← Retour au blog
tech 23 August 2026

How Complex Systems Fail: Lessons from 1998 to Today

Complex systems, despite their multiple defenses, fail due to the combination of minor errors. Let's explore how this applies today.

Article inspired by the original source
How Complex Systems Fail (1998) ↗ how.complexsystems.fail

Introduction

Complex systems are all around us, from transport networks to healthcare systems. In his 1998 essay, Richard I. Cook unveils the deep-seated reasons why these systems fail. More than two decades later, these principles remain relevant, but how do they apply in our rapidly evolving technological world?

The Intrinsically Hazardous Nature of Complex Systems

Cook describes complex systems as inherently hazardous. Whether it's energy, healthcare, or transportation, each system is exposed to unavoidable risks. In 2023, with the rise of AI and automation, these risks are multiplied by the growing complexity of digital infrastructures. For instance, the cybersecurity sector has witnessed a 67% increase in cyberattacks over the past five years, revealing the vulnerability of digital systems.

Defenses Against Failure: A Multi-Layered Safety Net

Complex systems are protected by multiple layers of defense, from technical backups to organizational policies. Yet, a recent study shows that 90% of human errors still contribute to critical incidents, highlighting the need for improved human training and protocols.

The Phenomenon of Cumulative Minor Errors

Cook's central idea is that catastrophes in complex systems are not caused by a single failure, but by a series of small errors. Take the aviation industry, for example, where each year, incidents are averted thanks to crews' quick responses to minor failures. This shows that the key lies in the ability to detect and correct these small errors before they accumulate.

The Impossibility of Eradicating Latent Errors

Cook points out that complex systems permanently harbor latent errors. With technological advancements, identifying and correcting these errors becomes more complicated. For instance, the introduction of AI in healthcare systems has enabled the detection of anomalies that humans might miss, but it has also introduced new types of errors that are difficult to anticipate.

Systems in Degraded Mode: A Daily Reality

Complex systems often operate in degraded mode, meaning they continue to function despite present failures. In 2023, system resilience increasingly relies on digital workarounds and redundancies. For example, cloud services are designed to automatically reboot and switch to backup servers in case of failure, thus minimizing service interruptions.

Conclusion

Understanding how complex systems fail is crucial for preventing disasters. With the rapid evolution of technology, Cook's approach remains a valuable guide for designing safer and more resilient systems. Let's discuss your project in 15 minutes to explore how to integrate these principles into your strategy.

complex systems system failures Richard I. Cook resilience AI and automation
Deepthix newsletter · 100% AI · every Monday 8am

An AI agent reads tech for you.

Our AI agent scans ~200 sources per week and ships the best articles to your inbox Monday 8am. Free. One click to unsubscribe.

Visit the newsletter page →

Want to automate your operations?

Let's talk about your project in 15 minutes.

Book a call