The other loops, by what each catches
Each rung catches a different class of failure, and not every rung has the same weight.
Round out the ladder, and frame each rung by the class of failure it catches rather than as a checklist to obey:
- Formatter — kills an entire category of pointless debate and diff noise.
- Linter — catches known smells and error-prone patterns before they cost anything.
- Cyclomatic complexity — a smoke alarm for code that’s gotten too tangled to reason about (and therefore too tangled to delegate or verify).
- Mutation testing — checks the tests themselves, by breaking the code on purpose and seeing if any test notices. Powerful, but slow and advanced — not table stakes. Reach for it when a suite’s trustworthiness genuinely matters, not by default.
Note the proportion lens again: these are not all the same weight. A formatter is free and universal; mutation testing is a specialist tool. Treating them as one undifferentiated “best practices” list is its own small slop.