Software architecture is full of principles that have shaped the way systems are designed for decades. But according to Barry O'Reilly, many of these ideas rely on assumptions about a future that cannot be predicted. In this interview from the Software Architecture Gathering 2025, he explains why concepts like loose coupling and traditional risk analysis deserve a closer look, and how his Residual Theory helps architects build systems that remain resilient in the face of uncertainty.

Key Takeaways:

1. Loose coupling is a valuable goal, but architects still lack a reliable way to determine where coupling should actually exist.
2. Complex software systems should be designed to handle uncertainty instead of relying on predictions about the future.
3. Business factors, such as competitors, customers, and market changes, can put just as much stress on a system as technical failures.
4. Risk probabilities often create a false sense of certainty because they reflect assumptions rather than objective facts.
5. Better architectural decisions begin with accepting uncertainty instead of trying to eliminate it.

Barry O'Reilly encourages software architects to question long-standing assumptions and rethink how they approach complexity, risk, and system design. His interview offers a fresh perspective on uncertainty and demonstrates why embracing the unknown can lead to more resilient software architectures. Watch the full interview below to learn more about the ideas behind Residual Theory.