Comments that narrate what the code does ("Loop through the users")

Comments and text: one of the ways AI coding agents bloat software, from lessloc's pattern catalog. lessloc catches it as narration.

What makes comments that narrate what the code does a bad practice?

The standards that call it a bad practice: Clean Code C3 Redundant Comment and Noise Comments; McConnell, Code Complete §32.4 (a comment that repeats the code is not acceptable in finished code); Ousterhout: Comment Repeats Code.

What does lessloc do about comments that narrate what the code does?

lessloc catches it as narration. Comments that give a reason are kept.

• narration (tier 3): Comments that narrate the code, decorate it or restate a name, closing-brace comments, and log lines that narrate each step.

Examples

Caught as narration:

def total(xs):
    # Loop through the items and add them up
    return sum(xs)

Written well:

def total(xs):
    # Refunds are negative, so the sum is already net
    return sum(xs)

Related patterns

The other patterns in Comments and text:

• Banners, dividers, closing-brace comments, emoji decoration: lessloc catches it as narration.
• Docs that restate the name: lessloc catches it as narration.
• History and messages to the requester: lessloc catches it as history.
• Comment blocks and comment-heavy changes: lessloc catches it as comment-block and comment-density.
• Log lines that narrate each step: lessloc catches it as narration (three or more in a file).
• Docs that still name what a change removed: lessloc catches it as stale-doc.
• Summary, report and notes files: lessloc catches it as junk-file and refuses it before it lands.
• Doc files and README growth nobody asked for: lessloc catches it as unrequested-doc.
• Docs deleted wholesale to shrink a change: lessloc catches it as doc-removed.
• Always-loaded instructions that grow: lessloc catches it as long-instructions.
• Commit messages that recap the diff file by file: The commit hook warns when a message lists the files the diff shows.