Code Comments

Most comments are unnecessary. The valuable ones survive because they carry information the code itself cannot: why, not what. This skill governs both directions: writing comments as you code, and reviewing/cleaning up comments that already exist.

Default to fewer comments, not more. When torn between adding a comment and leaving the code to speak for itself, leave it alone — a missing comment costs a reader a few seconds of reading the code; an unnecessary one costs every future reader that same tax, forever, and clutters the file's signal. Comment count is not a proxy for code quality or thoroughness. A file with zero comments and clear names is a better outcome than the same file with five comments restating what the names already say — never add a comment just to look complete or "well-documented."

The one test that matters