Managing the Unexpected summary

Organizations that handle surprises well are built on an invisible infrastructure that quietly shapes everyday work. It guides how people notice events, how they talk about what they see, and how they change their actions in response. At its core are four basic processes that, together, determine how the organization stays in touch with reality.

First come expectations. People carry a sense of what is normal, what would count as a warning sign, and what can safely be ignored. Those expectations save time, but they also determine which tiny anomalies get attention and which ones disappear into the background.

A key difference between reliable systems and fragile ones is how they treat small troubles. Instead of brushing off odd events as noise, reliable organizations treat them as early clues that something deeper may be wrong, and watch carefully for the first hints that action is drifting away from intention.

When something deviates in a small way from what people expected to happen, we have an anomaly. At first an anomaly is just a cue, something noticed for a moment. As people talk about it and connect it with other events, it can become a clue that a more serious failure is taking shape.

Simple stories feel safe, but in complex systems they can be dangerous. When messy situations are squeezed into neat labels like “routine incident” or “operator error”, important differences vanish and early signs of trouble become easy to miss. People narrow their attention to what fits the label and stop looking for details that might show the situation is changing.

A lot of this comes down to how we use categories. Labels help people coordinate, but they also pull them away from direct contact with events. Consider the outbreak of the virus that was eventually called West Nile virus. Initially, patients with fever, confusion, and muscle weakness were placed under an existing diagnosis that did not match their condition. Dead birds and local clustering were discounted because they did not fit the chosen label, and by the time those inconsistencies were taken seriously, valuable time had been lost.

Our third principle for reliable organizations is all about staying tuned to what is actually happening right now, not what the plan says should be happening. In complex systems, every action changes the conditions for the next action, so people need a live sense of how their work is affecting the overall system. That awareness is not limited to frontline operators; managers and support staff also work at "sharp ends" – points where decisions about budgets, staffing, scheduling shape what others can safely do.

You can see this type of awareness in places like surgery, mining, and power generation, where teams have to keep two things going at once: they need to keep the work flowing and, at the same time, protect the system from being pushed beyond its limits. To do that, they pay close attention to cues such as small delays, unusual noises, or awkward handovers, because these can signal that once-separate processes are starting to interact in unexpected ways. Sensitivity to operations means noticing those shifts early and adjusting before they cascade into something larger. 

Even in the most careful systems, some surprises break through the defenses, and what happens next depends on resilience. Here, resilience is about being able to keep going, improvise under pressure, and accept that performance may have to degrade in a controlled way rather than collapse all at once. Instead of assuming they can prevent every bad event, reliable organizations assume some things will go wrong and prepare to discover, on the spot, what needs to be done.

That preparation shows up long before an emergency. Reliable teams practice working with incomplete information, learn to recombine old skills in new ways, and develop a habit of watching how situations evolve over time. In the Sioux City accident, a DC-10 passenger aircraft lost almost all of its steering when an engine failure destroyed the hydraulic systems, leaving the crew with only engine power to influence the plane’s path. With no procedure that matched the situation, they had to invent a way to fly and land a barely controllable plane. In so doing they drew on long experience, a shared picture of what the aircraft was doing, and a constant cycle of trying something, noticing the response, and adjusting together.

Our final principle for reliable organizations revolves around a question: When demands speed up or conditions shift, who really gets to decide what happens next? In more reliable systems, authority moves toward the people who understand the situation best, not automatically to whoever has the highest rank. When that doesn’t happen, local specialists get overruled, technical warnings are softened, and control lands with the wrong person just when pressure is highest.

Treating expertise in a reliable way calls for a particular mindset. The organization needs to be seen as a network of different kinds of know-how, and the job in a surprise is to connect the right pockets of experience. That takes humility about the work itself: individuals are confident in their skills, but treat the situation as bigger and more uncertain than any one person can master. With experience, responders learn that even messy events eventually settle, which helps them avoid panic and hasty fixes.