🔗 The unreasonable effectiveness of the Highlander rule

Matthew Sinclair argues that a memorable three-word coding rule—"there can be only one"—outperforms complex specifications for managing AI-generated codebases. The principle eliminates duplicate code paths and functions by enforcing a single canonical implementation. Sinclair observes that Claude Code began identifying violations unprompted once the rule was established, suggesting metaphor-based guidelines compress meaning more efficiently than technical prose. The broader insight: pithy names carrying cultural weight communicate architectural intent to AI agents more effectively than dry specifications, demonstrating "unreasonable effectiveness" of well-chosen constraints.