Selection

Selection is deciding what deserves the cursor’s attention. A codebase is full of tempting edits; the craft is choosing the surface that matters and leaving the rest alone.

The active context should stay small and close. The files that surround the change — the types it touches, the tests that pin its behavior — are the working set. Everything else is noise until it becomes relevant. A tight selection makes edits precise and reviews easy.

When the right surface is unclear, selection becomes investigation: read the callers, check the conventions, and find where the change belongs. Changing the wrong layer — patching symptoms instead of the shared function — is the classic error of selecting too small.

Selection also means knowing when to stop. The disciplined edit is the one that ships alone, without dragging unrelated refactors along. Choose the change, and choose its boundaries.