On the difference between complicated and complex

A useful distinction I keep coming back to:

Complicated systems have many parts but are ultimately predictable. A car engine is complicated—lots of components, intricate interactions, but an expert can understand and predict its behavior. Disassemble it and reassemble it, and it works the same way.

Complex systems are fundamentally unpredictable because their behavior emerges from interactions between parts. A traffic system is complex—you can’t understand traffic by studying individual cars. The system behaves in ways that aren’t reducible to its components.

This matters for policy. Complicated problems yield to expertise and engineering. Complex problems require different approaches: experimentation, adaptation, resilience.

Many policy failures come from treating complex systems as merely complicated—assuming that if we just understand all the pieces, we can predict and control the outcome. But complex systems surprise you. They have feedback loops, emergent properties, and tipping points that resist prediction.

The question to ask: “Can I understand this system by studying its parts, or do I need to observe the whole system in action?”