Substrate mismatch

The tool or person you picked is good at the shape of the job but bad at the part that actually matters.

Six weeks after knee surgery, your rehab app is a green wall. Every session logged, 240 straight leg raises, a 41-day streak, little trophies. Your knee still buckles going down stairs. At the follow-up the therapist watches you do one rep and says your hip is swinging to spare the quad, and has been the whole time. The app counted every rep you did. It never once looked at how you did them.

Substrate mismatch is when the carrier you chose handles the format of the work well and cannot do the real operation inside it. The tell is a clean, complete record next to an outcome that has not moved. Name the actual operation in a verb, not a noun: not "track exercises" but "correct form under load." Ask whether your carrier can perceive the thing that verb acts on. If it cannot, move that one part somewhere that can, and let the tool keep the part it is genuinely good at. Here that is a weekly video sent to the therapist, or three supervised sessions, while the app keeps counting. What changes is that the record starts meaning something. The false alarm: when results are actually improving, a bare log is fine, and you do not need a human watching you do arithmetic.