S.MISCAST.C — Control/verification gap

Outputs turn into action without any test, limit, or way to catch a bad one before it causes harm.

You set every household bill to autopay two years ago and stopped looking. This week you noticed your internet provider has been charging a rate that expired in March, and a gym you quit last summer collected 11 more payments. Nothing alerted you. The system executed perfectly; it just had no way to notice it was executing something wrong.

A control/verification gap is the diagnosis for outputs that turn into action without any test, limit, monitor, exception route, or recovery plan to catch a bad one before harm. The cue is a phrase like "it just runs" with no answer to what happens when it runs wrong. To apply it, ask the spec's question directly: how would we detect a bad output before harm, and how would we recover after failure? Then look for the missing pieces, such as a monthly anomaly scan, a charge cap, or an alert threshold. When the diagnosis works, you stop debating whether automation was a mistake and start adding the bounds and checks it never had. Do not reach for this label when the outputs are actually fine and the real problem is that nobody agreed who may act on them; that is authority, not control.