera—may come from third-party suppliers, which themselves can be very large companies. Managing such supply chains is a massively large undertaking. Add to that the complexity that many elements need to be mined in areas affected by war, or under terrible social inequalities, and require conflict-free approvals.
What We See and What We Do Not See
This highlights the first problem with understanding complex systems: We rarely see the networks that operate behind them. On electric vehicles, we see no tailpipes; therefore, we think that the car has no emissions. We fail to see all that is needed to manufacture and then charge our “zero emissions” cars. We benefit from the amazing possibilities at our fingertips brought up by smart phones but cannot see energy need beyond those of recharging and fail to see the enormous machinery that is needed to run all the amazing searches and apps in our iPhones.
Who’s to blame for this situation? Bounded rationality theory says that humans do not factor all components in making decisions or forming opinion, but rather choose an option that fulfills their adequacy criteria.
Then there is the question of incorporating numbers in our thinking, and in this area, we may get lost in the realms of the very small and the very large. We can measure things more precisely than ever before, but with it brings a layer of abstraction: prefixes like pico to femto quantities, on the small side, are used to describe the spread of pollutants. Prefixes like mega to zetta, on the large side, describe the huge amounts of information exchanged and generated in the world.
It is hard to grasp a change in a prefix, even common ones. One million seconds is less than 12 days, a billion seconds is 32 years, a trillion seconds is 32,000 years.
An example: Is there gold in our bodies? The answer is yes, about 0.2 milligrams in a 70 Kg body. How small is this? About an inch in the distance from Los Angeles to New York City. One lesson is that just because something can be measured does not mean that it is significant. But sometimes it may be. One cannot give absolute rules. Nevertheless, it is hard to grasp the very-small and very-large scales, since those extremes are so far removed from our normal experiences.
And then there is the fallacy of trying to solve a complex problem by taking only one factor into account. We try to use this strategy to solve many problems in our daily lives, and sometimes it even works. But in complex systems, it can be disastrous. Ecology provides sobering examples. The hake fish is valuable commercial catch in South Africa. Fisheries, in their quest for more hake, have argued for culling the population of seals, since seals eat hake. A similar argument took place in the Northwest Atlantic in Canada: less harp seals should equal more cod, fisheries argued. But the problem is that seals don’t just eat these fish. Harp seals eat over 150 different species.
Experiments like this do not work because we fail to recognize the possible interactions. Just five species in a food web could have more than a million potential interactions with other species in their environment. At ten species, we move beyond the number of stars in the universe.
How to Better Understand Systems
The answer isn’t to teach more math. To better understand complex systems, we must learn to imagine every network that emanates from a node and to understand that each node may also be its own network. We must constantly think that there may be more than what is immediately in front of us.
Consider how we could apply this to many issues people feel passionate about today. Many of us care about the food we eat. We may care that animals are treated humanely, and we care about how food production affects the environment. Some respond by going vegan or eating only locally produced food. Others have created consultancies to calculate the CO2 footprint of various processes and food companies. But all this operates in a complex network of energy, water, public policy, and climate. We must understand that we cannot fix a problem by concentrating on only one node.
Another issue is the clothes we wear. Many people do not want them to come from sweatshops, since they involve terrible working conditions for humans and are bad for the environment. In all these discussions, we may have some idea of the nodes and links that we want to be factored in, though in many cases the actual objective function is left undefined. What do we hope to achieve or optimize? Too often we think we can change huge systems with small actions that may ultimately only make us feel better. When more factors are accounted for, it may be far from clear as to which items and actions in these examples would most influence the environment, climate, and human rights.
Things get even thornier when politics enters the picture. Consider the demands to universities for divestments driven by social or energy concerns. This may place us in a wide gray spectrum. Demands do not seem to factor in how dizzyingly complex the relations between companies, suppliers, and customers may be.
While scientific research can give us some intuition as to how complex systems react, we must have some humility as to what our powers of intervention are. An alternative is to change how we think. Design thinking provides an example of enlarging frames. Design thinking is currently evolving from “human-centered design,” to “life-centered” or “humanity-centered design.” A former colleague, Don Norman, and a current collaborator, Bruce Mau, are the center of this. The change of emphasis is significant, putting humans at the center or taking a broader viewpoint and putting all society, life, or the planet at the center. Not all problems demand the wider lens. The problem determines the viewpoint.
Take, for example, climate change. Human-centered design might focus on one aspect: a user who can see the effect of climate change on their eroding shorelines but perhaps cannot fathom the warming of entire oceans and effects on coral reefs. It is hard to grasp the entire complex networks when we can only see the parts of the network that affect us. Not all problems demand fixing everything at once, but seeing more of what is in front of our eyes is always a good thing.
There are ways to introduce complex systems thinking into your toolset. When you encounter a problem, consider the networks that lie beneath. Seek to identify the components of the system, question what you read, check sources and biases, expand your circle, and get information from multiple perspectives.
Then, in cases when choices are under our control, do small beta experiments to see how the system reacts. Be aware of what your objective function is. Be vigilant about outcomes, and be prepared to make modifications to your positions. Rigidity is rarely the best choice.
This might be a tough sell in an era where quick, extreme reactions get attention. But it’s worth a shot, especially when complex systems are at the heart of many of our global problems.
