1. Human-written code only — No AI code generation. Systems programming requires direct authorship and a clear understanding of what is built and why.

2. Correctness is non-negotiable — Software must work reliably and predictably. Bugs are failures, not features.

3. Understand your dependencies — Know what every dependency does. Every dependency is a potential point of failure. Build it yourself when you need control or deeper understanding.

4. Performance through design — Data-oriented, cache-friendly, manual memory management. Performance isn't an afterthought; it's a core design constraint for systems programmers*.

5. Explicit over magic — Code should be clear about what it does. No hidden behavior, no clever tricks that obscure intent.

6. Test comprehensively — Correctness doesn't happen by accident. Rigorous testing is mandatory, not optional, though software need not be test-driven.

7. Quality documentation or none — Bad documentation is worse than no documentation. Write simple, readable code and document non-obvious invariants and design decisions.

8. Build for longevity — Code should be maintainable and understandable decades from now. Future systems programmers* should not need archaeology to understand your work.

9. Systems languages — Statically typed, compiled languages with explicit control over memory and execution (Jai, Rust, C, C++, Zig, etc). These are the main tools of systems programmers*.

10. Be responsible for your software — Once you release software, keep it maintained or provide a better alternative. Don't abandon your work.

A systems programmer is someone who builds software with an explicit understanding of how it executes on real hardware. They reason directly about memory, data layout, data flow, control flow, and performance characteristics. This definition is inspired by ideas expressed by Jon Blow and Andrew Kelley, emphasizing responsibility, predictability, and deep understanding over convenience or automation.

Systems Programmers are the people keeping civilization from collapsing.

We don't use AI to write our software. When AI writes your code, you lose understanding of the details. You cannot be a systems programmer* if you don't understand the system you're building. AI-generated code creates confusion, obscures bugs, and accumulates technical debt that nobody fully comprehends.