In the early 1600s, during the Dutch Golden Age, tulips—a flower which had been introduced to Europe less than a century before—had become a status symbol, a luxury item coveted by all who wanted to flaunt their wealth. At the same time, the Dutch were busy inventing modern financial instruments. This became a dangerous combination when, in the mid-1630s, speculators entered the tulip market and futures prices on tulip bulbs—a durable commodity, given their longevity—began to skyrocket. At its peak, in early 1637, single bulbs of the most coveted varietals traded for prices 10-15 times the annual salary of a skilled craftsman (roughly the equivalent of $500,000 to $800,000 today). Even common varietals could sell for double or triple such a craftsman’s salary. And then, in February 1637, almost overnight, prices dropped by 99.9999%, the market collapsed, the contracts were never honored, and tulip trading effectively stopped. It’s generally considered the first recorded example of a speculative bubble. For centuries, theorists have argued various explanations, from outside forces (a bubonic plague outbreak led traders to avoid a routine auction in Haarlem), to rational markets (prices matching demand and never separating wildly from the intrinsic value of the commodity), to legal changes in the futures and options market about the structure of contracts (meaning futures buyers would no longer be obligated to honor the full contract). The Tulip Mania is one of the most famous stories in economics, and no one really knows why it happened in the first place.
Driving home from work, I (and probably most of you) often notice a curious phenomenon, which most of us just take for granted at this point. Every evening at rush hour, my commute slows down. Even when there’s no accident blocking a lane or two, even when the on-ramps are metered to ensure there aren’t dozens of cars trying to merge into the lane at once, even when there’s no dangerous weather, even when everyone is theoretically trying to get home as fast as they safely can, the cars around me on the highway are moving well below the speed limit. We call this phenomenon “congestion” or a “traffic jam,” and everyone has just learned to deal with it. Scientists have tried to model traffic for decades, with everything from fluid dynamics to phase theory. Economists have likened it to “tragedy of the commons” models. But no one has been able to produce a good mathematical model that matches empirical observations and can explain where it comes from in the first place in the absence of external triggering events.
Every summer, when the water in the north Atlantic is warm enough, and the winds are just right, and the atmospheric pressure is just right, sometimes—about a dozen times a year between June and November—a storm that, at any other time would remain just a storm, picks up speed and begins cyclonic motion. And if the conditions are just right (and no one is quite sure what “just right” means), that cyclone will develop into a hurricane. These massive tempests are to the original storm what the Great Chicago Fire was to the lantern that first lit the flames. While the normal storm would have made some people wet and maybe knocked some trees over, hurricanes can cause widespread death and destruction among whatever’s in their paths, whether it’s fishing villages in the Caribbean or the New Orleans metropolis. And, much like the Tulip Mania or traffic jams, while scientists have gotten reasonably good at identifying risk factors, no one is really sure what causes an ordinary storm to become a hurricane. It requires the perfect combination of the right factors in the right place at the right time. We can identify the (mostly) necessary conditions, but even when all of them are present, often a hurricane never appears. Sometimes one appears even when they aren’t all there. And yet, despite this apparent randomness, it happens like clockwork, a dozen or so times a year in the same six month timeframe.
Why do we care? What do Dutch tulip markets, highway congestion, and tropical cyclones have in common? The answer is all are natural features of what we call “complex” systems. In this series of articles, we’ll look at what complex systems are and how they differ from complicated systems. Markets, urban commutes, and weather patterns are all examples of different types of complex systems, and sometimes complex systems inherently exhibit unpredictable, wild, seemingly inexplicable behavior like bubbles and crashes, congestion and slowdowns, and out of control feedback loops. Not because anyone wants them, or because they design for them, or they screwed up and designed the system badly. But because that’s the nature of complexity.
Complexity is a difficult term to define, even though it’s been widely used in various scientific disciplines for decades. In the next article of this series we’ll look at the defining characteristics of a complex system. But for now, we’ll stick to the broad overview. Complexity is the state in which the components of a system interact in multiple ways and produce “emergence,” or an end state greater than the sum of its parts. Cars, buses, a multi-lane highway, public transportation, on- and off-ramps, surface streets, traffic lights, pedestrians, and so on are the components of the system. They all interact in many different ways in a densely interconnected and interdependent system—what happens in one area can have wide-ranging affects across multiple areas of the system as a whole. And thus, even though everyone hates traffic jams and everyone just wants to get home as efficiently as possible, the traffic jam nonetheless appears, like clockwork, every evening at rush hour. Congestion is an emergent property of the commuting system. It is more than the sum of its parts, completely different that the pieces making it up, the cars and the roads and so on. That’s complexity, in a nutshell.
Contrast this to the other major type of systems, which we call “simple” and “complicated.” A simple system is something like a simple machine. A pendulum is a simple system. A lever is a simple system. In these, the system is the sum of its parts. It allows us to do things we could not do without the system, but it is additive. There are limited interactions, and they operate by well-defined rules. A complicated system is just the extension of this, composed of many simple systems linked together. Whereas the defining feature of a complex system is interconnectivity, a complicated system is defined by layers. Hierarchical systems like military organizations are complicated systems: they may be very difficult to work through and figure out what goes where, but when you figure it out, you can see all the relationships and know what effects an action in one area will have elsewhere. Many engineering problems deal with complicated systems, and thus humans have become quite skilled at understanding these types of systems: we use mathematical tools like differential equations and Boolean logic, and can distill the system into its essential components, which allows us to manipulate the system and solve problems. It may be difficult and take an awful lot of math and ingenuity, but at the end of the day, the problems are solvable with such tools.
Complex problems, however, are not solvable with the traditional tools we use to address complicated systems, because by their very nature they work in fundamentally different ways. As I already mentioned, they are defined not by the components and layers, but by the interconnectivity and interdependency of those components. The connections matter more than the pieces that are connected, because those connections allow for emergent properties greater than the sum of the parts. They allow for butterfly effects and feedback loops and inexplicable changes. Complex systems are not all the same—complexity can occur in deterministic physical systems like weather patterns and ocean currents, or in nondeterministic social systems like ecosystems and commodities markets and traffic patterns, and even in deterministic virtual systems like computer simulations. Because, again, what matters for complexity is the connectivity, not the components.
And because complex problems are not solvable with the tools we use to solve complicated problems, we often get unexpected results, causing even worse problems despite our best intentions. This fundamental misunderstanding of how complex systems work has led to everything from inner city gridlock to economic collapse. Researchers have only been studying complexity for about three decades now, but it has revolutionized understanding in fields ranging from computer science and physics to economics and climatology. It’s amazing what you can do when you start asking the right questions.
In the next article, we’ll look at the characteristics of complex systems and a couple different types of them. Then we’ll look at the tools we use to understand them. And finally, since I’m an economist and this is my blog, we’ll look at the relatively new field of complexity economics and try to understand some the lessons learned about how markets actually work.