Code & Homicide: What Software Devs (..) from Offender Profiling
summary
Claims that 23–42% of developer time is wasted to technical debt and that static metrics alone are not actionable lead to a practical alternative: using CodeScene to map Phoenix’s ~40,000 lines and trace commits to reveal hotspots. A specific test function shows 69 commits and 274 lines, suggesting high ROI for refactoring; CoreCLR at ~8 million lines shows widespread red. The workflow applies offender methods like geographical profiling to prioritize files that actually get changed so fixes target impact.
AI-written summary · about
chapters (12)
- 0:00Intro
- 7:58Code at Scale: Why it's a hard problem?
- 9:27Quantifying technical dept via static analysis
- 11:45What is criminal profiling?
- 15:28Code Health: Beyond a single metric
- 18:11Visualizing code health of an Elixir Codebase
- 20:07Examples: A gallery of code
- 23:26What is criminal profiling? Continue
- 26:14Geographical Profiling
- 34:06Geographical offender profiling
- 38:19Refactoring recommendation
- 43:08What is legacy code?
related talks
40mTreat Your Code as a Crime Scene
38mCode Red: The Business Impact of Code Quality
48mA Crystal Ball to Prioritize Technical Debt
45mCode Red: The Business Impact of Code Quality
46mCode Red: The Business Impact of Code Quality
51mCode Red: The Business Impact of Code Quality
50mPsychotherapy for Code
59mManaging Code at Scale