At over 23 hours, this was a massive audiobook, but I think it’s one of the better books of software engineering advice I’ve read. The core framing is that programming is just the immediate act of producing code, while software engineering is what you have to do to keep that code sustainable over time and across a team, or as they put it, it’s programming if clever is a compliment and software engineering if clever is an accusation. Hyrum Wright is one of the co-authors, and you can tell he’s proud of Hyrum’s Law since it gets cited at least three times. It’s a solid observation though: with enough users, every observable quirk of your system becomes a contract someone depends on, and while Google obviously sees that at an extreme scale, I’ve seen it crop up even in my own smaller applications.
While it’s true that not everyone works at the scale of Google or has the same need for code longevity, they tend to point out when their considerations are primarily scale-driven, and they share a lot of hard-won advice. They don’t shy away from making lessons out of where they’ve made mistakes either, offering several case studies where they openly admit they wish they had handled things better. A huge chunk of the book is about testing, and I really liked their emphasis on testing public behavior and outcomes rather than implementation details. They lean heavily toward using real dependencies or high-fidelity fakes instead of mocking everything out, which matches my own experience with mock-heavy tests becoming brittle and useless.
The chapters on Google’s internal tooling were interesting too, even if most of us don’t have the luxury of having thousands of software engineers around to develop custom build systems, code search tools, and entire languages just to support our core products. Still, their arguments for simple trunk-based development, short-lived branches, and relying on fast rollbacks rather than fear of releasing felt grounded in reality.
Since this was another audiobook, my notes aren’t quite as filled out as normal, since I mostly just try to jot down core ideas when I’m listening on a run. There were plenty of good observations in here that I didn’t get down, but I think I managed to capture the main points that really struck me. It’s a long listen, but well worth the time.
Full book notes.