Developer credits commenter who ran his code and found three real bugs in it

A software developer published a blog post featuring a retry helper function with budget-guarding and exponential backoff logic, which passed his own review and that of several readers. A commenter then took the code off the page, ran it on a test bench with a stubbed clock and zeroed jitter, and returned with reproducible results exposing three distinct defects. The bugs included a budget check that could not catch overruns until after they occurred, silent discarding of backoff when a server sent a Retry-After header, and an attempt-count multiplication that could turn four logical retries into a dozen actual HTTP requests. The developer reproduced all three issues, confirmed the findings, fixed the code, and published a dated correction rather than editing the post silently. Reflecting on the experience, he argued that a commenter who actually runs the code provides far more value than one who only engages with the argument, reframing corrections as the highest-value output a comments section can produce.