Number of Tries Using Coding Agents May Say More About Standards Than Competence (codemanship.wordpress.com)

A recent discussion among software developers highlighted an intriguing perspective on the use of AI coding agents like Claude Code. While many believe that developers who achieve good results quickly demonstrate greater competence, some argue that the number of attempts may instead reflect their personal standards for quality. This conversation has inspired deeper contemplation about how different approaches to AI coding can lead to varying perceptions of skill and effectiveness. The interplay between quality standards and coding efficiency raises important questions about the nature of creativity and technical proficiency in AI-assisted development. Just as film directors differ in their approach—some favoring fewer takes for immediate satisfaction while others pursue perfection through multiple iterations—the same can be observed in developers using AI tools. This suggests that the divide between those who complete tasks swiftly with AI and those who prefer a more hands-on method may be more about subjective quality assessments than raw technical ability. As the AI/ML community continues to explore the implications of these tools, understanding these differing perspectives could help refine how tools like Claude Code are integrated into development workflows.