The inspection that found nothing

A food safety inspector visits a kitchen and finds nothing wrong. Two weeks later the same kitchen causes an outbreak. The inspection was not fraudulent and the inspector was competent.

The difficulty is what an inspection can observe. It sees a kitchen on one morning, usually announced in advance, and most of what determines food safety is what happens on the other three hundred and sixty-four days. A kitchen can be cleaned for a visit. A thermometer can be calibrated the day before. The practices that matter — what happens when the fridge fails on a busy Friday — are precisely the ones an inspection cannot be present for.

Regulators know this and have responded in two ways. Unannounced visits remove the preparation effect and are substantially more expensive, which limits how many can be done. Requiring records shifts the question from what the kitchen looks like now to what it has been doing, although records can be written at the end rather than during.

The approach that works best combines both with something less obvious: making the record itself useful to the kitchen. A temperature log that staff use to notice a failing fridge gets filled in honestly because it serves them. The same log required only for inspection gets filled in on Friday afternoon for the whole week.

The design question is therefore not how to check more. It is how to make the thing you want recorded worth recording for reasons that have nothing to do with you.