The Wonderful Wizard of LoC: Paying Attention to the Man behind the Curtain of Lines-of-Code Metrics
summary
Paper argues that counting lines of code often misleads: a study of top-conference papers finds 44% reported code and among those most use LoC while only two used effort estimation. LoC varies wildly with language, formatting, and boilerplate so results aren’t comparable across projects. Recommends declaring intent, preferring qualitative descriptions or alternate metrics (AST node counts, functions touched, commit heatmaps), and when using LoC reporting the exact measurement procedure so comparisons actually work.
AI-written summary · about
chapters (7)
- 0:00The problem with code metrics
- 1:23Why we measure code
- 2:08Methods for measuring code
- 4:23Common practices in papers
- 5:40Challenges with line counting
- 6:59Recommendations for authors
- 8:55Specialized reporting cases
related talks
33mCode Quality Metrics to Measure and Quantify Quality of Code
01h 00mCode Quality in Practice
28mActionable insights vs ranking How to use and how NOT to use code quality metrics
39m7 Secrets of Maintainable Codebases
29mYour Metrics Are Lying: How to Stop Reporting and Start Learning
40mTreat Your Code as a Crime Scene
32mPrioritizing Technical Debt as if Time and Money Matters
50mThe Code is Documentation Enough