Short, opinionated, and right about most things. The core idea is Deep Modules: hide complexity behind simple interfaces. This is ETC from another angle. If your module has a complex surface, every consumer pays that tax forever. Make the surface small and the internals can evolve freely. I read it in a weekend. I have been thinking about it for years.

I already thought in systems before reading this. What Meadows gave me was the vocabulary. Stocks and flows. Feedback loops. Leverage points. The places to intervene in a system, ranked from least to most effective. When I write about NixOS or Emacs or OneLoop, I am reaching for these ideas even when I do not name them directly. Short and elegant.

Hamming spent decades at Bell Labs studying what separates great scientists from merely good ones. His central question: why do so few people make significant contributions while so many are forgotten? The answer is mostly about how you allocate your attention and whether you have the courage to work on important problems instead of safe ones. This maps directly to what I wrote about transmittable vs. untransmittable knowledge. Hamming is rigorous, contrarian, and deeply practical.

The best systems engineering book of the last decade. Kleppmann builds from simple primitives (logs, hashes, SSTables) all the way up to distributed systems, always grounding abstractions in the physical realities they paper over. Even if you never build a distributed database, the way Kleppmann thinks will change how you design any system. This book is first principles applied to data.

Interviews with 19 legendary programmers from the 1980s. Bill Gates, Andy Hertzfeld, Charles Simonyi, and others. What makes it timeless is the focus on how they think about problems, not what they built. You start to see patterns across how these minds approach complexity.

I live the Unix philosophy every day. Reading it from the source is different than absorbing it osmotically. Kernighan and Pike do not just describe the philosophy. They demonstrate it, building real tools from shell pipelines, showing how composability creates power that no monolith can match. Every time I write curl | jq or compose tools in Emacs, I am living this book. My OneLoop agent (read, write, edit, bash as four primitives) is an example of Unix way of doing things.

The spiritual ancestor of everything I love about Lisps. Homoiconicity. Functions as first-class citizens. Building abstractions upward from the smallest possible primitives. SICP teaches you to think about computation itself, not any particular language. If you are a Clojure programmer and you have not read it cover to cover, fix that.

Not a software book, and that is the point. Norman's central idea is that good design is invisible and bad design blames the user. This applies to everything from Makefile help targets to agent tool interfaces. Your self-documenting system? Your one obvious path? Your inspectability obsession? Norman gives you the design vocabulary to articulate why something feels wrong and how to fix it.

Weizenbaum wrote ELIZA in 1966, then spent the rest of his career warning about delegating human judgment to machines. He is not anti-computer. He is pro-human. He argues that some things should not be computed, not because they cannot be, but because doing so corrupts the thinking. Reading this in 2026, in the middle of the AI hype cycle, feels like finding a kindred spirit from fifty years ago. This is the philosophical depth behind "the agent is not the point."

The original essay on why adding people to a late project makes it later. But it is really about something deeper: the irreducible complexity of communication and the necessity of conceptual integrity in software. Brooks' concept of conceptual integrity (a system should reflect one set of design ideas, not a committee's compromises) is my oneness principle by another name. The anniversary edition includes "No Silver Bullet," which is even more relevant in the age of "AI will solve everything."

—

These are not the only books I have read. They are the ones I keep coming back to. The ones where I will catch myself mid-conversation reaching for an idea and realizing it came from one of these pages. If you have read them, you know what I mean. If you have not, start with 1, 2, and 9.