Why Do Complex Systems Develop Unexpected Behaviors?

How feedback loops, small changes, and hidden connections produce outcomes nobody planned.

Complex systems develop unexpected behaviors because their parts interact, respond to one another, and change the conditions under which other parts operate. Small disturbances spread through these connections, feedback loops amplify or correct them, and accumulated changes sometimes push the entire system into a new state.

Scientists call this emergence: patterns or behaviors that arise from interactions among components rather than from any single component acting alone.

The system hasn't stopped following rules. Its rules have produced consequences nobody anticipated.

The Traffic Jam That Nobody Caused

In 2008, Japanese researchers placed 22 cars on a circular track approximately 230 meters long.

Drivers were instructed to maintain a steady speed and distance from the vehicle ahead.

There were no intersections, traffic lights, accidents, or lane closures.

Initially, traffic flowed smoothly. Then small variations in speed began to grow.

One driver slowed slightly. The next driver responded, followed by another. Eventually, vehicles began stopping and starting, creating a traffic jam that traveled backward around the track.

The jam emerged without a physical bottleneck or a single driver causing the breakdown.

The crucial condition was traffic density. Once vehicles were sufficiently close together, small disturbances grew instead of disappearing.

Research: Sugiyama et al., New Journal of Physics (2008)

Every driver was responding to nearby conditions. Together, those responses created a pattern no driver intended.

Why Understanding the Parts Isn't Enough

A complicated system and a complex system are not necessarily the same thing.

A mechanical watch contains precisely engineered components. Understanding how they fit together explains how the watch operates.

Now consider a city.

Residents choose where to live based on affordability, employment, schools, and preferences. Businesses respond to residents. Transportation changes where people work. Housing prices respond to demand.

Each decision changes the conditions influencing other decisions.

There is no single person determining the city's overall behavior, yet recognizable patterns emerge.

Economist Thomas Schelling demonstrated this principle in his influential 1971 paper, Dynamic Models of Segregation.

His models showed how individual residential preferences could collectively produce substantial segregation, even when those preferences were less extreme than the resulting pattern.

The model does not explain every cause of real-world segregation, including discriminatory policies and historical inequalities. It establishes a narrower principle.

Individual intentions and collective outcomes can differ dramatically.

Understanding what people want is not enough. We must understand how their choices affect one another.

When Solutions Become Part of the Problem

Unexpected behaviors also develop when systems repeatedly correct themselves using delayed information.

Consider a retailer experiencing an unexpected increase in product demand.

The retailer sees inventory falling and orders more stock. But delivery takes time, so inventory continues shrinking. The retailer places another order to correct the apparent shortage.

The wholesaler observes those larger orders and responds similarly. Its supplier does the same.

When the original shipments finally arrive, the retailer has more inventory than needed. Orders are cut, and the adjustment travels back through the supply chain.

This is one mechanism behind the bullwhip effect, where relatively modest changes in customer demand produce much larger fluctuations in upstream orders.

The underlying feedback is balancing: each company tries to close the gap between actual and desired inventory.

But delivery delays mean corrections continue before earlier corrections take effect. The system overshoots, then corrects in the opposite direction.

Demand forecasting, bulk ordering, shortages, and promotional pricing also contribute to the bullwhip effect.

Research: Lee, Padmanabhan, and Whang, Management Science (1997)

No participant needs to behave irrationally. Local decisions based on incomplete information generate instability across the system.

Why Small Changes Produce Disproportionate Effects

Complex systems frequently respond nonlinearly: the effect of a change depends on the system's current condition.

The traffic experiment illustrates this.

When vehicles have sufficient room, a slight reduction in speed is absorbed by following drivers.

At higher density, drivers have less room to adjust. The same disturbance grows into a stop-and-go wave.

The critical difference is not the size of the disturbance. It is the condition of the system receiving it.

This explains why a system appears stable for long periods before suddenly behaving differently.

Accumulating pressure changes the system's ability to absorb disturbances. What mattered little yesterday becomes consequential today.

The effect of a small change depends on where it enters the system.

How Changing One Relationship Changes the Whole System

In 2018, researchers returned to the circular-track traffic experiment with a new approach.

Instead of instructing every driver to behave differently, they introduced a single vehicle using automated speed control.

The vehicle adjusted its speed to smooth traffic flow rather than repeatedly accelerating and braking in response to the car immediately ahead.

The result was a substantial reduction in stop-and-go waves across the group.

Research: Stern et al., Transportation Research Part C (2018)

The researchers had not eliminated human driving imperfections. They changed how disturbances traveled through the system.

That distinction matters beyond traffic.

An organization experiencing recurring problems might replace individual performance incentives that undermine cooperation with shared goals. A supply chain might give manufacturers access to actual retail sales instead of forcing them to infer demand from fluctuating orders.

The objective is to change the interactions producing the problem, not merely demand better behavior from every participant.

What We Design Versus What We Create

Organizations design policies, processes, incentives, and responsibilities. They do not directly design every behavior that emerges when people interact under those arrangements.

A policy intended to increase accountability might encourage employees to conceal mistakes. A bonus intended to improve productivity might create competition between teams that depend on cooperation.

Emergence also produces benefits. Cooperation, innovation, and collective problem-solving develop through interactions no individual fully controls.

The lesson is not that complex systems are unmanageable. It is that managing their parts is insufficient.

When an unexpected behavior keeps recurring, asking who caused it misses the deeper explanation.

What is it about the system that keeps producing this outcome?

📚 Bookmarked for You

Thinking in Systems: A Primer by Donella H. Meadows: Explains feedback loops, delays, and why well-intentioned interventions produce surprising consequences.

Complexity: A Guided Tour by Melanie Mitchell: Explores how interactions create emergent behavior in natural, technological, and social systems.

🧬 QuestionStrings to Practice

Trace an Unexpected Outcome

Use this sequence to investigate a recurring organizational problem.

What unexpected behavior keeps appearing? → Which decisions contribute to it? → How do those decisions influence one another? → Where do feedback loops or delays amplify the behavior? → Which relationship should we change first, and how will we test the result?

---

Explore another question each day at QuestionClass.

← Home