Consistency makes it easier to automate a process. It is easier to write automation for cattle than for pets because there are fewer surprises and variations to be aware of and fewer permutations to test. Automation brings about opportunities for self-service system administration. Web sites and other tools can empower users to get their needs met without human intervention.
You can also look at this another way: Before we can improve things, we must make things consistent. Making improvements to something inconsistent is like wrestling a pig: It’s messy and you probably won’t win. Once things are consistent, we can make them better—optimize them—and we gain the freedom to experiment and try new things. Our experiments may fail, but if we do not try, there is no way to improve. At least with each failure we learn something. This is not a rationalization that makes us feel better about our failures: The experiments that are a success are valuable because the system has been improved (optimized); the experiments we revert are learning experiences that guide us as we make future improvements.
You’ll see this pattern of chaos⇒defined⇒repeatable⇒optimizing throughout this book. It is also the basis of “The Three Ways of Operational Improvement” described in Section 12.3.5, and is the basis of the assessment levels in Section 55.3.2.