Measuring Engineering Impact from Git History Alone

Why commit counts, PR counts, and lines of code fail to capture real engineering strength As engineering teams grow, one question becomes unavoidable: How do you evaluate engineering strength using observable signals instead of intuition? Most organizations fall back on weak proxies. Commit count Lines of code Number of PRs All of these are easy to measure — and easy to misinterpret. A typo fix and a system-wide architectural change both count as "one PR". A generated lockfile can add thousands of lines. Commit habits vary wildly between engineers. Yet inside every team, people still have a sense of who the strongest engineers are. "This person writes code that lasts." "That person touches everything but somehow nothing improves." Those intuitions exist, but they are rarely measurable. I wanted a model that uses only git history to approximate real technical influence in a codebase. Internally, I sometimes jokingly call it an engineer's "combat power." But what it actually measures is