https://jlshen.com/Writing by Junlong Shenenhttps://jlshen.com/writing/code-vs-price/https://jlshen.com/writing/code-vs-price/Sat, 03 Oct 2026 00:00:00 GMTI tested the usual ways of counting AI coding output (PRs, lines of code, how long code survives) against 8,526 code changes a company actually paid for. They barely track the price, and survival doesn't predict later bugs.https://jlshen.com/writing/only-the-denominator/https://jlshen.com/writing/only-the-denominator/Fri, 02 Oct 2026 00:00:00 GMTCompanies cap each employee's AI spend because no one can yet say what the output is worth; vendors measure it differently and nobody has checked them, and the people most worth measuring are the heavy users the caps bind.