In this blog post, the author argues against the traditional notion that software engineers must fully understand their codebase. Drawing a distinction between small and large codebases, the author contends that in large systems, complete understanding is unattainable and suggests that partial understanding can be effective. He references Peter Naur's idea of 'theory building' but critiques its application to modern software, emphasizing the importance of partial theories in navigating complex codebases and using tools like LLMs. The post ultimately stresses the need for engineers to adapt their understanding to fit the context of their work.