At Chaos Conf 2019, Subbu Allamaraju, Vice President of Expedia, presented a talk focusing on the importance of forming failure hypotheses within Chaos Engineering, emphasizing that success in this field requires a cultural shift towards learning from incidents. He highlighted that while Chaos Engineering is often perceived as a method of inducing random failures to test system resilience, it should instead be approached as a scientific process similar to A/B testing, where hypotheses are tested in controlled environments. Allamaraju argued that organizations should focus on understanding the "as it is" state of their systems, rather than merely the "as designed" state, through the analysis of real-world incidents. He underscored that many failures are triggered by changes and that complex production environments often lack clear fault lines, making second-order effects difficult to troubleshoot. To effectively prioritize and implement Chaos Engineering, he suggested articulating the value of such work by focusing on critical areas and making value-based decisions. This approach helps in understanding the system's actual behavior, avoiding symptom-chasing, and recognizing the roles of people, processes, and tools in maintaining system reliability.